Live-Cell Imaging and the Limits of Structural Biology | Eric Betzig on Super-Resolution Microscopy

0:00Almost everything you learn in biology

0:02textbooks is a hallucination. You guys

0:04have probably seen on the web. You know,

0:05there was those beautiful things of

0:07here's a cargo on a chin walking along a

0:11microtubule like this and it's all like

0:13in this vast empty space. I don't know

0:16any cell that's a bunch [laughter] of

0:17vast empty space. I'm sorry. It's

0:19crowded as [ __ ] There's 10 billion

0:21protein molecules. There's 10 billion

0:23carbohydrates. It's by far the most

0:26complex matter in the known universe. We

0:28understand the interiors of neutron

0:30stars far better than we understand the

0:32interior of cells. There's a reason why

0:35only 9% of the drugs that enter phase

0:37one come out phase three cuz we don't

0:39know what the we're doing. We don't know

0:41the real mechanisms that are going on.

0:43And when you start to [ __ ] look at

0:45the dynamics, not just the structure,

0:47you realize that you had it all wrong.

0:49And you realize that so many of the

0:51things that they thought they knew, they

0:53you can't be sure that they know. We

0:54have to reinvestigate all of it. This

0:56week, Mike and Misha sit down with Eric

0:58Betsig, Nobel Laurate and UC Berkeley

1:01professor who pioneered super resolution

1:03microscopy. Eric traces his path from

1:05Cornell to Bell Labs, including two

1:07stretches of unemployment that produced

1:09some of his best ideas. He reflects on

1:12his work in the auto industry in the

1:13living room, where he built a microscope

1:15to beat the defraction limit. Now he's

1:18looking ahead to his new cell

1:19observatory, using AI to turn pabytes of

1:22live cell imaging into a queryable model

1:24of biology.

1:26Please welcome Eric Betsy.

1:30>> So Eric, you've been working like

1:33battling with defraction limits

1:35throughout your career. What is the

1:38defraction limit? How would you explain

1:39it? Well, it's was first figured out

1:42around the end of the 19th century is

1:45and it kind of makes sense that if

1:47you're going if light is waves and

1:48you're trying to get an image through

1:50light, then you can think of the light

1:54colloquially as like little fingers

1:56trying to feel the sample, right? And if

1:58the fingers are too fat, you're not

2:00going to feel the structure below the

2:02level of those fingers, right? And so

2:05basically about half the wavelength of

2:07light is what Ernst Abbe mathematically

2:10determined is the limit. You know he did

2:13it but you know as soon as people

2:15understood the wave nature of light

2:17going back to Huygens or or certainly

2:20Maxwell all of this could have been

2:22easily deduced. Um but he was the guy

2:26who really realized from a from a real

2:28imaging perspective that that was a

2:30fundamental limit. you know, in in in

2:32many fields, of course, you know, when I

2:34first started getting in in the 80s in

2:36graduate school, there was obviously

2:38already semiconductors and other

2:41structures that were pushing those same

2:43limits in terms of their manufacturer

2:45and so forth. So, it was already known

2:47as an issue. And of course, this is why

2:49the electron microscope was invented in

2:51the 30s with Ruska and others, right? Is

2:53to get around that limit. Um uh in

2:56biology it was perhaps less appreciated

3:00that that you'd necessarily want to go

3:03further but you know the defraction

3:05limit is such that it's roughly 100

3:07times smaller than a cell. So you can

3:08still learn a lot but it's 100 times

3:11larger than the molecules that make up

3:13the cell. So obviously there's an

3:15incentive to try to do better.

3:18Maybe if you could give us like some

3:20taste of like maybe some numbers in

3:22terms of let's say there's like

3:24available off the shelf microscope

3:27someone can get like with objectives and

3:30how

3:30>> how small those numbers are. So yeah, so

3:34with visible light you're going to be

3:35down to around 200 nanometers. Okay. So

3:38again a protein molecule might be four

3:41nanometers. So you're 50 times too

3:43coarse to get to that level. But a cell

3:46might be 20 microns, which is 20,000

3:49nanometers. So then 200 nmters means

3:52you're you're roughly, you know, can see

3:55100 spots across the the width of a

3:58cell. Um, but you have to have a good

4:00microscope. I mean, the kind that you'll

4:01have in a high school is probably not

4:03going to cut it, right? You need good

4:05objectives. It's it it an art and a

4:08science for those guys. And that was

4:10really the thing is in fact there's an

4:12interesting story is that when when um

4:15Abbeby came up with with the

4:18understanding of this limit not only did

4:20he do that but he also understood

4:22exactly how to create objectives um

4:27lenses that would actually reach that

4:28limit. He was working he was a professor

4:30at University of Yana, but he was

4:32working with Carl Zeiss who had his

4:34microscope company in Yana. And um and

4:37so he had all these prescriptions about

4:40how many lenses, how to grind them, you

4:42know, how to space them exactly to get

4:44to the defraction limit. Um Zeiss goes

4:47and follows those exact rules. And it

4:51was crap. Um and they're like,

4:53[laughter]

4:54why is this the case? And they realized

4:57was because the quality of the glasses

4:58were not good. Okay, that that um that

5:02uh you know this is often the way as

5:05experimentalists. You have a mental

5:06picture in your head but reality bites

5:08you in the ass. And so they went to auto

5:11shot who was making um glasses

5:14particularly in that time there was a

5:16lot of street lamps that were gas lamps

5:18and so he was making the the bora silk

5:21glass that was the the covers the lamps

5:23and they went to him and say hey can you

5:25make us better glasses and so basically

5:28shot figured out exactly how to make

5:31exactly the types of refractive indices

5:33they needed. Once they did that bang

5:35they were right to the defraction limit.

5:37And so, you know, uh, Abbeby, he did

5:40okay. Zeiss, he made a fair amount of

5:42money, but the real winner was shot

5:45because shot is today still one of the

5:48largest makers of glasses for every

5:51application in the world. Shot and

5:52Corning are the two big names in this

5:54stuff right to this day. And so he

5:56became uber rich because of his ability

5:59to make these glasses.

6:00>> And was it like impurity impurity

6:03issues?

6:03was part of it, but but also al also

6:06because you have to introduce impurities

6:08as well specifically to get certain

6:10refractive indices, right? You need to

6:12go

6:12>> anywhere from, you know, glasses is is

6:15normally 1.5 index and they would go

6:18down to below 1.4 all the way up to 1.9,

6:21right? And so you have a lot more tools

6:24at your disposal as an optical engineer

6:26if you have all these different

6:28refractive indices to make your lenses

6:30from.

6:31>> Why did it matter that much? because

6:32they could also use like, you know, the

6:34curvature.

6:35>> It's not enough. You don't have enough

6:36enough knobs to tweak to get to where

6:39you need to be with that alone. You

6:41actually need to be able to have

6:43different indices of glass to make it

6:45work.

6:46>> So,

6:46>> where are we now with objectives?

6:49>> Oh, I mean, so there's there's many many

6:53um uh fringe benefits of winning the

6:55Nobel Prize. So um one of them is I've

6:58had a long relationship with Zeiss

7:00because they licensed our patents on

7:02palm they licensed our patents on

7:03lattice light sheet but um there

7:07normally I go to Yana but one trip I got

7:09to go to Obercockin and Obercockin is

7:12where they make the big EUV lenses that

7:15then go into the ASM ASML machines that

7:18then go into making Nvidia chips and so

7:21forth. And um that is the closest I've

7:24ever seen to alien technology in my

7:26life. I mean these lenses are are you

7:29know taller than me. They're this fat

7:30around. They have um they're all

7:33reflective lenses because you can't use

7:35glasses there, right? But the coatings

7:38to make it reflective at EUV

7:40wavelengths. They have like 30 layers

7:42and because of the curvature even though

7:45it it's like sub a monotomic level of

7:50coating thicknesses varying across the

7:53diameter and then they have to put

7:55little impurities in so the so the atoms

7:57don't diffuse across the boundaries

7:59between the different layers and that

8:01this whole thing is literally gigantic

8:03and it's precise down to the atomic

8:06level in what they do. It's just

8:08absolutely, you wouldn't believe that

8:10this would be physically possible, but

8:12it they do it and it's it's incredible.

8:14>> Have you thought about a biological

8:17imaging microscope based on these giant

8:19mirrors?

8:23>> You could well if if a living cell could

8:26live under EUV wavelengths, life would

8:29be great, wouldn't it? But sadly, that's

8:30not the case.

8:31>> But I guess you could image like cryo

8:33cryosamples.

8:34>> Yeah. Yeah. Yeah. Yeah. But well, one of

8:36one of one of my main points that I make

8:38in all my talks today, right, and one of

8:41the reasons I pivoted from from

8:43structural super resolution is it's my

8:47firm belief that you cannot understand

8:49life without looking at it live. Mhm.

8:51>> And I I feel like a big problem in

8:54biology today and in pharma today is

8:57that there's been um in the because of

9:01the defraction limit because we reached

9:04this wall at the end of the 20th century

9:06and it was 100 times too small to see

9:08the molecules.

9:10Science pivoted from this holistic look

9:13at cells with that they could look at

9:15live cells to doing reductionism. And EM

9:18is one of those reductionist tools.

9:20You're doing structural imaging.

9:22>> But you will never understand life

9:24without looking at it live. Okay. So

9:28having high resol Everybody was thinking

9:30what's the point of a microscope? High

9:31resolution. No, it's one of the things

9:34you're interested in, but you're also

9:36interested in in speed because the cell

9:39is moving. You're interested in

9:41non-invasiveness. So what's the point of

9:43looking at a cell if you're killing it

9:45while you're looking at it? So all of

9:47these other metrics matter too. And so

9:49UV is not the path to glory to study

9:52living things. In fact, while we've

9:54learned a lot from the reductionist

9:56tools like like EM

9:59and and particularly like structural

10:00biology like Cryomm and Alphafold and

10:03all of that, there's a whole slew of

10:05companies five miles across the bay here

10:08trying to do drug discovery based on

10:10protein structures determined by

10:11Alphafold.

10:14the those proteins and that stuff is

10:17such an infantessimal part of the whole

10:20dynamic complex system that creates

10:22life. They're burning money for nothing.

10:25I I guarantee that unless they pivot to

10:28pull in other types of information,

10:30biological information, those companies

10:32will will not be here in 5 years. Okay.

10:35>> So what kind of information do you see

10:37needs to be there to like develop

10:39>> to be able to see the dynamics and to

10:41see the interactions that actually

10:43happen in living matter? That's the only

10:47imagine you try to reverse engineer an

10:49internal combustion engine. And if all

10:52you have is the reductionist tools that

10:54that rule biology today and ruled in the

10:5720th century, those are biochemistry,

11:00how different proteins and other

11:01molecules interact with one another. Um

11:05structural biology such as um alphafold

11:09or or um or em or stuff like that. And

11:13molecular biology, the central dogma,

11:15knowing DNA makes RNA makes proteins and

11:18all of that. But those are are

11:20incredibly reductionist tools and would

11:22be like trying to under to reverse

11:24engineer life from that would be many

11:27many orders of magnitude harder than

11:30having a random pile of internal

11:31combustion engine parts and trying to

11:34figure out how internal combustion

11:35engines work. Okay, that's where we are

11:38today. That's where most of the field

11:40is. And it's it drives me insane because

11:43I keep making this this point over and

11:45over again that there com it we need to

11:47go back to a holistic understanding of

11:50this complexity that exists in the cell

11:54instead of being focused on just little

11:57parts. So you mentioned icon is is a

11:59company we have u so the head of ICON is

12:03um is Roger Permutter who used to be

12:05head of research at Merc. is one of the

12:07most prolific and successful drug

12:09discoverers of of the 20th century and

12:12he has he has a line which is you know

12:15it's a miracle if we ever find a drug

12:18that works because we have no idea what

12:20we're doing and that is accurate because

12:23of this focus on such a crazy level of

12:26reductionism instead of understanding

12:28the system holistically. Do you know we

12:31were discussing this yesterday like uh I

12:33mean obviously we've kind of skipped a

12:35lot of the what what is super resolution

12:37microscopy or anything but

12:39>> is it do you have you ever thought about

12:41or like have have you ever kind of like

12:43come across an idea of somehow making a

12:45cell where it's genetically engineered

12:48to be friendly to microscopy. So meaning

12:51like you genetically engineer as many

12:53proteins, as many structures as possible

12:54in the cell to be easier to image so

12:57that you can image like all of them at

12:58on the in the limit it would be like

12:59they're all fluorescent in different

13:00wavelengths or something.

13:01>> Right. Right. Well, that's that's

13:02exactly the way we do engineer cells

13:05today, right? Is to put fluorescent tags

13:07on specific proteins so they'll light

13:09up.

13:09>> And that's one of the most limiting

13:11parts of optical microscopy still

13:13because in in visible wavelengths

13:16there's a limited number of colors that

13:17you can do. and there's 20,000 different

13:20types of proteins in the cell. It would

13:22be great if we could see them all at

13:23once, but that technology does not

13:25exist. And that is a nobel waiting to

13:27happen if somebody can directly

13:29interrogate proteins in some way and

13:32deduce them without having to put those

13:34tags on. But more generally, if you

13:36start modifying a cell in order to make

13:39it tractable to image it, you're not

13:42looking at real cells anymore.

