Michael Filler and Matthew Realff discuss Fundamental Manufacturing Process innovations. We explore what they are, dig into historical examples, and consider how we might enable more of them to happen. Michael and Matthew are both professors at Georgia Tech and Michael also hosts an excellent podcast about nanotechnology called Nanovation.

Our conversation centers around their paper Fundamental Manufacturing Process Innovation Changes the World. If you’re in front of a screen while you’re listening to this, you might want to pull up the paper to look at the pictures.

Key Takeaways

1. Sometimes you need to go down to go back up
2. The interplay between processes and paradigms is fascinating
3. We need to spend more time hanging out in the valley of death

Links

Fundamental Manufacturing Process Innovation Changes the World(Medium)(SSRN)

Michael on Twitter

Matthew Realff's Website

Michael Filler's Website

Nanovation Podcast

Topics - The need for the innovator to be near the process - Continuous to discrete shifts - Defining paradigms outlines what progress looks like - Easy to pay attention to artifacts, hard to pay attention - Hard to recreate processes - The 1000x rule of process innovations - Quality vs price improvements - Process innovation as a discipline - Need to take a performance hit to switch paradigms - How to enable more fundamental manufacturing process innovations

Transcript

[00:00:00]

this conversation, I talked to Michael filler and Matthew Ralph about fundamental manufacturing process innovations. We explore what they are, dig into historical examples and consider how we might enable more of them to happen. Michael and Matthew are both professors at Georgia tech and Michael also hosts an excellent podcast about nanotechnology called innovation.

Our conversation centered around their paper called fundamental [00:01:00] manufacturing process. Innovation changes the world, which I've looked to in the show notes and highly recommend the fact that they posted it on medium. In addition to more traditional methods, give you a hint that they think a bit outside the normal academic box.

However, I actually recommend the PDF version on SSRN, which is not behind a paywall only because it has great pictures for each process that I found super helpful. If you're in front of a screen, while you're listening to this, I suspect that having them handy, it might enhance the conversation. And here we go.

the, the place that I'd love to start is, to sort of give everybody a, get them used to both of your voices and sort of assign a personality, a personality to each of you. so if each of you would say a bit about yourselves, and the. The, the sort of key bit that I've loved you to say is to, to focus on something that you believe that many people in your discipline would sort [00:02:00] of cock an eyebrow at because clearly by publishing this piece on medi you sort of identify yourself as not run of the mill professors.

 

Oh boy. Okay. So we're going to start juicy, real juicy. So I guess I'll go since I'm speaking, this is Mike filler speaking. Great to be here. so I've been a professor of chemical engineering at Georgia tech for a little over 10 years now. my research group works in nanoscale materials and device synthesis and scale up.

So for say electronics applications, Yeah. I mean, this article, which we'll talk about emerged from, you know, can I say a frustration that I had around electronics really is where it started for me, at least, that. We have all this focus on new materials or new device physics or new circuit. And I know your listeners are probably thinking about morphic computing or quantum computing, and these are all very cool things, but it seemed to me [00:03:00] that we were entirely missing the process piece.

The, how do we build computers? and, and, and circuitry. And, and so that's where this started for me was, starting to realize if we're not dealing with the process piece, that we're, we're missing a huge chunk of it. And I think one of the things is that people, people miss that where within working within the context of something developed 50 or 60 years ago, in many cases, and it's it's was really hidden to a lot of people.

And so that, that was where I came at this. Great. All right. So, yeah, so I'm, also a professor of chemical and biomolecular engineering at Georgia tech. my background is actually in process systems engineering. And, if you go back to the late 1960s, early 1970s, actually frankly, before I was a much more than in shorts, there was a, that was a real push towards.

The role of process systems engineering in [00:04:00] chemical engineering in it really arose with the, with the advent of computing and the way that computing could be used to help in chemical engineering. And then slowly over time, the, the role of process systems engineering has become, I think, marginalized within the chemical engineering community, it's gone much over towards.