13:43[laughter]

13:45You're looking at the thing you created,

13:47right? and and so uh yeah that's I I

13:50don't think that's the path to go

13:51>> is is there

13:53have people tried this where they try to

13:56like make the proteins more absorbing or

13:58more some more maybe like I mean the

14:00thing is like if you want to add let's

14:01say a green fluorescent protein that's a

14:02big protein so it's going to start

14:04changing of course

14:05>> no kidding it's a bowling ball

14:07>> but maybe two kilo dolton bowling ball

14:09added to added to your protein that's

14:11right

14:11>> yeah is there is there any has that

14:13actually y has that type of approach of

14:14actually genetically engineering the

14:16cells yielded any kind have like

14:17interesting results yet.

14:19>> Well, I I mean again, we genetic

14:20engineer all the time to put those

14:22labels on to put you you you do all sort

14:26all sorts of as many controls as you can

14:29to try to show that the cell is still

14:31behaving physiologically with that

14:34particular tag on. And a lot of times

14:36you'll find it isn't okay. And a lot of

14:38times it's a black art about the linker.

14:41for example, the the little the little

14:43peptide linker between the protein and

14:45the bowling ball and and all sorts of

14:48crap that you have to just there it's

14:50it's an art, not a science in terms of

14:52figuring out what works and

14:54recapitulates as much as we can tell the

14:57native physiology. But there's it's it's

15:00just it's just like you know the

15:01uncertainty principle. Anytime you

15:02observe anything, you're going to

15:04perturb the thing. And so it's really

15:06really important to make sure that you

15:08you do as many controls as you can to to

15:12believe your physiological.

15:13>> Yeah.

15:15>> So what kind of like cells or objects

15:18organism you are targeting?

15:20>> So many many many things. So we start of

15:24course with with cultured cells because

15:26that's the easiest right. Um and

15:29certainly with palm and so forth, that's

15:30where we had had to start super

15:32resolution started with with single

15:34culture. In fact, at first it starts

15:36with dead cells, fixed cells, chemically

15:38fixed so that so they're not wiggling

15:40around because the super resolution

15:42techniques at least initially and still

15:45in many cases today are slow and so

15:48things would blur out if you tried to

15:50look at super resolution of something

15:52that's wiggling and moving real fast. So

15:53you chemically lock it into place and

15:56then you image that, right? Um, so

15:58that's where you start. But those

16:00fixation protocols which started with

16:02electron microscopy are incredibly

16:05perturbative to the ultra structure at

16:07the nanocale you want to see. It's fine

16:09if you're looking at at regular optical

16:12resolution, but when you when you dig

16:14down to super resolution, you're looking

16:16at artificiality.

16:17And that was one of the reasons I

16:19pivoted from super resolution because

16:21most of the time people electron micros

16:24have known this for years. And so their

16:26gold standard for them is not to

16:27chemically fix cells. It's to high

16:29pressure freeze cells so that in in

16:32milliseconds you're locking in the

16:34structure in vitrius ice, not crystallin

16:36ice and then look at it that way. But

16:39that's a lot of work to do. Harold and I

16:41have done it at Janelian done

16:43correlative em and super resolution. But

16:46it is not it's not a protocol for the

16:49faint of heart. It takes a lot of

16:50technology and a lot of skill to make

16:53that kind of thing work. Do you try to

16:56combine both microscopes

16:58system?

16:59>> Yeah. Yeah. Yeah. So, well, you go from

17:01one to the other, right? So, so you

17:03vitriusly freeze the cells. Um, then you

17:06put them in your super resolution

17:09cryostat that's now at 4° Kelvin. And

17:12so, you do cryo cryopalm we call it,

17:15right? And you and you bleed out all the

17:17molecules, find out where they are. Then

17:19you take that and then you take it out

17:20while it's still frozen vitriusly. You

17:23do freeze substitution and you end

17:25basically end up putting it in resin in

17:28a way that still preserves that ultra

17:29structure. Then that gets sliced and put

17:32into Herold's three-dimensional focused

17:35iron beam milling electron microscope.

17:37So then you can get the

17:38three-dimensional profile that way. And

17:40then you take the two data sets, which

17:43of course have all sorts of like weird

17:45little distortions and stuff. And then

17:48you do kind of a a warping and and

17:50register them all together. And then you

17:52got your correl that's this one right up

17:55there. You see that? That's correlative

17:57superresolution. That orange thing with

18:00overlapping them there.

18:02>> Yeah. They're exactly mapped on one

18:04another right there. Yes.

18:05>> So you have to slice basically the

18:08>> in the in the EM. So there's a a focused

18:11ion beam like a gallium beam that comes

18:13and takes a nanometer or two off.

18:15>> How thin is the layer? It's like

18:18>> a couple nanometers. Really thin. And

18:20then you image that. You image that.

18:21and Herold is the god of fibsim. So, so

18:26connetoics is a big thing in

18:28neuroscience right now. So, um one of

18:32the first connetos was made possible by

18:35Harold's FIBSM at Genealia where I used

18:37to work.

18:38>> So,

18:38>> and that has to be done couple

18:40nanometers at a time.

18:41>> You couple nanometers image that

18:43surface. Couple nanometers image that

18:45surface. And

18:48>> it's it's less crazy than those lenses

18:50that I talked about at Zeiss, but it's

18:52pretty crazy. Yes.

18:53>> Not very scalable.

18:54>> Uh it's not terribly scalable. That's

18:56[laughter]

18:57great.

18:57>> It's not going to be a consumer product

18:58anytime soon.

18:59>> No. No. But none of what we're talking

19:01about is yet.

19:03>> Uh may maybe we can go back to like uh

19:06to your jumping all over the place.

19:07>> You're a physicist, right? So how did

19:08you get into all this like wet, you

19:10know, fluffy stuff?

19:12>> Yeah. So So curiosity, right? So um

19:16yeah, I started as a physicist and you

19:18know as a kid I wanted to be an

19:20astronaut cuz I grew up with Apollo and

19:22Star Trek, right? And uh but by the time

19:25I was in in college and by the time I

19:28graduated from college um the the

19:31shuttle was coming online and I knew at

19:33the time the shuttle was the biggest

19:35mistake you could possibly make. It was

19:36a horribly engineered

19:38>> system, a complete waste of money.

19:41nothing but a nothing but a a um a

19:45political gift to the legacy aerospace

19:48contractor.

19:49>> Did you like the Soviet rockets better?

19:50>> Uh well the Soviet this the Soyos has

19:54been flying for what 50 freaking years

19:56or something, right? That's pretty good,

19:57right? But obviously SpaceX has

20:00completely changed the nature of the

20:01game, right? So this is very exciting to

20:05me because I think that uh you know I

20:08honestly believe there is a finite

20:10chance that I could still reach my dream

20:12of making it to space before I die and

20:15that is the number one thing on my

20:16bucket list is is to make it to orbit.

20:19Yeah. So

20:20>> take you had to take a break from that

20:22dream.

20:23>> Yeah. But but anyway anyway so getting

20:25back to your point, how did I get into

20:26the wet and squishy stuff, right? So, I

20:29went to graduate school. Um, and I've

20:32always wanted to be like I wanted to do

20:35physics not because I wanted to be a

20:37fineman or anything. I wanted to do

20:39engineering physics. Okay? I wanted, you

20:42know, my dad was an engineer. I wanted

20:44to be an engineer. I like making things.

20:47Um, and so at that time there were only

20:50two um applied physics departments in

20:53the country. One was at Stanford and I

20:56[ __ ] hated California. I wanted to

20:58get the hell out. And then the other was

20:59Cornell. And Cornell was a lot more like

21:01Michigan where I grew up. So I went to

21:03Cornell. And there was two professors.

21:06One was a guy who was an electron

21:08microscopist. And the other guy was a

21:10raman spectroscopist.

21:12And um they had this crazy idea that if

21:16you could use your electron microscope

21:18to drill a hole smaller than a

21:20wavelength in a black film, then you

21:23could press it against a cell and you

21:25would have a little nano flashlight that

21:26would only illuminate one spot. You

21:28drive it around, you get a super

21:30resolution image. And you know the idea

21:33is what if we could make an a microscope

21:36that could look at living cells with the

21:38resolution of electron micros. All you

21:40have to do is say that sense and you go,

21:42"Oh my god, that would be incredibly

21:45revolutionary, right?" And you say,

21:46"Okay, that's the kind of thing I'd like

21:48to engineer." So that was my entrance

21:50into working in super resolution for my

21:53thesis. And so I did near field

21:55microscopy as it was called because

21:57you're in very close to the aperture.

22:00You're not in sort of the sort of

22:02farfield optics limit. you have to get

22:04into the details of the of electric

22:07fields that decay exponentially away

22:10from the thing. So that's called near

22:11field optics. Did you use a fiber or

22:13>> I and not at that at that time we used

22:16um so one of the things that was new at

22:18that time was some something in

22:21electrophysiology called patch clamping

22:23where you could basically look at single

22:25ion channels in a membrane by pulling

22:27glass pipets down to the nanometer scale

22:30and then pop that thing as an electrode

22:33on top of the ion channel actually

22:34measure sort of single ion currents

22:37going through pores in in cells. Okay.

22:40And so we realized that well if we coat

22:43that thing with metal then the hole in

22:45the end of that glass thing is going to

22:47be our little aperture and that's a lot

22:49easier than having to go to the electron

22:51microscope and drill these holes in this

22:53in this thin film. And the thin film if

22:56you looked at it the wrong way would

22:57shatter and so um the pipets were a lot

23:00more robust and simpler. And so that's

23:02what I did. And then I took that

23:04technology uh in my own lab at Bell Labs

23:08and then realized once I was there that

23:11we could get much better delivery of

23:13light by instead taking optical fibers

23:15and pulling them like a piece of taffy

23:18and then it would taper down and break

23:20with cleave with a flat end and then

23:23illuminate the sides and then it would

23:24be optically guided in the wave guide

23:26down to that aperture. So we'd have a

23:28much brighter light source. And then

23:31that was where I developed my reputation

23:33as a super resolution guy. And the

23:36experiments I did there, although I

23:38didn't know it at the time, kind of set

23:40the stage for what would become palm and

23:43other super resolution.

23:44>> How small can you get in terms of the

23:46>> 20 nmters?

23:47>> I see. And that was the your like

23:49resolution limit essentially.

23:50>> Yeah. Yeah. Yeah. Yeah. it tough to get

23:53the tr the the the key thing that killed

23:56um near field for biology is that again

23:59because it is near field the light that

24:01comes out of that hole spreads

24:02incredibly rapidly. So if you're even 20

24:06nanmters away from the surface, you

24:08know, you've lost much of your

24:10resolution.

24:11>> And I didn't know a lot of biology at

24:13the time, but I knew cells were a lot

24:15rougher than 20 nanometers, right? And

24:17so there was there really wasn't any way

24:20I was going to be able to follow even

24:21get the surface and follow the contours

24:23of the surface with this this big ass

24:25probe and and and get the resolution

24:28that I needed. And there was no way I

24:30was going to get to the interior of the

24:31cells. So, I did some experiments on

24:34cells then, but I picked cells that were

24:36known to be crazy flat, right? And and

24:38not the whole cell, but just out at the

24:40periphery where it's just kind of,

24:42>> you know, and demonstrated that I could

24:45do it.

24:46>> But, um, it was really for biology

24:48pretty much a dead end.

24:50>> By the way, related maybe to defraction

24:52limits. Seems like it's a good

24:53demonstration that if you have a tiny

24:55hole

24:56>> like 10 20 nm like 10 times smaller than

24:58the wavelength

24:59>> that the light wants to expand. Is there

25:02really really fast

25:02>> I mean I can derive it like let's say

25:05from

25:05>> yeah it falls

25:06>> from the formulas but

25:07>> it's an effan field it's exponential

25:09right

25:09>> right is there intuitive u explanation

25:12for

25:13>> uh I I don't know if I don't not really

25:16unless you want to think of it as like a

25:17fire hose right or something like that

25:19right if that's your level of intuition

25:21but generally speaking it you know it

25:24was it was something that that was a

25:25concern of mine from day one in fact the

25:28first thing I did when I got in there as

25:30a graduate student is I I I couldn't do

25:34a round hole, but I developed um a

25:36theory for figuring out how fast the

25:38defraction is from a from a slit. Um and

25:42I did that on an IBM the an IBM PC, the

25:458088. [laughter]

25:48And so I was able to do do that

25:50calculation said, "Yeah, it's pretty

25:52fast." But I didn't necessarily have

25:54enough confidence to think that well let

25:57let's just do the experiment and find

25:59out how bad it is.