What I call science and engineering science in a way from the process systems piece of it. And so, you know, as Mike would, would berate me with the, with his travails over, over what he was trying to do with nano integration and nanotechnology, I realized that what he was doing was describing a lot of the same frustrations I felt with the way that process systems engineering was being marginalized and pushed to the edges of chemical engineering with the.

Focus more around fundamental discoveries rather than actually how we translate those fundamental discoveries, into, functioning, processes that then lead to outcomes that affect society. So for me, it, it, it [00:05:00] was a, it was a combination of, talking to Mike and then my own frustrations around how my own field was somewhat marginalized within the context of chemical engineering.

Got it. And, sort of to, to anchor everybody and, and start us off. could you just explain what a fundamental manufacturing process innovation it's. So the way we think of fundamental process innovation or manufacturing process innovation is actually rethinking how the steps in a process are organized and connected together.

And so that has become the paradigm which we have. we have set for fundamental manufacturing process innovation, and these innovations come in in different categories that enable us to put these processes together. And one of the examples of which for example, is. I'm factoring taking something that has been done together at one process step and separating it into two different steps that occur maybe at different [00:06:00] times or in different places.

And by so doing, we actually enable us to make, a tremendous change in the way that that process operates. So it's really around. The strategy for organizing and executing the manufacturing steps and using a set of schema is to sort of understand how over history we have been able to do that. Do you want to add to that mic?

Yeah. I want to take a step back outside of manufacturing. So one of the examples we give at the outset of the piece is not in manufacturing, but in shopping something that every single person listening to this can wrap their mind around, I think. and I still love the example cause it just kind of. I miss it every single day.

and this is all pre COVID thinking of course, but the idea that say a hundred years ago, and a lot or Western societies, you would go to let's call it the general store. and you'd walk in, go up to the counter. And, if I have a list maybe, and you'd handle lists to the purveyor, and they would go [00:07:00] in the back rows of shelves and they'd pull off what was on your list and they'd bring it out to you, you pay for it and you go on your Merry way.

And then, you know, several decades ago, this started to change, probably half century my ex ex ex. Exactly sure. The timing, but, to, to a model, where instead of a single shop keeper, having to interface with many individual, shoppers, it was now many shoppers who did the traversing of those aisles themselves, right?

This is at least in Western society is what we are familiar with today as the grocery store or the target or the Walmart. And what you do is you. Trade one thing for another in doing that right. Instead of, the person, the, the purveyor, getting things for you, which from a customer's perspective is very nice.

Right? you, you, you no longer have that, right. You're being told. Okay. He used to, yeah, he or she used to get it for you now. You're going to go and traverse the ALS yourself. But you do get something in return as the [00:08:00] shopper. And that is a lower costs because now one store at the same time can be, open to many, many people stopping shopping simultaneously.

So, selection goes up, costs go down and there's a benefit for the customer, and the shopkeeper. So this is an example of a process innovation it's the it's still shopping, but it, it takes the old process paradigm and inserts a new one. Excellent. And so you, in your paper, you illustrate eight major historical, fundamental process innovations.

And I would love to sort of frame the conversation by walking through them so that, a just because they're great history and B, so that everybody can sort of be anchored on the very concrete, examples while at the same time, I'll, I'll sort of poke at, The, the more sort of abstract questions and ideas around this.

so the, the first, [00:09:00] the first one you talked about is the shift from the new Komen to the watt steam production process. So like, what was that? And, and why was that important?

it was important because, what it did was it changed fundamentally how we could make power. So the newcomer engine had, the condensation of steam in the same vessel, as, as the, as what was being the vacuum was being pulled to enable the, Pulling of water up from the coal mines in Britain, turns out it's actually 10 mines rather than coal mines, where this was first developed.

And what, what did was to factor that's one of a fundamental process schema factor, the two pieces so that the vacuum pulling and the condensation happened in different vessels. And as a result of that, he was able to increase the efficiency of the steam engine by, an order of magnitude. and, and through other innovations that then followed from that.