26:00>> How were you getting the probe so close

26:02to the cell without like like how do you

26:04you like

26:06>> that's the other right? So what you need

26:08is an independent feedback mechanism to

26:10regulate the height of the aperture

26:12above the surface. So um there's there's

26:16two answers to that question. The first

26:18is um if you at Bell Labs I had lots of

26:22hits. I had a number of you know science

26:24papers

26:26you know at at one you know I I I became

26:30I became well known as a scientist at

26:32Bell Labs but if you look at all of the

26:34applications I published they have one

26:36thing in common the samples were really

26:39really flat [laughter]

26:41okay so we did like high we had the

26:44world record for high density data

26:45storage at one time with near field

26:47optical data storage because you're

26:49looking at very flat you know

26:52ferroelectric film that you can switch

26:54the bits in, right? Um and uh but still

26:58it was it was uh limited to very flat

27:01things. But

27:01>> so your samples were the magnetic

27:03samples or

27:03>> Yeah. Yeah. Yeah. Yeah. Exactly.

27:05>> Seems like it was a big theme uh to

27:08develop data storage systems.

27:10>> Oh yeah. Yeah. I you know back I you

27:12know flew out here and had talks with

27:14Seagate about commercializing and [ __ ]

27:16like that. None of it none of it ever

27:18went anywhere, but at least at the time

27:20it was it was something that was worthy

27:23of thinking about, right? But but to get

27:26back to your question, in order to get

27:28that regulation that I developed at Bell

27:31Labs and it turned out so this was

27:34nearfield was a form of what is called

27:36today scan probe microscopy. The most

27:39famous example of that is scanning

27:41tunneling microscopy, right? And that's

27:42the one that that won the Nobel in '86.

27:45That was while I was um where was I? I

27:48was still in grad school in ' 86, right?

27:50And so at first we were trying to you

27:52know because our our tips were our

27:54nearfield tips were coated with metal in

27:56order to make them opaque. Well, now I

27:58can try to use that to do tunneling

28:01microscopy against the sample if I have

28:02a conductive sample. That didn't work so

28:05well because those tunneling distances

28:07are not nanometers, they're angstroms

28:10[laughter] which gets even harder,

28:12right? But another technology that

28:15developed out quickly out of out of

28:18scanning tunneling microscopy was

28:20something called atomic force microscopy

28:22>> where instead with a with a tip that's

28:24attached to a spring you can kind of

28:27feel the forces at at maybe not the

28:30atomic level but sort of at the

28:32nanometer level you could kind of do

28:34that. And so the trouble was that was

28:38done with a tip on a can lever. So the

28:40canal lever is very floppy, right? And

28:42my probe has to go this way and it's

28:44stiff this way, right? And so what I did

28:47instead is I dithered it in the floppy

28:49direction. And then as I came close to

28:51the surface, there would be enough

28:53forces between the tip and this that it

28:55would slightly change the resonant

28:57frequency of that thing, you know.

28:59>> So damping it.

29:00>> Damping. Exactly. And so you could see

29:02that slight change in the damping

29:04frequency and use that as a regulatory.

29:06That worked really well. So that worked

29:08on cells, worked on everything. So sheer

29:10force feedback is a thing today. And

29:13that was developed specifically for

29:15Nearfield first. Yeah.

29:19>> When did you realize that like Nearfield

29:21wasn't going to do it?

29:23>> Because eventually I got to the point

29:25after six years in which I had pretty

29:27much exhausted every flat thing worth

29:29[laughter] looking at.

29:32>> Okay.

29:34But at the same time uh and this gets

29:36now into sort of philosophy of science

29:39and and why I'm not really a scientist

29:42right is um is um

29:46I I I I dropped near field is in part

29:50because I dropped science altogether. I

29:52got really fed up with science. Um

29:54because you know when I when we first

29:56started in graduate school to do near

29:59field everybody told us we were nuts.

30:01there's no way fraction limit blah blah

30:04blah this will never work. Um and um it

30:08was enough to to convince the people at

30:10Bell with the results I had to do it.

30:13And um and it just took off. I mean

30:16again it was it was a terrific best time

30:18of my life was at Bell. Um and and

30:21things were going great, but I knew the

30:23limits, right? I and and uh and they

30:26were real, but at the same time, you

30:28know, it's it's how science is very

30:31fattish, okay? And every new thing just

30:34people jump on, right? Just tons and I

30:37don't mind that. That's fine. But they

30:40have to be careful and they're not.

30:42Okay? And most people who jump you

30:44there's first off if you're willing to

30:47drop what you're doing as a scientist to

30:48chase something else. What was the value

30:50of the thing you were working on in the

30:52first place? Right? What is motivating

30:54you in order to do this? It's fine if

30:56there's an opportunity particularly if

30:58that opportunity reads upon what you

31:00were doing. But if you're if you're just

31:02like whoosh, let's do this all of a

31:04sudden, right? So the field blew up

31:06overnight. It was it was part of the

31:07blow up of scan probe microscopy in

31:09general with STM with AFM with near

31:13field all of this. But the trouble is is

31:16that most of the people are not careful.

31:19And one of the things I one of my stock

31:21phrases from that error was it's very

31:24easy to get an image and very difficult

31:26to get a meaningful image. So it's very

31:28easy to see artifacts in your images all

31:31the time and thing and this was just

31:33endemic and I kind of felt like every

31:35good paper that we published was the

31:38justification for a 100 pieces of crap

31:41that followed in its wake

31:42>> and everything that I was doing was a

31:45net negative to society and a waste of

31:47the taxpayers's money. Um so that's the

31:50way I still feel about most of science.

31:52>> Do you have here an image taken with

31:54that

31:54>> with nearfield? No, I don't I don't

31:56think I have a nearfield one anywhere.

31:58Uh Nope. Sorry. Yeah.

32:01>> I I mean I have them somewhere. I have

32:03some Oh, actually, you see those two big

32:06ass black books right right above where

32:09he is there.

32:10>> That's my thesis. One of the biggest

32:12thesis in the history of applied physics

32:15at Cornell University. If later on if we

32:18want and I can give you there's plenty

32:20of images in there from my nearfield

32:22microscope at Cornell. Okay. So,

32:24>> and it was written without LLMs.

32:26>> It was Yeah, it was, you know, you know

32:29what? It was written on a Wang word

32:31processor. There wasn't even, you know,

32:33didn't even have a a PC to do it on

32:35then. So, yeah.

32:37>> Yeah. I think this complaint about the

32:39like people publishing essentially

32:41artifacts, right? Because like I've

32:42heard it before actually from an AFM guy

32:44that I'm friends with that

32:46>> he worked on um he had a job for a while

32:48doing like hero experiments at Brooker.

32:50So, they want to show like the power of

32:52the microscope.

32:53>> Exactly. So he would put a lot of effort

32:54into like let's say imaging the helix of

32:56DNA or whatever. But then he was saying

32:58that like yeah if you look at papers

32:59like you're often just looking at

33:01artifacts.

33:02>> Absolutely.

33:03>> Yeah. What's pro microscopy is loaded

33:05with artifacts.

33:06>> Is is there like a do you have like a

33:09hack that would you know help reduce the

33:12amount of artifacts in literature? Have

33:13you ever thought of like a if you if you

33:15could like wave a wand? So, one of my

33:18favorite lines and I and I only learned

33:21it a few years ago, but I think it

33:23explains everything in the world is from

33:26Charlie Mer. Show me the incentives and

33:28I'll show you the outcome.

33:30>> Okay.

33:31>> So, the incentives in science are not

33:34incentives to to to

33:38increase the the store of knowledge.

33:41there to get grants, to um get awards,

33:45to do things of this sort, right? And if

33:49one can do that,

33:50>> receive the Nobel Prize.

33:52>> Yeah. Whatever. [ __ ] Yeah. Yeah.

33:54[laughter] Award. I I can rant forever

33:56about how toxic I think awards are to

33:59science.

34:00>> Um every award that's ever given out,

34:03there's 50 people who think they should

34:04have gotten it and are bitter about it,

34:06right? And it's all subjective. It's a

34:08[ __ ] beauty contest, right? Who's to

34:11science is a collective work of many

34:13many people. Who's to say one guy should

34:15get all theing credit for for what an

34:18entire field has done? Or worse, a guy

34:21who runs a lab of 50 people, right?

34:24These giant super groupoups, right?

34:26Where it's all like, you know, field

34:28marshal at the top and lieutenants and

34:29this and that and that all the way down.

34:32Some some kid down at the bench has some

34:34ID and the guy up here gains all the

34:35credit, right? I mean, what the [ __ ] is

34:38that, right? The whole system is is

34:41frankly

34:42corrupt. Were you already [laughter]

34:44Were you already Is this what you were

34:46thinking when you were leaving IBM? Like

34:48when you No,

34:50>> sorry.

34:50>> Bell Labs. Yes.

34:51>> I was thinking this when I was in

34:52graduate school. Are you kidding me? The

34:54scales fell off my eyes pretty damn

34:56early, right? But I felt that that's

34:59this was a lot of the reason I I didn't

35:01just leave Bell Labs. I didn't just

35:02leave nearfield. I left science, right?

35:04I mean, I was done with it, right? So,

35:07yeah.

35:07>> What did you do? Where's your

35:09>> I worked for my dad. So he he uh timing

35:13was good because um uh he uh um worked

35:18for a machine tool company in Michigan

35:21which um so a machine tool is so

35:25Michigan of course is home of the big

35:26three in the auto industry and they were

35:29producing millions of cars a year. So

35:31that means you need to make millions of

35:32brake calibers calipers a year and you

35:35may need to make a million you know

35:37intake manifolds a year. That's a lot.

35:40Okay. You don't do that by monkeys

35:43typing out Shakespeare by having a

35:45thousand guys each on its own milling

35:48machine and personally doing that. You

35:49have to have automation. So CNC machines

35:52existed then but they were not robust.

35:55Okay. You could you could not produce

35:57the million parts a year with the

35:59technology with CNC computer numerically

36:01controlled machine tools at that time.

36:04So you what what my where my dad worked

36:07is you you actually design the machine

36:11specifically for that part. So I two

36:14years before the car ever is in the

36:16showroom the big three says okay we've

36:19designed this this is this is for

36:22example the brake caliper. Okay, they go

36:25to these different companies and they

36:26say, "Quote for me a machine customized

36:30specifically to make a million of these

36:32brake calipers a year." And so they do

36:34that and that machine will be,

36:37you know, 10 times the size of this

36:39room. Okay? And because it has and it's

36:42working in parallel doing here it's

36:45doing a milling operation here it's

36:46doing a drill here it's tapping or at

36:48different stations as it goes around and

36:50every 30 seconds pop part comes up pop

36:53comes up. So these machines even back

36:55then cost hundreds of thousands to

36:57million dollars and you have a year in

37:01order to get that machine running and

37:03it's got to run 247 on the floor or you

37:06are out of business because they will

37:08never buy from you if you stop the

37:10assembly line because they don't have

37:11that brake caliper.

37:13Okay? So you have to make really really

37:16robust machines. So my dad my dad did

37:18that and then um and then he decided to

37:22make his own company doing that. And so

37:24that happened while I was at Bell. Um

37:27right when right when I started Bell he

37:29had left the other company and started a

37:31competing company. So that was 1988.

37:36You guys are probably too young to know

37:38but 1990s was the era of us boomers

37:41becoming real consumers. The minivan was

37:44invented. you know, the SUV was

37:47invented. It was boom time for the auto.

37:50So, my dad's timing could not have been

37:52better. And so, his company, so by the

37:56time of 97, he had 300 people and was

37:59doing 70 million in sales from zero,

38:01right? and like you know that's great

38:03and so he could af when when I left so

38:05therefore in 95 when I left Bell they

38:10were already on their upswing and he

38:12could afford a leech like me to come

38:14work there and try to see if there was

38:17some way I could use you know my physics

38:19experience and like that to to make

38:21machine tools better. And so, you know,

38:24when I was completely fed up with

38:26science, I always always kind of knew I

38:29would have this as a potential backup

38:31cuz he was always saying, "I want you to

38:32come work for me."

38:35And eventually was like, "Okay, Dad.

38:38Yeah, I'll come work for you, but can I

38:40just kind of try to be the guy who's

38:45your R&D department? Okay. I'm I'm going

38:47to try to think if there are other ways

38:49of doing things or other markets we

38:51could do." And so he was willing to

38:53humor me with that, right? And so and so

38:56I did that and I did that for six years

38:58and and I I developed two projects in

39:02that time. The first one was um when the

39:06when the parts come off the machine,

39:08they had damn well better bolt together

39:10at the rest of the parts. Well, right.

39:13And so they couldn't inspect every part,

39:15but what they do is they take every one

39:17in a zillion parts and they go to a

39:20clean room where there's there's

39:22something called a coordinate measuring

39:23machine, this thing that's on a gantry,

39:25and it has this little sapphire ball at

39:27the end and it touches the part here,

39:28touches the part there D, and it kind of

39:32feels its way around and figures out to,

39:34you know, sort of 10 microns or so

39:38precision where everything is, right?