The steam engine became [00:10:00] significantly more efficient. Now, what did that do? Well, the first thing it did was is it meant that you could pump water out of deeper mines, so you could actually now get coal out of deeper mines and so you can increase coal production significantly. The other thing it did, of course, it meant that for the same amount of power, the engine could actually get quite a bit smaller.

In fact, it could get small enough that he could actually move itself on rails. And so what that also then enabled was. Stevenson and essentially the invention of railways without the steam engine. You wouldn't have railways with railways. Now, suddenly you can bring the coal, which you've now enabled yourself to dig out of Deepa mines.

You can now bring that to manufacturing sentence. So there's a whole follow on set of innovations. And in fact, a complete reorganization it's called the industrial revolution. That is, that is based on these kinds of process innovations. And this was one of the most central ones, right? To that actual outcome was the idea of factoring these students.

[00:11:00] Two steps leading to much greater efficiency in the way that a steam engine could be used. And, and that, there's actually two pieces that I think are fascinating about that. And one is, this phenomenon that you see over and over again, where what I would sort of call a continuous efficiency increases, right?

Where it's. It was, it was like a fairly steady, increase of efficiency. But then, because as you point out, it eventually got efficient enough that it could power, a rail car that all of a sudden made this like discrete difference in what the process was actually capable. and I feel like you see this in school, many of the examples that you give, and I like, I just love that.

And then the other piece. That I believe is the case. Is that, what was, what was, new Cummins apprentice, right? That I'm not sure about that actually. I mean, I think he was familiar with new [00:12:00] comes work. but I don't know if he was actually his apprentice or not in that particular context. W w and the reason that I ask is that, like, do you think that what would have been able to.

Create this, this process innovation, if he hadn't been like, sort of actively working with the new Komen engine in the first place. No, I think the answer is, is that without that he, he, you know, you have that you had to have a starting point. And I think he understood, once, once he sold the starting point that, yeah, that was a, there was a way in which he could make this more efficient.

the other thing about the, the, the efficiency and the scanning of efficiency is what we see in a lot of these fundamental process innovations is that there is a step change, but not only that, it shows how then off to that fundamental process innovation has happened. That, that can be this continuous increase, right?

So there is, it unlocks an enormous potential to suddenly change the game in terms of the efficiency. So, [00:13:00] so the point being that say the original engine was maybe less than, than 3%. Maybe one, 2% efficient. And what, what did with the sort of next version was increased that by an order of magnitude, and then suddenly with that innovation now by better manufacturing, higher pressure vessels, et cetera, you could actually then go into an even higher level of, of efficiency.

Not only that, but it drove the development of the sort of discipline of thermodynamics. Now you have to analyze the engines on their efficiencies and understand what could lead to greater efficiency in the future. And so, and you know, entirely scientific discipline was built on top of the, of innovations that were occurring in heat engines.

Yeah. Well, I think there's an important point here in the efficiency discussion, right? And Matthew and I have chatted about this. A fair amount is that you kind of have the efficiency piece and as you're pointing out, Ben, it's really critical. Look it up for some threshold with a lot of these, but efficiency is kind of zero to a hundred, [00:14:00] right?

And then you have the whole cost throughput piece. And as we show in the piece, you have many orders of magnitude possible gains on that side of the equation. and some of it goes hand in hand with efficiency, but I sometimes think that that is there's an overemphasis, often on efficiency.   you gotta get through the threshold and then recognize that the driving down of costs or increasing of throughput can happen, you know, a million X, you know, as, as for example, the planar process of integrated circuit shows it's more than a million X decrease in cost over time.

Yeah. And, and this, this idea is that, that you point out about almost sort of like the process innovation, defining a paradigm that then sort of sets the pack for things is, is a theme that we'll like, let's, let's almost like poke it that as we, as we go through through everything else. and, before we move on, I guess the last piece, [00:15:00] sort of going back to.

like Watts familiarity with the process in the first place. And sort of tying it back to to today is, I guess what, what's your take on sort of like the, the, the familiarity that the people who are working on cross as possible process innovations have with the processes now, Let's see, I probably phrased that a little bit weird, but, I guess my concern is that there's, there's more of a separation between the people that we expect to do the innovating and the people who are working on the processes.

So, so yeah, this is a really critical point. I mean, what we have done in the modern innovation enterprise right, is we've split, so-called fundamental research with applied research. and, these examples, many, the ones that we give are really squarely between the two and they need both [00:16:00] to function.

And so this is, for this kind of innovation or real. I think a real issue with the current way, things are set up, because it requires some knowledge of the science that's kind of emerging. It requires some knowledge of some engineering, and it's a matter of integrating these things. And it's not, so much, I think what the prevailing view of the world is, which is fundamental innovation gets developed and leads to some specific technology.

It happens between the two. and so that's, that is, that is, That is a theme I think, and these innovations and it's something that I think today is harder to do. we could talk for a long time about why it's harder to do, but it's harder to do today. Cool. Well, we'll, we'll we'll. Circle back on that, as, as we get sort of closer to the present.

so can I say one more thing? This is such a good example, but everyone knows the, the watt engine and we are very careful to call it, the watt, what do we call it? We call it the [00:17:00] walk process, right? We call it the what process, what process for energy generation or something like that. But yeah, we focus on the process and I think this is one of the reasons why these kinds of manufacturing innovations are missed all the time is that you focus on the engine, the physical thing that carries out the process and you're missing that.

Oh, actually, what, what did was he factored these two steps? It's still a machine like new Coleman's machine, but in the end, what made it so powerful was the underlying process that. It carried out. And I think that that is one of the reasons why these manufacturing innovations are missed in manufacturing versus in other areas where process is talked about much more frequently.

So I wanted to make sure, well, actually, as long as we're on that topic, I want to, sort of the talk like the. call out the sort of obsession with novelty in academia, where like, [00:18:00] if like, it's, it's really important to call out the, the process innovation. Because if you look at it just as like steam power, then you could sit, like you could sit a lot, like what's novel, like near your dinner, your power from steam, new colon generated power from steam.

and so, so we, like, we need to. Really sort of pay attention to what's going on on the inside and like how that really different, even though on the outside, it does not look that different. For sure. And, and I think the point that we arrived at there is, is, is when we went back into deep history and asked ourselves, well, what do we call the ages of the past?

And we call them things like the INH. We don't call it the smelting age. Right. Right, right. We could, we could call it by the process, but we don't, we call it by the thing that was made. you know, we don't, we don't talk, we talk about Flint's and we talk about Flint arrows. We don't talk about the ways in [00:19:00] which those flints were shaped into arrowheads, the flaking and the, and the, and the.

But essentially those kinds of processes, which we don't even know in many cases how to reproduce and they lose that knowledge for, for many, many years. In fact centuries, the one example we use in the paper is that a Roman concrete, you know, we were able to, to look at Roman buildings, but we were not able to reproduce them because we had lost the, the recipe.

We lost the recipe for making, concrete, with the, with the sort of dissipation of the Roman empire. And so in fact, we couldn't reproduce these buildings, so we could look at them, but we couldn't reproduce them because we had lost the process. Well, I think that that's so key to point out because it's almost what, like similar to the, the streetlight effect where, it's, it's so much easier to look at and point out and talk about the, the artifact.

but it's, it's, not as legible what work went into making and even, even now, like, even [00:20:00] now, when you like, literally when everybody's writing everything down, it's still, there are so many little things that go into these processes, that are sort of illegible. and I think that it's. Easy to forget about that and think like, Oh, well, you know, someone wrote it up.