39:40But you're only looking at one out of a

39:42zillion parts and then you hope

39:43statistically you look at enough parts

39:45and you hope the variability of the

39:47process is good enough. It wasn't. It it

39:50clearly wasn't. Okay. And so what I came

39:52up with was because I was still filled

39:56with optics and still filled with super

39:58resolution, right? um uh you know I knew

40:02in fact one of the two experiments I did

40:05at Bell which led to the idea for the

40:07Nobel Prize was I was the first person

40:10to see single molecules

40:12um at room temperature and furthermore I

40:15was able to localize their positions to

40:18about a 50th of the wavelength of light

40:20because I knew what the energy

40:23distribution inside of the near field

40:24aperture was. So I could fit a profile

40:27to that and therefore figure out where

40:28the molecule was to much better than the

40:31width of the aperture itself by doing

40:33that fit. Right? Just like you could

40:34find the center of a Gaussian to much

40:36better than the width of a Gaussian if

40:38you have enough signal noise and the

40:40nature of that Gaussian.

40:41>> Did you do it with near field

40:42microscopy?

40:43>> I did that with near field microscopy.

40:45>> How do you also prove that it's a single

40:46molecule?

40:47>> Oh, that's easy because they two there

40:49were two things that were really

40:50conclusive on that. I loved that. That

40:53was that was probably one of the best

40:54papers of my life. Um the the first one

40:57is is is it bleaches it doesn't go down

41:00slowly where the molecules bleach.

41:03They're there and then the next instant

41:04they're gone. Okay. So you fried it.

41:07Okay. The other is is that fluorescent

41:10molecules are dipoles. That's how they

41:12work. And so it had a dipole

41:13orientation. So actually what happened

41:16is because of that dipole orientation it

41:19was actually mapping out the electric

41:21fields in that subwavelength aperture.

41:23So I could actually map the field the

41:25fields in the aperture to almost one

41:27nanometer precision as I drove it along

41:29that molecule and I compared it to a

41:32theory that Hans Beta developed in the

41:3440 for what defraction would be through

41:35a subwavelength hole and it matched up

41:37exactly. It was an awesome paper. So you

41:40could see the the double near field

41:42>> distribution and and and I could turn

41:44that around and since I knew from Beta's

41:46theory what the electric field profile

41:48was like I could not only determine the

41:51position of the molecules I could

41:53determine the orientation of every

41:54[ __ ] dipole. [laughter]

41:58So it was really cool.

41:59>> Is it the single dipole or

42:00>> single dipole? No no no it's the floor

42:03fours are generally just a single

42:05dipole.

42:05>> And how do you collect the fluoresence

42:07in that situation? Is it through the

42:08same fiber just coming?

42:09>> You can either come back up or you can

42:11have a detector on the opposite side if

42:13it's on a transparent substrate. Got it.

42:16>> Yeah.

42:16>> Um

42:17>> the other key thing then then was um

42:20having these um uh avalanche photo

42:23dodes. So before that was all photo

42:26multiplier tubes which sucked at both in

42:28terms of quantum efficiency and noise.

42:30But the avalanche photo diode absolutely

42:32changed imaging when that came on the

42:35scene. And so again, it was being in the

42:37right place at the right time. And one

42:38of the other things I would say is one

42:40of the reasons I've been successful as a

42:43scientist is I've always been one of the

42:45first adopters of new technology. I'm

42:47always got my, you know, my sniffing

42:51around for whatever is the latest widget

42:53that I can and h and how

42:55combinatorically it adds with all of the

42:57other widgets and all the other

42:58experience I have to do something new.

43:00>> How would you explain it? How do you do

43:02it? Like to be in the right

43:03>> Well, again, it's it's just it's you I

43:07hate going to conferences, but I go to

43:10conferences to go to the trade shows,

43:12right? Because that's where you kind of

43:13learn what's the new and you talk to the

43:15people. You even learn things that

43:16aren't quite they're not quite ready to

43:18release yet and so forth. And so, yeah,

43:20>> I started actually to do the same.

43:22[laughter]

43:22>> Yeah. Okay, there you go. Yeah,

43:24absolutely. Yeah, I I learned way more

43:26from those guys than I ever learned in

43:28any talks at at, you know, something

43:30like, you know, the optics conferences

43:32in Moscone or whatever, right? Yeah.

43:35>> But what uh how did this uh you were

43:38going to you were going to talk about

43:39how that microscopy work translated to

43:42the like CNC?

43:43>> Oh, yeah. Yeah. Yeah. We got off the

43:45that thread. Exactly. So, um, so instead

43:49of just measuring one part every so

43:51often, I developed this thing that had

43:54about 30 little cheap CCD cameras that

43:57were all around the thing and then I

43:59could measure to sub pixel precision

44:03where say the edge of a part was or

44:05where a hole was located or all of that

44:07and I could do it in a fraction of a

44:10second. That's it could be done for

44:11every part

44:11>> for so therefore we can look at every

44:13part that comes off the machine and we

44:15can reject those parts that don't that

44:18don't fit the the the uh tolerances of

44:21the of the of the blueprint. Right? And

44:24so I built one of those um we put it on

44:27a machine that we had shipped to New

44:29York. Uh and um and what happened it

44:34starts rejecting parts, right? Because

44:37what happened is um it the the

44:41tolerances that the companies would spec

44:44this was the era of of GE and the six

44:48sigma [ __ ] um with Jack Welch and

44:51all of that, right? Six sigma is crazy.

44:54You know what six sigma means in terms

44:55of precision? It's it's like it's like

44:5810us 6 or something. [laughter]

45:01It's it's crazy [ __ ] right? I mean,

45:03nobody can can achieve that type of

45:05stuff, right? So So basically it's

45:07rejecting parts because of this, right?

45:10And um and tacitly everybody knew that,

45:15right? And so with the old method of

45:18just using the the coordinate measuring

45:20machine to to statistically get some

45:22parts, as long as the [ __ ] things

45:25bolted on to to the axle, life was good.

45:29Okay? And and so they speck them so

45:32tight that even if you were a factor of

45:35five off in terms of tolerance, it was

45:38still plenty good that it would end up

45:40bolting onto the thing. Right? That was

45:41the way the world worked. And so what

45:44they did is because it it rejected

45:47parts, they turned it off, right? They

45:50turned it off and then they let the

45:52parts go through and then they measure

45:54them the old way. And so so that was my

45:56first lesson is that okay, well the

46:00thing that is king is is productivity. I

46:04I developed something which hampered

46:07productivity, right? So instead for my

46:10second thing I did with my dad is can I

46:13do something that increases

46:14productivity. Okay I've learned my

46:17lesson. Okay. So there was an

46:20opportunity there as I learned is that

46:23you see in order to have the robustness

46:25that these machines needed to have. They

46:28were incredibly big. Okay. Um there's

46:32you're talking about each each you would

46:34and you it would do stuff in parallel.

46:36So if you have like on one surface eight

46:38holes are being drilled at once, you

46:40make a spindle that has eight spindles

46:44together linked together and in the

46:45exact positions of where those holes

46:47will and it all goes in in parallel

46:49together and it drills all of those

46:50holes at once, right? And so um that's

46:53the only way you get the speed. Um and

46:56so um so you're moving masses for these

47:00in order to make it robust, in order to

47:02make aggressive milling cuts across, you

47:05know, cast iron or something, uh you

47:08need to make things that that are really

47:10heavy. So each station around these

47:14machines might be a ton a piece, right?

47:18Um maybe more because again, you're

47:20you're making three axis moves, right?

47:22So you have three different one axis on

47:25top of the other on top of the other to

47:27make these moves. And so it's like a

47:29rocket, right? Where you have three

47:30stages, right, in order to make this

47:32this type of move. And so the bottom

47:34stage is moving all the upper stages,

47:36right? In order to do all that. So it

47:38has to be even bigger and more more

47:40robust in order to do that, right? So

47:42that's how it worked. Um and um and in

47:46that day, uh you had two options for

47:49trying to move this stuff. You could use

47:51a big electric motor which was tied to a

47:55ball screw and the ball screw would then

47:57move the stage forward or you could use

48:00hydraulics and use a hydraulic cylinder

48:02that would just push it forward. So you

48:03you you could use either a hydraulic

48:06cylinder or ball screw. The trouble with

48:08the hydraulics was it was cheap but it

48:11wasn't very precise. Okay. So generally

48:14what you do is you just move it forward

48:16until you come up against stop but you

48:17couldn't really make milling moves with

48:19this. just kind of drilling moves. And

48:21the trouble with the ball screw stuff is

48:23it was really limited in speed because,

48:25you know, you're putting all of the

48:27force of this tons through a screw,

48:30right? And so the screw could get ripped

48:32up if you don't do it slowly enough,

48:34right? So it meant that a lot of the

48:37most of the time in which you're making

48:39these parts, you're not making chip.

48:41You're just moving the masses to the

48:43point where it's going to make the chip.

48:44And there's an old saying in the

48:46business that if you're not making

48:48chips, you're not making money, right?

48:50And so the duty cycle of actually making

48:53chips was small. And so, but I had

48:57learned enough control theory while I

48:59was in grad school in Bell Labs that I

49:02realized that there was no need to do

49:04all this openloop hydraulics. And there

49:06were some new, again to the things that

49:09are just new, there was some very

49:12impressive servo valves developed by

49:13Bosch in Germany. And so I was able to

49:16do some nonlinear control theory coupled

49:19to those servo valves, coupled to

49:21hydraulic cylinders, coupled to the

49:23final thing, which is which is energy

49:26storage. So normally when you're doing

49:28an electric move, right, is is is if if

49:31I wanted, for example, move that whole

49:33big column of metal forward in a certain

49:36given of t period of time. If I wanted

49:39to to do it in half the time, that means

49:41I'm going at twice the speed. Okay, but

49:45twice the speed means I've got four

49:48times the kinetic energy, MV squared,

49:50right? That I have to put in there. But

49:52I have to put that energy in in half the

49:55time. So the peak power goes up as the

49:59cube.

50:00>> So that's ruinous with an electric

50:02motor, right? Because now I have to have

50:04electric motors that are eight times

50:07bigger to do half the time and their

50:09mass goes up eightfold to have that much

50:13extra horsepower. And so now it's like

50:15having a rocket in which your rocket

50:17fuel is really shitty and your mass

50:19fraction is really bad. Okay, so that

50:22was not the PE. But hydraulics,

50:24hydraulics, you can fit a 100 horsepower

50:26hydraulic motor in the palm of your hand

50:29because basically it's just delivering

50:31what's the original source. And with

50:34hydraulics, they have accumulators.

50:36Basically short-term batteries for

50:39hydraulics. And it's nothing more than a

50:41big ass cylinder which stays remote,

50:45right? And it's got a little nitrogen

50:47bladder in the top. and you use your 20

50:50horsepower electric motor to drive a

50:52pump to to basically compress that gas

50:55up there and now it's a spring, right?

50:58And then when you want to make your

50:59move, boom, you can blast that out at

51:02hundreds of horsepower through the

51:04hydraulic lines and move the load. So I

51:06made this thing, it's somewhere here,

51:08somewhere on one of the walls.

51:10>> Do you do like acceleration?

51:11>> There it is right here. That's it. Okay,

51:14that's fast flexible adaptive servo

51:16hydraulic technology. So you see that

51:19thing there? See that? See this big ass

51:21mass here? A human sits like right here.

51:23See this big thing here? That's a couple

51:25tons. I could move that at 8 gs of

51:28acceleration and position it anywhere

51:30within a meter cubed to 5 micron

51:32precision.

51:33>> That's sick.

51:34>> Yeah, it was sick. It was sick. It's

51:36been taken over since by linear electric

51:38motor technology has this is why you

51:41know Starship started with hydraulics

51:43but they quickly pivoted you know for

51:45the for the uh flaps and so forth and

51:48and and um moving the Raptors around to

51:52electrics but um it was only long since

51:55I left that electrics could get to that

51:58point but man it worked great. It worked

52:00great. It was just amaz I was so proud

52:03of that machine and so I spent three

52:06years developing it and two years trying

52:08to sell it and in the end I sold two

52:12[laughter]

52:12two because it was you well couple

52:17reasons a there's a lot of a lot of push

52:20back from the UAW because we had to use

52:223,000 PSI hydraulics instead of the 300

52:26PSI they were used to. Now, aerospace at

52:29that time was already using 10,000 PSI

52:31hydraulics, and there's no problem with

52:333,000 PSI, but that doesn't mean that

52:35they're not going to fight it anyway.

52:37The UAW was definitely not on board with

52:41that. Um,

52:41>> was it because it would produce more

52:43parts per person, so there's like fewer

52:44jobs?

52:45>> That's possible, but I I think it was

52:47just the bottom. It wasn't just the UAW.

52:50It was also So, so this gets into the

52:53into the human aspect of how orders are

52:56placed and so forth, right? is is that

52:58typically what happens is again we the

53:01the big three say hey we're going to uh

53:04make uh this particular new car here are

53:06the parts here's the part print and what

53:10happens is um the uh the the guy who is

53:15responsible for getting that part a

53:18million parts per year to the company

53:20he's up and coming he's probably about

53:2240 years old he's a middle manager this

53:24is going to help define his career

53:26Right. So, should I go with what's

53:30always worked or should I put my neck

53:32out [laughter] and and do this crazy

53:35thing that this crazy kid who Well, he's

53:37not a kid anymore, but he's wildeyed and

53:39nutty and talking about and I look at

53:41his machine and and you we would bring

53:43people by and and they would look at it

53:45and say, "Okay, here we're going to make

53:46this part for you." And and they would

53:49literally jump out of their skin when it

53:50first moved because it was it was it was

53:53like a hummingbird, you I mean, and it

53:56had to be tied not just to the concrete

53:58floor, but into the rebar in the in the

54:01in the thing for for the for the back

54:04reaction, you know, of the inertia,

54:06right? And so, um, it just scared the

54:09[ __ ] out of people, okay? Particularly

54:12the people who were going to sign the

54:13check in order to buy it, right?

54:16[laughter]

54:16And and so yeah, it and and I I you

54:20know, I I'm a good scientist, but man,

54:23am I a horrible businessman and an even

54:25worse salesman just just it's just not I

54:28cannot I cannot make people feel

54:31comfortable, right? In terms and that's

54:33still a problem today in trying to go to

54:36philanthropy and and raise money is

54:38because I come across as too [ __ ]

54:40crazy, right? and and even

54:43philanthropists who claim that they're

54:45that they're risk tolerant are

54:48fundamentally they don't want to look

54:49stupid, right? And so uh this is a

54:52problem. But yeah,

54:54>> and so what what happened next? So did

54:56you go back to back to science or

54:58>> Yeah. So that's where we then get into

55:00the palm story, right? is is so after

55:03six years of that, I apologized to my

55:05dad for wasting a couple million dollars

55:08of his money and so forth, but the

55:10company was still doing fine. And and um

55:13you know, of course, he wanted me to

55:14stay in in what he always wanted. He

55:16wanted his son to take over his business

55:18someday, but I did he was just he was so

55:22good. He was the inverse of me. He was

55:25he was so people friendly and he could

55:28make everybody feel comfortable and like

55:30him and um he he just he just was really

55:34terrific. There's no way I [laughter]

55:36was ever going to fill his shoes, nor

55:39did I want to, right? It it was I I

55:41wanted to do some I still had in my head

55:44I'm a scientist. I'm a technologist. I I

55:47want to make the warp drive. Okay. I I

55:50this it's not my dream to run a machine

55:52tool company. I wanted I want to build a

55:54warp drive. Okay.

55:55>> So, what happened to the business?

55:57>> Oh, that's a sad story. So, um uh so it

56:01went fine again for several years after

56:03I left, but um then uh um uh you know,

56:10my dad got old. Um the other

56:14technologies, remember I said eventually

56:16um particularly CNC's started to take

56:19over more and more of the business. They

56:21were still doing fine, but my dad was

56:22getting old. Okay. And um and the

56:25business was changing. And of course

56:28through the '9s, not only were the

56:30minivans doing well, but there was

56:31encroachment of the Japanese and so

56:34forth. It was putting pressure on the

56:35big three. And um and so the company the

56:39company was, you know, they got up to 70

56:42and then they stayed around 70 for most

56:44of the time until my my dad retired. 70

56:46million. And um and then it was time for

56:49him to retire. And um uh so he ended up

56:54selling the company and then and then

56:57this was 2005. And then 2008, what

57:00happened in 2008?

57:02>> GM goes bankrupt.

57:03>> The great the great financial crash. You

57:06you got it guys. Okay. It was it was it

57:11was a bloodbath in Michigan.

57:14>> A bloodbath. unemployment hit over 30%.

57:18Okay. Um my dad's company, the the

57:21people who went into receiverhip, 300

57:23guys,

57:25>> 300 guys, many of them who I knew out

57:28out of a job. I was by then a Janelian

57:31and living high on the hawk. Okay. Um

57:34but a lot of those guys and it was a

57:36scramble for those guys. A lot of them,

57:38you know, just odd jobs here and there,

57:40whatever. Um, we never got the reckoning

57:44we deserved for what happened in 2008.

57:47There are not enough heads who rolled

57:48for that wholeing thing. Basically, you

57:51know, moral hazard went out the window

57:54and allowing people to actually [ __ ]

57:56take take the penalty for what theying

57:58did never happened.

58:00>> Um, it's one of the things that

58:03>> I I I kind of feel like the United

58:05States lost the threat of what it was

58:07starting in 2008 and has not recovered.

58:10that there is no there is no reckoning

58:12for for um for bad behavior financial or

58:16otherwise. And I think it really started

58:18right then. Um and again, you know,

58:22eventually those guys found things, but

58:24but you know, it was it was it was

58:26brutal. Really brutal. Um so anyway,

58:29let's let's switch topics from that. But

58:32yeah. Yeah.

58:33>> Coming back to the poem.

58:34>> Yeah. I just want to say one more story

58:36about that, right? Is is another another

58:38thing that

58:40In some ways they they I while certainly

58:43I I I I believe the financial types have

58:46primary responsibility for this, the big

58:50three in the unions did not help

58:51themselves. And a story for that was on

58:54one trip when I when I put in one of my

58:57vision systems. Um uh this is a a a um a

59:03transmission plant in Cooko, Indiana.

59:06Um, Cooko, Indiana is literally in the

59:09middle of nowhere. It's in corn fields.

59:11It's right next to Seymour, Indiana. Do

59:13you guys know John Melanchamp, the

59:15singer, at all? You guys,

59:17>> but I I did my page at Purdue, so we had

59:19a lot of

59:21Yeah. Right. So, Cooko is, trust me,

59:23Urbana Champagne is New York City

59:26[laughter] compared to Cooko. Okay. So,

59:29so anyway, um, so I made the grievous

59:34error, you know, because I went with

59:36some of the service guys who were doing

59:37other things on that machine. I made the

59:40grievous error of taking one of our, we

59:43had a fleet of cars that the service

59:45guys would take out, right? Um, when

59:47they went on jobs, I took a Ford to a

59:50Chrysler plant. Okay, so I'm working all

59:53day trying to get my machine up and

59:55running in this Chrysler plant. I have

59:58never seen a closer approximation to

1:00:00hell in my life. It was August. It's 120

1:00:05degrees inside of this plant. The plant

1:00:07is probably like 250,000 square feet,

1:00:09chocked full of machinery and people.

1:00:13Probably 120 dB in there. Every

1:00:16frequency from subsonic grinding to

1:00:18supersonic milling. Um

1:00:21>> stuff for the brain.

1:00:22>> There there's coolant mist in the air

1:00:24everywhere. Chips. Um there were there

1:00:29were a thousand people working in that

1:00:31plant and I worked there from 6:00 in

1:00:34the morning until 11:00 at night and and

1:00:38um and I go out to my car bone tired

1:00:41ready to go to the hotel and there was

1:00:46it was it was 90° outside that day and

1:00:49so there was somebody had poured chili

1:00:52all over my windshield that had dried

1:00:55because I took afford to a Chrysler

1:00:57plant. Can you imagine if I took a

1:00:59Toyota [laughter]

1:01:03but but the thing I learned from there

1:01:05and this is something which I'm sorry

1:01:06still stick. I am never going to be an

1:01:08academic. I I am an academic. I have the

1:01:12[laughter] first the first class of the

1:01:14day I have to teach this. There's no way

1:01:16I would ever be at this place if it were

1:01:18not for my wife who wanted to be here.

1:01:20Okay.

1:01:20>> You're trying so hard not to be an

1:01:21academic.

1:01:22>> Yes. But

1:01:22>> it pulls you back. What? [laughter] But

1:01:24but but but the but the but the thing is

1:01:27is um those people work their ass off

1:01:30and they work under just miserable

1:01:32conditions, right? Eventually that

1:01:34that's gone. That transmission plan is

1:01:36also gone. All of those people were

1:01:38thrown out of work in the end, right? As

1:01:40as the every as everything changes and

1:01:42and more and more stuff gets done

1:01:44overseas, right? I get it. I believe in

1:01:47competition, okay? But um those people

1:01:50worked really really hard and then all

1:01:54of a sudden skipping past the palm I'm

1:01:57at Janelia working for Hughes a1 billion

1:02:00dollar building that's just luxuriously

1:02:03appointed the first thing that happened

1:02:05when I was the first group leader on

1:02:06staff and when the first six of us were

1:02:09together they took us in a room showed

1:02:11us six different executive desk chairs

1:02:14and said pick the one that you like the

1:02:16best when I was in my dad's company. I

1:02:19worked on a nogahhide stool that had the

1:02:22juke poking out of it into my ass for

1:02:25six years because it was good enough. It

1:02:28was good enough. Money mattered, right?

1:02:30And all of a sudden, I'm in La La Land.

1:02:31And then I start going to conferences

1:02:33again, right? And I go to these

1:02:35conferences. They can't be put the

1:02:38conference in Cooko, Indiana. It has to

1:02:39be in [ __ ] Creed or someplace, right?

1:02:42So, everybody flies to Creed. They're

1:02:44eating their lobster and they and they

1:02:46talk about the dumb rubes and the

1:02:48flyover states who vote for Trump. Well,

1:02:50why the [ __ ] do you think they're voting

1:02:52for Trump? This is this is Marie

1:02:54Antuanette all over again. They have no

1:02:57most academics have no appreciation, no

1:03:00gratitude for the grant money that they

1:03:03get and not an understanding

1:03:05[clears throat] that that money comes

1:03:07from those people in Cooko, Indiana and

1:03:09a 100 million other people just like

1:03:11them all around the country who scrabble

1:03:14every day and they think that their tax

1:03:17money is being well used to support

1:03:20research.

1:03:22So anyway, sorry

1:03:24there was many people.

1:03:26>> Is that the reason why you don't apply

1:03:28for apply for the grants?

1:03:30>> I I have never never wr written a grant

1:03:32in my life. I will go to my grave never

1:03:35never writing a grant.

1:03:36>> And it that's only part of it. The other

1:03:38part of it, of course, is is that is

1:03:41that peer review is toxic, right? It it

1:03:44enforces conformity. They never want to

1:03:47take a risk, right? Because you if you

1:03:49stick out, your peers are never going to

1:03:51support that. They they all they all

1:03:54want to just support the stuff that

1:03:56everybody else already does, right?

1:03:58That's how it works. Again, show me the

1:04:01incentives and I show you the outcome.

1:04:02They are not incentivized to support

1:04:05crazy ideas, right? So, you don't get

1:04:08crazy ideas. you get. But

1:04:12the lack of gratitude of academics

1:04:15towards the money they have or or a

1:04:18willingness to realize that every penny

1:04:20they spend is off of off of the labor of

1:04:24people doing all sorts of [ __ ] I I I go

1:04:29I have a a a vacation house near and I

1:04:32cross the the central valley every

1:04:34couple every couple weeks. the people

1:04:37working so hard in the fields there,

1:04:39particular again in the summer and and

1:04:41the smell from the from the the

1:04:43pesticides and the and the and the

1:04:46fertilizers and so forth. And Jesus, I

1:04:51it it it it just it just drives me crazy

1:04:53that that people don't realize the basis

1:04:56of of of what everything of of of of the

1:05:00comforts that we have, where the hell it

1:05:02comes from and and or or an appreciation

1:05:05of of of all of the technology behind. I

1:05:08I wrote I wrote a tweet just the other

1:05:10day about uh about SiriusXM, right? and

1:05:13how much I love being able to to not

1:05:16have to be just limited to FM and CDs

1:05:20and tapes like when I was younger, but

1:05:22have this infinite variety of musical

1:05:24sources and so forth and the technology

1:05:27in order to do that, right? in or in

1:05:29order to in order to have

1:05:3310,000 satellites going on in orbit and

1:05:36the switching from satellite to

1:05:37satellite and a phased array antenna I

1:05:39have in my backyard for just a couple

1:05:41hundred bucks and all of this stuff and

1:05:43people just don't realize that there's

1:05:45magic all around us, right? And and and

1:05:48they're completely oblivious to that

1:05:50magic, completely un ungrateful or un

1:05:53unappreciative of that magic. they

1:05:55expect it and it's so easy to lose it if

1:05:59we don't if we don't have gratitude for

1:06:01it or an understanding of it and so it's

1:06:04just nuts. Anyway, I'm going off topic.

1:06:07We haven't even gotten to Palmer or

1:06:08anything [laughter] and we're way off

1:06:10the thread of microscopy and have been

1:06:12for quite some time. So, [laughter]

1:06:15>> this is good. I I actually I I uh I

1:06:18worked in Michigan a bit uh where? So I

1:06:21was I had like built a factory in China

1:06:24and I was then directed by the CEO of

1:06:27this company that had acquired my

1:06:28company and I built the factory for them

1:06:30in China and then he was like a

1:06:32>> he was from Michigan and he wanted to

1:06:34bring manufacturing back there. Oh,

1:06:35cool.

1:06:35>> So we tried to move the factory to

1:06:37Detroit and this was like 2015ish 2016.

1:06:41It was probably 2016 and that was like a

1:06:44I when I tried to do it in Detroit I was

1:06:46like yeah I think some of these things

1:06:48had to go out of business.

1:06:49>> Yeah. So we ended up we failed to do it

1:06:51in Detroit and we ended up doing it in

1:06:53West Michigan which was much better. And

1:06:55the explanation

1:06:56>> or

1:06:57>> uh it was Grand Rapids. The explanation

1:06:59that I got which I'm curious like what

1:07:00you think of this is that

1:07:02>> the flavor of Christianity in West

1:07:04Michigan

1:07:05>> this is just what I was told so I never

1:07:06verified this but it doesn't allow you

1:07:08to join a union because you can only be

1:07:10in one organization which is like your

1:07:12>> I think it's like some type of Baptist

1:07:14church or whatever. I don't know.