Therefore we know everything that can be known about it. Yeah. History is kind of similar, right? The history. Yeah. We, we, we look back on history and we don't see the generator of the history. Yeah. So it's, it's often very hard to get our true handle on what it was that led to certain phenomenon. We, we, we look back and we start to come up with theories.

and I mean maybe sometimes they're right. Sometimes they're wrong. We don't have, we have some ways of knowing and other areas. We have no way of knowing because it's, what happened is lost to time. Yeah. Sorry. This is kind of very similar in terms of the fleeting nature of processes. Yeah. And, and, and the fact that it's not easy, I think it should be born out [00:21:00] by anybody who's ever tried to read the materials and methods, sections of academic papers, because you will discover that very rarely do the researchers actually document the materials and methods in sufficient detail to actually reproduce them.

There's a, there's a, there's something that they do in the lab that they just forget to write down. That's actually absolutely critical to make the, the, the, the material process work. you'll just discover that they, Oh yeah, we soaked it in methanol for 60 minutes. Oh, I'm sorry. We left that out. you know, there's, there's there are, there are easy to leave out these steps that turn out to be crucial, but they're not the final artifact that's being exhibited in the paper.

Yeah. Yeah, there's this, there's this, sorry, there's this, this kind of discussion in, today in science about irreproducibility and we have this reproduction crisis and okay. Maybe we can be doing a better job, but I think a lot of it it's just, as Matthew's describing it's stuff that is not obvious you, as the experimenter are doing the experiment.

You, even, if you wrote [00:22:00] down absolutely everything you thought you did. There are things you didn't even realize you were doing that were central to the process and it gets lost. And that, that to me is likely the main source of a lot of these, these issues. Yeah. I wonder what would happen if we actually had a system where you just videoed, literally everything that someone did in a process and then, like captured every key stroke on their computer and it would be it.

Yeah, but , I wonder, I wonder whether it would just be completely, unintelligible or  whether there'd be something useful that came out of it. Just for the sake of time. I love, yeah, let's move on the second of eight. so, the, the, the second process you talk about is, the, the, the foreigner process for continuous papermaking, which I did not know anything about before I read this.

so yeah, like what, what was that, why was it important? So, so here is it's a lot like, what,   Gutenberg good with the press. but, [00:23:00] paper prior to this innovation was Preston single sheets and dried as single sheets. basically a fully integrated process on one sheet of paper. And, what, continuous papermaking did was it took each of those steps and separated them into individual components.

So that's a factoring schema, as we describe in the paper, where you first throw down the slurry of pulp. Right. And then, there's a section where you let the water drain. you consolidate the Pope down into something that's like a sheet, and then you push that sheet through rollers. and then you dry it, but each of those steps are different, right?

The pulp deposition, the rolling and the drying are separated in space and time now. Whereas before they were more or less in the same space. And so that, that factoring allows you to scale up by orders and orders of magnitude, that production rate of paper. And so we talk a lot about Gutenberg's press, being central to mass literacy and it clearly [00:24:00] was.

But, and, and we're not the first people to point this out, Tim Harford, who I like a lot who writes for the financial times and his own books, has talked about this where, you need to have the continuous paper. Manufacturing piece so that you could get those books to so many more people. And it was really both of those together that, that led to that. The other point I was going to make about that is, is it also revealed that we, that we were going to that as soon as we were able to, you know, produce, paper at large rates, we needed some sort of raw material that could also be produced. At large rates. And so this idea that you are going to continue to use rags as the, as the input, suddenly became difficult.

And so people had to scout around for other forms of fiber that you could use. And that's really what led to the whole, you know, creation of, of the pulping industry that, that takes what. Well on the face of it, a tree doesn't exactly. Look like paper, takes a tree and turns it into something that you can make a make paper out of.

[00:25:00] So again, it's this upstream and downstream it's the, the downstream effect is, is. The societal mass literacy, the upstream effect is, is the, is the creation of a, of an entire industry around, you know, turning trees into, into pulp. and so some people might disagree with doing that, but, but the bottom line is, is that's what enabled, the, those two pieces to be driven was the creation of the, of the, of, of papermaking in the, in the middle of that.