1:07:15>> And so because of that, they never had

1:07:17like the same penetration of unions. And

1:07:19so people were just much more

1:07:21reasonable. Like when I worked with

1:07:22them, they were flexible, you know, just

1:07:24the typical like like reasonable.

1:07:25Whereas like in Detroit,

1:07:27>> everybody was kind of there was this

1:07:29like goo you would move through, you

1:07:31know, when you tried to do anything. You

1:07:32would just be in a room with people.

1:07:33>> I would call it solid cement,

1:07:35>> something [laughter]

1:07:36like that. And I was like, "Yeah, I kind

1:07:38of get why this didn't work out."

1:07:40kind of like,

1:07:41>> you know, the whole thing felt very hard

1:07:44to get through. So, and and I I think

1:07:47the Ohio plants that started and and the

1:07:49ones in Tennessee, it it was a new story

1:07:52then, but there was a lot of legacy

1:07:53stuff obviously in southeastern

1:07:55Michigan, right, that made it extremely

1:07:57difficult.

1:07:58>> Yeah. I think it's just hard to keep a

1:08:00culture like obviously Detroit was

1:08:01incredible for whatever 50 years or

1:08:03however long, but it's hard to keep that

1:08:05culture, right? people got.

1:08:07>> When I was born, Detroit was the fifth

1:08:09largest city in the United States.

1:08:11>> You know, it had 3 million people. Okay.

1:08:13>> Now it has 600,000.

1:08:15>> Yeah. Yeah.

1:08:16>> Yeah.

1:08:16>> Wow. It's crazy.

1:08:17>> Yeah.

1:08:18>> The 632 Nm podcast doesn't have any

1:08:21sponsorships. It's an art project and

1:08:23we're going to keep it that way. But we

1:08:25wanted to do an unpaid advertisement for

1:08:27Tesla Autopilot. Both of us use it.

1:08:29We've both done probably over 1,000

1:08:31miles on autopilot now. And it's just an

1:08:33incredible technology. It's gotten

1:08:35amazing over the last two years and I'm

1:08:37personally very passionate about it

1:08:39because I got into, you know, was hit by

1:08:41somebody who ran a red light and my

1:08:42pregnant wife was in the car and it's

1:08:44just something that really shouldn't

1:08:45happen anymore. Like we have the

1:08:46technology to stop this. You know, if

1:08:48that person was driving a Tesla, it

1:08:49wouldn't happen. So, luckily it all, you

1:08:51know, it all turned out fine.

1:08:52>> Driving in Boston is a hell and I think

1:08:55that definitely helps with finding the

1:08:57right way. It's mindbending and

1:08:59life-saving technology. I highly

1:09:01recommend. You have your hands free,

1:09:03your mind free, you can think. Yes, you

1:09:05have to like pay attention to the road

1:09:06to an extent, but it really drives

1:09:08itself. But they really improve it all

1:09:09the time. I don't think I've had to take

1:09:11over. Probably in the last 6 months or

1:09:13something like that and does literally

1:09:14all of my driving is on autopilot.

1:09:16Anyway, go out and try this. I I think

1:09:18every car should have it, not just

1:09:19Tesla. Like this software should be in

1:09:21every car on the road.

1:09:22>> If you are a fan of the 632 nm podcast,

1:09:26you most likely love quantum computers.

1:09:29And now I'm in the lab at Quera

1:09:31Computing where we build the most

1:09:33advanced neutral atom quantum computers.

1:09:35We have plenty of job openings and I'm

1:09:38hiring for the position of the quantum

1:09:40machine builder. If you have a relevant

1:09:42skill set and want to contribute to the

1:09:44race for building the first fall

1:09:46tolerant quantum computer, join us and

1:09:48enjoy the rest of the episode.

1:09:52>> How did you end up building a microscope

1:09:54in a living room in a living room? All

1:09:55right. So now we'll pivot. All right.

1:09:57So, so picking up that thread. So, I

1:10:00failed it for my dad, right? And I left

1:10:02and and so I was unemployed, you know, I

1:10:06was unemployed for about a year after I

1:10:08left um after I left Bell and I was

1:10:11unemployed for two years after I left uh

1:10:14my dad's company. And the first it's

1:10:16like, well, what am I going to do? I

1:10:18blown up my scientific career and I've

1:10:20blown up my backup plan of working for

1:10:23my dad. Is it good to be unemployed to

1:10:25get some fresh ideas?

1:10:27>> That's a major understatement. By far

1:10:30the best periods of my life were my two

1:10:32periods of unemployment. So for example,

1:10:35getting to the palm, the first key idea

1:10:39that led to Palm and in fact the Nobel

1:10:42committee cited two papers for why I was

1:10:44going to share that prize. Both of them,

1:10:47one each was from each of those two

1:10:49periods of unemployment and they were

1:10:52done by me and me alone. Well, not know

1:10:55the second one was with Harold, but you

1:10:56know the the the key was that um that I

1:11:00was unemployed in both cases. So, so the

1:11:03first one is um after I left Bell and I

1:11:07was trying to think of well, am I going

1:11:09to work for my dad? What am I going to

1:11:10do? My wife, my first wife was at Bell

1:11:13and still working. So, at least we

1:11:15weren't going to starve. And we had had

1:11:17a baby by that time. And so, the baby's

1:11:20like 6 months old. And um so I'm a house

1:11:23husband. And so, I was pushing the baby

1:11:26in a stroller. And there were two

1:11:28remember I said I just had this hit of

1:11:31streak of hits with with with uh

1:11:34nearfield. Um, one of the I mentioned

1:11:36the single molecule one already, but

1:11:38another was one I did with my friend

1:11:39Harold where we looked at um, again

1:11:42because Harold was a low temperature

1:11:44physicist, we were looking at

1:11:46semiconductor lasers at cryogenic

1:11:49temperatures to understand exactly where

1:11:52in these quantum well layers the light

1:11:55emission was occurring. And um, people

1:11:58have studied that optically for ages and

1:12:01understood the spectrum of the emission

1:12:03and like that. But when we did it with a

1:12:05near field probe, we found out that that

1:12:08spectrum, which normally looks like a

1:12:09normal spectrum, you know, it's got

1:12:11little humps on them like that. When you

1:12:13look at it with nearfield, it completely

1:12:15broke up into sharp spectral lines.

1:12:17Boom, boom, boom, boom, boom, boom. And

1:12:19every time you moved the probe by even

1:12:2050 nmters, the lines would change

1:12:23dramatically.

1:12:24>> And so what we found was that um that

1:12:27the light wasn't formed anywhere. It was

1:12:30formed in discrete spots which were

1:12:32usually just different points of

1:12:34roughness, single monollayer changes in

1:12:37the roughness like potholes in the

1:12:38quantum wells that would change the

1:12:40quantum confinement slightly and hence

1:12:41change the the spectrum of the emission.

1:12:44And even though even with our nearfield

1:12:46probe that was 50 nanometers,

1:12:49you know, those those potholes were too

1:12:51close together to resolve, but because

1:12:54the emission from them was so narrow,

1:12:58>> you could still resolve them in a

1:13:00multi-dimensional space based on

1:13:02wavelength is the additional dimension.

1:13:04So you could individually study them

1:13:06because you had enough spectral

1:13:09resolution in the one dimension coupled

1:13:11with enough spatial resolution thanks to

1:13:13the near field in the other dimensions

1:13:15to resolve all of them.

1:13:16>> Interesting.

1:13:17>> So I was pushing and then the second one

1:13:20was the fact that I was the first guy to

1:13:22be able to see single fluorescent

1:13:24molecules and then localize them to

1:13:27subways and dimensions. So while pushing

1:13:29the daughter in the stroller, it just I

1:13:32wasn't really even thinking about it,

1:13:33but it literally just popped from

1:13:35nowhere that you could combine those two

1:13:38ideas. And so if I had some way, say the

1:13:41molecules

1:13:42I know I knew fluorescent molecules at

1:13:44room temperature had broad spectrum, but

1:13:46say they didn't. Say that I was at a

1:13:48cryogenic temperature and they were

1:13:49equally sharp, right? Well, then I could

1:13:53then at low temperature with a

1:13:55spectrograph see these single molecules

1:13:58and with my nearfield tip I could see

1:13:59just a few of them at a time underneath

1:14:02the thing and they'd be isolated because

1:14:03they're different wavelengths. So I

1:14:05isolate them. But now I can find the

1:14:07center of the emission even though it's

1:14:0925 nanometers big with the near field.

1:14:11Hell, I could have done it with with

1:14:13regular light, you know, defraction

1:14:15limited light, but I can still find the

1:14:17center of emission to much better

1:14:19precision than the width of the

1:14:21emission. Okay? And so then I could find

1:14:23the position of every molecule to

1:14:25nanometer precision. So this would be a

1:14:28path to super resolution, okay? Is that

1:14:31you could do. So I published that paper

1:14:33in optics letters and I described how

1:14:36you might do this at cryogenic

1:14:38temperatures and so forth.

1:14:40But [clears throat]

1:14:42it would have been a hero experiment. I

1:14:43could have gone back to Harold who was

1:14:45still at Bell. I was still in the

1:14:46neighborhood. I because my wife was

1:14:49working there. I could have tried to do

1:14:50that experiment but it would have a been

1:14:52a hero experiment to do it. And the and

1:14:54the second thing is you're [ __ ] at

1:14:56four degrees Kelvin. there's not a lot

1:14:58of biology, live cell biology to study

1:15:00at Fort Kelvin, right? And so and so uh

1:15:04um and so I just published the paper and

1:15:06left it at that. And then that's when

1:15:09then I went to work for my dad. Okay. So

1:15:11fast forward six years, we talked my

1:15:13dad's story, I'm back unemployed. Okay.

1:15:16And um and

1:15:18>> so it's like no one really take it like

1:15:20seriously.

1:15:22>> Actually, there were there were a couple

1:15:23papers. So there was um this guy

1:15:26Brackenhof um in the Netherlands used a

1:15:30con focal microscope around 2000 or so

1:15:33to cryogenically look at um like six

1:15:37florores inside a volume and and and see

1:15:40them there. Right? So that was really

1:15:41like the first and there was a couple

1:15:43other things where people used like

1:15:44blinking of molecules to see two

1:15:46molecules in one spot but not at the you

1:15:49know if in order to have super high

1:15:52resolution you need to have lots and

1:15:54lots of molecules to decorate your

1:15:55sample or else it's just a bunch of dots

1:15:57right so this is something called the

1:15:59Nyquist limit right is that you have to

1:16:01sample at at least half of the finest

1:16:04resolution you want to have a complete

1:16:06picture what's going on so that means

1:16:08you need a mole if I want to have 50 nm

1:16:10ter resolution. I need a molecule every

1:16:1225 nanometers localized to a few

1:16:15nanometers to have resolution at that

1:16:16level. So you need to have lots and

1:16:18there was no good way to get lots in the

1:16:20same defraction limited region at the

1:16:22time. How would you like sample them to

1:16:24like uniformly cover the space?

1:16:28>> Again, you would you decorate them at

1:16:30high density. But at at the original

1:16:32concept, it's I I generalized the

1:16:34concept even that paper to have any sort

1:16:37of discriminating

1:16:39third dimension. If there's sufficient

1:16:41resolution in that discriminating third

1:16:43dimension, whatever that means is it

1:16:46could have been the lifetime of the

1:16:47molecules. It could have been the

1:16:48polarization of the molecules. It could

1:16:50be then you can discriminate in that

1:16:52space but you but you have to have the

1:16:55more the higher the density of molecules

1:16:58the more resolution you need in that

1:17:00third dimension to discriminate them.

1:17:01Okay.

1:17:02>> Was there like a possibility to make a

1:17:03single layer or they were just like

1:17:05stack on top of each other? No, no, even

1:17:06in a single layer there's, you know, the

1:17:08defraction limit is big, uh, you know,

1:17:09100 times bigger than the molecule,

1:17:11right? And so,

1:17:12>> and so you need to be able to

1:17:14discriminate, have resolution of a 100

1:17:17times better than whatever you had in

1:17:19the third dimension in order to do that

1:17:21discrimination to say that this molecule

1:17:23is different from this one within that

1:17:25same spot. Right? So, um, so there

1:17:28wasn't a great way to do it. maybe

1:17:30cryogenically could have done it. But

1:17:32then fast forward and during my second

1:17:36round of unemployment. Um well, first

1:17:39it's why go back to science? I hated

1:17:41science. I hated everything about

1:17:43academia, right? Is I think I've made

1:17:45that clear by this point. Um and um but

1:17:49the the the key for me was um I realized

1:17:54that I really miss science. I [laughter]

1:17:57>> It's like love and hate.

1:17:59>> It definitely It's love hate. It's

1:18:01definitely I really missed doing

1:18:03science. Okay. Um and I wanted to see if

1:18:06there was some way somehow I could get

1:18:07back to doing science and so so I

1:18:10reconnected with Harold and dur after I

1:18:13left in in ' 95. Um this was this was

1:18:17the time period in which um uh you know

1:18:20in '84 is when they broke up the

1:18:22monopoly that that provided the

1:18:24financial underpinning for Bell Labs and

1:18:26and so they they continued to support it

1:18:29through the mid90s but um by then they

1:18:32were starting to think they need to be

1:18:35uh financially relevant and contributive

1:18:37to AT&T's bottom line. And so um then

1:18:41that so they among other things they

1:18:44purchased national cash register right

1:18:47and then so so the the the the

1:18:51environment changed so quickly at Bell

1:18:52that when I started in 88 in in in 90 I

1:18:57had another scientist another friend

1:19:00come by and said you know what you are

1:19:02doing is technology and that doesn't

1:19:04really matter here at Bell the only

1:19:05thing that matters is how many fsrev

1:19:07letter papers you publish Okay. So, fast

1:19:10forward two years after that after the

1:19:14NCR purchase and the head of of Bellab's

1:19:17research, Arnold Pensas, who did the the

1:19:20background theory for the Big Bang and

1:19:22won the Nobel, every year gives a gives

1:19:25a um a State of the Union address or

1:19:30something. And he used my work as the

1:19:32sort of applied work that we should all

1:19:34aspire to because I had had my first few

1:19:37hits then, right? A couple years after

1:19:39that with the National Cash Register

1:19:42purchase a right after I left, Harold

1:19:45was asked to spend some fraction of his

1:19:47time not doing low temperature STM, but

1:19:50instead see if he could figure out how

1:19:51to use spectroscopy to figure out which

1:19:54fruit was in the shopping cart so he

1:19:55wouldn't have to put those little

1:19:56fourdigit stickers on it and would

1:19:58automatically know. That's how quickly

1:20:00the culture changed. Okay. [laughter]

1:20:03So, so anyway, so there was this

1:20:05diaspora out of Bell and by 2000 The

1:20:08Henrik Shonne scandal was the final nail

1:20:11in the coffin. And so everybody was gone

1:20:14after that, right? And so Harold had

1:20:16gone into working for a company that

1:20:19makes test equipment for the disc drive

1:20:21industry in San Diego. But so I

1:20:23reconnected with Harold cuz he's been my

1:20:26best friend forever. Um and uh and you

1:20:30know, he was feeling dissatisfied to a

1:20:33degree, right, with that. and we was

1:20:35thinking, well, he'd like to get back

1:20:36into science, too. So, we started going

1:20:39to different national parks where we

1:20:41would meet up and, you know, hike around

1:20:45and think about ideas and and

1:20:48[clears throat]

1:20:49I developed I decided, of course, I

1:20:52couldn't get away from microscopy. I I

1:20:54knew I didn't want to do microscopy, but

1:20:56then I started to think about it and and

1:21:00uh the thing that changed my mind was

1:21:02starting to read the scientific

1:21:04literature for the first time in a

1:21:05decade and I ran across the paper on

1:21:09green fluorescent protein. Green

1:21:11fluorescent protein came out the exact

1:21:14same time I quit Bell Labs and I did not

1:21:17look at the scientific literature at

1:21:19all. The idea that you could snip a

1:21:21piece of DNA from a glowing jellyfish

1:21:23and have that attached to any protein of

1:21:25interest in a live cell.

1:21:27>> My jaw was down on the ground for a week

1:21:30after learning that in 2003.

1:21:34Okay.

1:21:35>> I was like, "Oh my god." Because one of

1:21:37the hardest I tried so hard to do near

1:21:40field biology, but in addition to all

1:21:42the problems with the microscope, you

1:21:44just couldn't decorate the fluorescent

1:21:46molecules onto the proteins at high

1:21:49enough density to to do super

1:21:50resolution. Nor because you were

1:21:53bringing them in exogenously. A lot of

1:21:55times they'd stick to the things that

1:21:56aren't the protein you want. The idea

1:21:58that finally you have 100% certainty

1:22:01that that glowing spot is the protein.

1:22:04It was like, "Oh my [ __ ] god, this is

1:22:06going to revolutionize imaging." Okay. I

1:22:09said, "Shit, I got to do imaging." So I

1:22:12tried to come up with an idea and I came

1:22:13up with an idea about how to interfere

1:22:15multiple light beams from different

1:22:17directions to create different types of

1:22:19optical lises. This is used now a lot in

1:22:22in the AMO field, right, to do stuff.

1:22:25But at that time it wasn't. And so I

1:22:29came up with these theories of optical

1:22:30latises to make a massively multif focal

1:22:33exitation field that I could try to do s

1:22:36you know parallelized three-dimensional

1:22:38imaging of living cells. And so I tried

1:22:41to get Harold to come in with me on that

1:22:43idea. Um, and and he said, "It's it's a

1:22:47nice idea and I'd like to help you, but

1:22:49I'd be chewing your cut cuz it's your

1:22:51idea." Um, but he tried to help me. And

1:22:54so, um, he there was so part of that

1:22:58diaspora out of Bell Labs was a lot of

1:23:01good people were everywhere, people I

1:23:03knew. I had contacts everywhere because

1:23:06I had made a name for myself at Bell and

1:23:08these people had all gone to academia or

1:23:11or government labs or whatever. And so,

1:23:14man, did I use that contact network.

1:23:16[laughter]

1:23:17So, so um one of them was was Harold was

1:23:20friends with this guy Greg Boinger who

1:23:22had been at Bell and had since become

1:23:24head of the National High Field Lab in

1:23:27Tallahassee. And so, Greg had been

1:23:30trying to recruit Harold to become a

1:23:32scientist at the Magnet Lab. And um in

1:23:36an earlier visit, he met this weird

1:23:38dude, Mike Davidson. Um, and Mike's job

1:23:41was to use microscopes to look at the

1:23:44grain boundaries in the wires that would

1:23:46go into the magnets because if they

1:23:48weren't the right type of structure, the

1:23:49magnets would just blow apart when

1:23:51you're running at 30 Tesla, right? Um,

1:23:53and so, but Mike's real love was live

1:23:58cell imaging. And so Mike had made

1:24:00himself independently wealthy by using

1:24:03his microscopes to look at cocktail

1:24:04mixes under polarized light microscope

1:24:07and print those on neck ties and then

1:24:09sell them online. And so he made

1:24:11millions off of this. Okay. And so then

1:24:14Mike um was uh um uh used that money to

1:24:21follow his dream of doing live cell

1:24:22imaging. And live cell imaging meant GFP

1:24:25in that era, right? Green fluorescent

1:24:27protein. And so Mike became a cloner.

1:24:29And in fact, he because Mike is was kind

1:24:33of self-taught, barely made it through

1:24:35college, self-supported all the time. He

1:24:38took all these kids who were flunking

1:24:40out of Florida State and hired them as

1:24:42techs to be cloners to start knocking in

1:24:45fluorescent proteins on everying protein

1:24:47under the sun. So he had the world's

1:24:49biggest library of fluorescent protein

1:24:52fusions, about 3,000 different knock-ins

1:24:54by that time. And so we went to visit

1:24:57and so Harold thought, well, you know,

1:24:59Mike's into live imaging. Maybe he'll

1:25:00give you some space in the lab to do

1:25:02this. So we went there, we we hit it

1:25:04off. Great. And um and he then told us,

1:25:07you know, I I said how much I love

1:25:09fluorescent proteins. And and he said,

1:25:11"Yeah, well, there's this kind of weird

1:25:12one that just came on the scene. It's

1:25:14what happens is when you when you shine

1:25:16the normal 488 laser on it, it doesn't

1:25:19glow green. Nothing happens. But if you

1:25:21first shine 405 light on it, it

1:25:24activates it. And then it glows green

1:25:26>> and so okay well that's interesting. So

1:25:28we finish the trip and Harold and I are

1:25:30in the airport in Tallahassee and we and

1:25:32it hits us both at the same instant. Oh

1:25:35my [ __ ] god. If you turn down that

1:25:37violet light really low only a few

1:25:39photons at a time are going to hit the

1:25:41sample. You're going to only

1:25:43photoconvert a few of the molecules.

1:25:46Statistically they'll be separated.

1:25:47Their fuzzy balls will be separated by

1:25:49more than the fraction limit even in a

1:25:51regular optical microscope. And then you

1:25:53can find the center of those fuzzy

1:25:54balls. Then you turn those molecules

1:25:56off, turn on another subset, do that

1:25:59again and again and again, and you get a

1:26:01super resolution image with a standard

1:26:03microscope. No funny fancy tricks at all

1:26:06other than having this photosw

1:26:08switchable fluorescent protein. And so

1:26:11we said, well, [ __ ] that optical lattice

1:26:14[ __ ] Um, let's do this. Okay. And now

1:26:17now Harold's fully invested because it's

1:26:20our shared idea, right? And so um and so

1:26:24we were terrified. We were like my god

1:26:27that paper is like is like four, five,

1:26:29six months old. Why hasn't anybody done

1:26:31this already? I I I pitched that idea

1:26:33back in '94. How how this would be

1:26:36possible if you had something like

1:26:38nobody had connected the dots. Um it was

1:26:41like we were, you know, we're nothing.

1:26:43We have no lab, nothing, right? And

1:26:45anybody who had a reasonable lab, this

1:26:47would be trivial to do. And it's like

1:26:50holy, we got to do it now. And so, um,

1:26:53and so, uh, you know, we figured we'll

1:26:56do it at Harold's place because when

1:26:57Harold left Belle, he took all of his

1:26:59equipment with him because Belle didn't

1:27:00have any use for it. So, a lot of that

1:27:02was a lot of the spectroscopy and optic

1:27:04[ __ ] we did when we did our our quantum

1:27:06well experiment back in the day. So, we

1:27:07had some lasers and detectors and other

1:27:10[ __ ] and and then we had to put about

1:27:1250k each of our own money into it for

1:27:14other [ __ ] that didn't happen and

1:27:15machining the microscope. And there it's

1:27:18that guy right there. That's Harold's

1:27:20living room right there. That's where we

1:27:22did it. Okay.

1:27:22>> You said EMCCD was the

1:27:24>> the EMCCD was the big ouch because that

1:27:26was 30 grand and I I struck a deal with

1:27:28Andor to at least return it to them at

1:27:31half price if it didn't work out. But

1:27:33that was the whole thing. We put that

1:27:34together and in under we went from the

1:27:37idea to shipping it. So that was the

1:27:39other part of it is we didn't know [ __ ]

1:27:41about biology. We didn't know how to

1:27:43clone a cell to get fluorescent proteins

1:27:45in or anything. We needed a [ __ ]

1:27:46biologist to work with us. Mike was

1:27:48willing to help and that was great. Um,

1:27:51but I had another in which is that. So,

1:27:54who invented this photoactivated

1:27:56fluorescent protein? Two biologists who

1:27:59were at the National Institutes of

1:28:00Health. Well, guess what? I had a friend

1:28:03from Bell Labs who had gone to NIH,

1:28:05[laughter]

1:28:05>> Bell Labs Network,

1:28:06>> Bell Labs Network. So, so I I I I I

1:28:10called up I called up uh uh him, Rob

1:28:13Tiko. He was an STM guy um and who had

1:28:17pivoted to to other things and and I

1:28:20said um uh um you know uh uh I'm trying

1:28:25to find a job. Um can I come give a

1:28:29would you host me to give a talk? And um

1:28:32and in in fact I had set that up before

1:28:34I I I I misspoke because actually I

1:28:37spoke to him before we had the idea for

1:28:39this because it was one of the many ways

1:28:41I was trying to sell my optical lattice

1:28:43idea and get into a lab is I asked him

1:28:45could I give a talk to try to pitch my

1:28:47lattice idea. Okay. And so, um, when the

1:28:50day came to give that talk, um, I I I

1:28:54said, "Rob, would you please ask these

1:28:56two people, George Patterson and

1:28:58Jennifer Lipincot Schwarz, to come to my

1:29:00talk because I'd really really really

1:29:02[laughter] like to meet him." And so

1:29:05they came to the talk and and I I went

1:29:07to them afterwards and can I take you

1:29:09guys to lunch? [laughter] And I and I

1:29:11said, "Okay." And took him to lunch. I

1:29:13said, "My buddy and I have this crazy

1:29:15idea and we need you guys. we need you

1:29:17guys badly for your for your

1:29:18photoactivated flare surprise. And

1:29:20Jennifer said, "Fantastic. Sure. Bring

1:29:22it by." And so, um, we were already

1:29:24building the scope then. And so, from

1:29:26conception to shipping that thing out to

1:29:29NIH was 3 months within

1:29:32>> Oh, so you you ship the

1:29:34>> We ship that whole thing to NIH, right?

1:29:36We built it there.

1:29:37>> Why not? Just to get the the

1:29:39>> Because there's because there's a lot of

1:29:41biology and [ __ ] you got it. It's

1:29:42easier to take the microscope to the

1:29:44biology than take the biology to the

1:29:46microscope. Okay.

1:29:47>> So, we did that and um and within less

1:29:51than a month, you know, we we had a

1:29:54sample in we uh we Jennifer's people

1:29:58would rather than taking a whole cell,

1:29:59we did cryossectioning to take thin

1:30:02sections, plate them out because you're

1:30:04worried about the third dimension,

1:30:06right? There'd be a lot of out of focus

1:30:07stuff like that. So, we wanted to make

1:30:08sure we were really two-dimensional. So

1:30:11we had these these slices that were then

1:30:13on on the sample.

1:30:14>> Could you use conf focal approach?

1:30:16>> Uh not really. Again, just

1:30:18>> you want to see like the whole field of

1:30:19view.

1:30:19>> Yeah. Yeah. Exactly. Exactly. And so um

1:30:22and so uh we did that and um uh so we

1:30:27had some slices through uh through some

1:30:30through some loss. That photo is

1:30:33somewhere in here I think. Yes, right

1:30:36there. There it is. Okay. Um that was

1:30:39one of the first but in the first so we

1:30:42turned on that the for the violet light

1:30:44and then turned on poof these molecules

1:30:47came on we said [ __ ] we got it it works

1:30:49it's going to work and then we just

1:30:52raced like hell we had only a 10 m laser

1:30:55so we spent so we worked round the clock

1:30:58just babying the microscope because the

1:31:00focus would draft [laughter] because

1:31:02it's just this cheap little thing that

1:31:03we put together right and so we would it

1:31:06was like it was like you

1:31:08November and it was cold and it was a

1:31:09concrete it was an old dark room that

1:31:12was converted and we we kept it secret.

1:31:14Her lab was big but we kind of kept what

1:31:16we were doing secret except for George.

1:31:18Uh and so uh um they would see us going

1:31:22back and forth and they didn't know who

1:31:23these two old dudes were because you

1:31:25know everybody's just a posttock or a

1:31:27grad student, right? We're these old

1:31:28guys talking to George, right? They

1:31:30called us the gruesome tsum. [laughter]

1:31:34And so we were doing this and uh um and

1:31:38uh and once we had that everything went

1:31:41like so so yeah we we went within 3

1:31:45months we had all the data that was in

1:31:46that science paper.

1:31:47>> So you realized that before you did the

1:31:49post-processing of the

1:31:50>> Oh yeah.

1:31:51>> Oh yeah. Yeah. The post-processing was

1:31:53easy. Yeah. So um yeah it was it was

1:31:57obvious we had it. Um

1:31:59>> what was the most difficult challenging

1:32:00part in building the setup?

1:32:02>> Nothing. It was it was like a gift from

1:32:05God, right? It was just trivial. Um

1:32:08everything was easy. Um it was it was

1:32:11just the ripest plum you could imagine

1:32:14[laughter]

1:32:15that was just waiting to be plucked.

1:32:16>> Why did it take so long to actually

1:32:18collect the data? What was the

1:32:21>> Well Well, because because we were

1:32:24unemployed and had to shell out of our

1:32:26own pocket for a 10 m 561 laser to

1:32:30excite the fluorescent protein. Okay. So

1:32:32we didn't have much. So so we dribbled

1:32:35those molecules out. Nowadays you you

1:32:38know you got even f you know the the

1:32:41SCOS cameras are much faster than the

1:32:43CCDs back then. The the lasers are far

1:32:46more powerful. You can do it just you

1:32:49know Icon takes a [ __ ] pabyte of data

1:32:51today. Okay. So it's a different world

1:32:54than it was back in 2005.

1:32:56>> How did you control the lasers? Because

1:32:58you need like switch on and off.

1:33:00>> Yeah. Yeah. There's just TTL pulses that

1:33:02would go to to control the laser from a

1:33:05PC. And so we had, you know, there's a

1:33:07PC somewhere in that, you know, you can

1:33:09at least see the screen of the PC,

1:33:10whatever.

1:33:11>> So they had like the kind of triggering.

1:33:14>> Yeah. Yeah. Yeah. You could you could

1:33:16trigger the lasers, the 405 and the and

1:33:18the 561 and so forth. Yeah. It was

1:33:20really simple.

1:33:21>> How did you build the the software part

1:33:24to to

1:33:24>> just just ourselves, you I mean, you

1:33:26know, you just have National Instruments

1:33:29cards and, you know, [ __ ] and you

1:33:32control it through, uh, either Lab View

1:33:35or Lab. Lab View was around. I I I just

1:33:38did it with Mat Lab, but uh, yeah. So,

1:33:41yeah. And also the localization code was

1:33:44done in Mat Lab and Yeah. So, yeah.

1:33:47>> And for the objective,

1:33:49special objective. But yeah, you need a

1:33:51good you need a good We used a high NA

1:33:53turf objective and that was another one

1:33:55of the ouches. Okay. In terms of that

1:33:57was probably at least 10 grand,

1:33:59something like that for that objective.

1:34:00Yeah.

1:34:01>> What was So that those images proved

1:34:04that the method could work. What was

1:34:05kind of the first use of the method that

1:34:08generated like biologically interesting

1:34:10results?

1:34:11>> That depends on who you talk to,

1:34:13[laughter]

1:34:13right? I I would argue that that most

1:34:16single molecule super resolution hasn't

1:34:19revealed much anything. In fact, I would

1:34:21argue that most super resolution in

1:34:23general has not revealed much of

1:34:24anything.

1:34:25>> Okay. Um we can go deep down that rabbit

1:34:28hole if you want, but my feel my feeling

1:34:31is is that um

1:34:34even back in 2014, I felt like it was

1:34:36incredibly premature to give a Nobel

1:34:38Prize for that. I think the only reason

1:34:40they did is because well they broke this

1:34:42fundamental limit but I think anybody

1:34:45who who knew anything um from an optics

1:34:50point of view would know that it's not a

1:34:52fundamental limit in the same way that

1:34:54the uncertainty principle is fundamental

1:34:57right it was a practical limit that

1:35:00could be you you know circumvented by by

1:35:04a trick right and the trick is you can

1:35:07localize to better precision than the

1:35:08wavelength Right. And that's the trick,

1:35:10right? Um, so anyway, where was where

1:35:13were we going with this?

1:35:14>> What was the biological first thing?

1:35:17Yeah. So, so again, so overall, I the

1:35:21first in my opinion, the first real hit

1:35:22and the one that led to ICON was when we

1:35:26started not to look at dead and fixed

1:35:29samples. The reason you look at dead and

1:35:31fixed samples is you're only looking at

1:35:32a few molecules at a time. And it takes

1:35:34a [ __ ] long time to bleed out every

1:35:35molecule to get that Nyquis criterion of

1:35:38having every molecule on the thing

1:35:40decorated to put together that whole

1:35:41image and you're throwing a lot of light

1:35:44at the sample and if it were alive it

1:35:45would be cooked with all of that light

1:35:47on it for that time.

1:35:48>> So um so yeah I know we're never going

1:35:52to make it through everything [laughter]

1:35:53at the rate I'm talking but um

1:35:55>> it's a great story you know was like

1:35:57where else can you find it?

1:35:58>> Exactly. This is like what it's not

1:36:00written in any paper.

1:36:01>> Yeah. Well, most of this stuff is you

1:36:04can find somewhere if you dig hard

1:36:05enough. But um but uh the uh the first

1:36:10real hit was and this this really kind

1:36:13of set the stage for the whole rest of

1:36:15my career up to the present which is

1:36:17that um at at um when I was at Janelia

1:36:23um the president of Janelia is funded by

1:36:26the Howard Hughes Medical Institute and

1:36:28at that time the president of the Howard

1:36:30Hughes Medical Institute was uh Bob Teen

1:36:34who is famous as um as a biochemist who

1:36:39was one of the key guys to unravel the

1:36:42methods of transcription, how different

1:36:45proteins come together to um recruit the

1:36:50uh polymerase which is necessary to

1:36:51unzip DNA and then produce RNA. Okay. Um

1:36:56and uh uh Tee had of was a biochemist

1:37:02but he understood the potential of

1:37:04trying to see at a single molecule level

1:37:07how this transcription is actually

1:37:08occurring and so um so we started to use

1:37:12palmlike techniques in order to do that.

1:37:15Um and uh and so the in their models in

1:37:20their biochemical models they had come

1:37:23up with this idea that many different

1:37:26transcription factors their proteins

1:37:28that come together at the start of the

1:37:31gene first before it recruits the

1:37:33polymerase. And so there's a sequence of

1:37:35events that has to occur with different

1:37:37proteins. And they believed it formed

1:37:40this large larger

1:37:43call it micromolecular complex but

1:37:45multiple proteins dissimilar proteins

1:37:48that had to start at the beginning of

1:37:49the gene before the polymerase would

1:37:51come in. And it was believed that this

1:37:53would take minutes to hours to happen.

1:37:56when we started to look at the indiv by

1:37:58by palm like stuff in in live cells um

1:38:02at these transcription factor molecules

1:38:05none of them were binding to the DNA for

1:38:07more than a second or two and so it was

1:38:09like holy [ __ ] our whole model of how

1:38:12this how transcription works is

1:38:14completely wrong and in fact this is the

1:38:18take-home story of this whole talk okay

1:38:20of this whole morning all right is that

1:38:24almost everything you learn in biology

1:38:27textbooks is a hallucination

1:38:29because they it is it is it's because or

1:38:33at least cell biology because what

1:38:36they're doing is they're taking three

1:38:38reductionist tools biochemistry,

1:38:40molecular biology and structural biology

1:38:43and then hypothesizing

1:38:46what how those little pieces tiny tiny

1:38:49little bits come together both

1:38:51structurally, stoometrically and

1:38:54dynamically to create the cell. They

1:38:57have no direct knowledge of the

1:38:58stoeometry or the arrangements, spatial

1:39:02arrangements or the dynamics. All of

1:39:04that is hallucination. Okay, they little

1:39:07bits. I'm I'm exaggerating, but largely

1:39:10speaking, you you guys have probably

1:39:12seen on the web, you know, there's those

1:39:14beautiful things of of you know, oh, all

1:39:17these molecules coming together. Here's

1:39:19a here's a cargo on on uh a kines

1:39:24walking along a microtubule like this.

1:39:27You know, if you've seen these things

1:39:28and it and it's all like in this vast

1:39:30empty space. I don't know any cell

1:39:33that's a bunch of vast [laughter] empty

1:39:34space. I'm sorry. It's crowded as [ __ ]

1:39:37There's there's there's there's there's

1:39:39100 trillion water molecules in every

1:39:41cell. There's there's 10 billion protein

1:39:44molecules. There's 10 billion

1:39:45carbohydrates. There's 10 billion uh uh

1:39:48um lipids. There's metabolites. There's

1:39:52it's by far the most complex matter in

1:39:55the known universe. We understand the

1:39:57interiors of neutron stars far better

1:39:59than we understand the interior of

1:40:01cells. It's it's crazy complex. Um and

1:40:06yet

1:40:08we've got a whole industry of farm.

1:40:10That's back to Roger's point. There's a

1:40:12reason why only 9% of the drugs that

1:40:15enter phase one come out phase three

1:40:17because we don't know what the [ __ ]

1:40:18we're doing. We don't know the real

1:40:20mechanisms that are going on. And and

1:40:22and this is the real hit from single

1:40:25molecule stuff was the realization that

1:40:27that's the case. That when you start to

1:40:30[ __ ] look at the dynamics, not just

1:40:31the structure, you realize that you had

1:40:34it all wrong. And you realize that so

1:40:36many of the things that they thought

1:40:37they knew, they you can't be sure that

1:40:39they know. we have to reinvestigate all

1:40:41of it, right? And so that's where I

1:40:43pivoted to live imaging, right?

1:40:45>> How did you what was the like

1:40:47improvement in the method that actually

1:40:48allowed you to look at transcription

1:40:50live? Like

1:40:51>> well, you were saying you had to have

1:40:52>> there was no improvement. There was it

1:40:54was basically applying palm type

1:40:56techniques to photoactivate a subset of

1:40:58these transcription factors and look at

1:41:00them in a live cell instead of trying

1:41:02instead of having to get to that Nyquis

1:41:04criterion of trying to get every

1:41:06molecule right in order to look at a

1:41:09structure. I didn't care about that. I

1:41:11just want to understand their kinetics.

1:41:13Okay? And so if I want to understand

1:41:15their kinetics, I can do a subsample,

1:41:17right? And if if that if that principle

1:41:20was right that they would form these

1:41:22these stable complexes, I should be able

1:41:25to see that. But they weren't they

1:41:27weren't they weren't stabilized

1:41:29anywhere. They were they were staying

1:41:31for a second or two. And so the model

1:41:33they've built up is that basically you

1:41:35it's the

1:41:37mitochondria is not the powerhouse of

1:41:39the cell. Brownian motion is the

1:41:41powerhouse of the cell. That's how

1:41:43everything happens is statistically

1:41:46ever, you know, molecules diffuse at,

1:41:48you know, on the order of 20 micron

1:41:50squares a second. That means you'll go

1:41:52across a whole cell in in like 3

1:41:54seconds. Okay? But in that time, you're

1:41:56not going in a straight line. You're

1:41:58bouncing off of literally one trillion

1:42:01other molecules in those two seconds to

1:42:03get to that other side. Your

1:42:05instantaneous velocity based on Maxwell

1:42:07Boltzman is like a 100 meters a second.