Yeah. And something that. So, did you have a sense of how people were thinking about papermaking? Oh, for, before for generic came up with process that is like, did, did they realize that it should be possible to make paper more efficiently? Or was it just like, just that's the way it wants? because I feel like so many of these process innovations.

[00:26:00] There are people just sort of accept whatever level of whatever process we have. And we're like, Oh, like that's the way it is. Yeah. Maybe we can make it a little better until something new comes along. One of the things we were careful to do in the piece. And I'll be honest because we're not historians is to, to try to stay away a little bit from like the, the, the driving forces.

Right. And kind of what people were thinking. I'm really focused on the mechanisms. And that's one of the things, you know, I've really enjoyed learning from people who are in the, the progress studies community, that emerging community. in general, I find that they really know a lot about history and that's great.

and we really wanted to make sure we could pay attention to mechanism at the, at the actual innovation level. and so I guess I'm saying that as a long winded answer to say, I don't know how they thought about it. but, you know, but I think that there's kind of been a shift over time. you know, Matthew was sending me, show me something from scientific American recently.

[00:27:00] They just, what was their anniversary? Matthew? A 175th. I can't remember what that is in Latin, but, but it's, it's a very long and complicated word. Yeah. But  DECA. Yes, exactly. Quickie and no versary. Yes. It's something like that. I buy, if I pumped up, I could go get my issue and they have it in there, but, but it is, it's quite a complicated word.

That's all I remember. And they have a article in there talking about the shift in how people speak, spoke about science and engineering. And, h hundred years ago, there was this kind of more engineering processing, which that was far more common. And then around at the time of world war two, it kind of shifted, be more about science and the emphasis on science.

At least as far as that magazine goes, but I think the magazine is probably fairly representative of the endeavor as a whole. And so, yeah, that's, that's kind of fascinating. You're saying, did they appreciate, whether the process could be [00:28:00] better? And my gut feeling is they maybe in, in the 18 hundreds, they appreciated that it could be better, more.

Did they have an appreciation for how much better that's that's probably dubious. Right? I think most of these, if you went back and asked the original innovator. Did you know, you were setting us on a pathway or a trajectory that led to, you know, the world, as we know it today, I think they'd probably be like, wow, no, I did not expect that.

I just was trying to make an extra buck. Yeah. But I think it's like, it's actually almost like a powerful, admonition people to sort of like, keep in mind the different schemas that you lay out and just to like walk around the world. Saying like, Oh, like, could this, could this apply here? and it almost like gives you a bit of humility that it might be possible that like these could always happen.

that's for us, that's kind of [00:29:00] emerging from doing this and we're, we're continuing to work on, on, on next pieces basically is a kind of a thousand X heuristic. Whereas you have a two D technology today and you ask yourself, can I do it a thousand X cheaper or a thousand X faster? with the way we do it today?

if the answer is yes. Okay, great. And you're really competent that if the answer is no, it may be time for a process innovation. Maybe to us a thousand X is, is sufficiently beyond someone,   you know, giving you the pop out answer. Of course, we've made progress in the last 10 years and I expect more progress.

Well, that's kind of a cop out answer. A thousand X is quite a bit faster or quite a bit a higher throughput. So that's, I think that's a good metric for anyone working on any technology. and I think COVID COVID is a great example of what we've been experiencing in the last, however many months. It feels like two years, and you know, we needed rapid vaccine [00:30:00] manufacturing.

We needed rapid testing, basically a thousand X faster. And we didn't really have that capability in hand and people have done tremendous work right in the, in the intervening months to try and get us a lot closer. I know Matthew has done some work on this. but when the whole thing started, we hadn't really thought about it so much yet.

How could we speed up this a thousand X? And so for us, it's a pretty good heuristic is that, is I like that a lot. That is a very powerful heuristic. and it's also like it's, it's aggressively ambitious, which really, really does speak to me. cool. And so, let's, let's talk about the, the Bessemer process for steel manufacturing, which, His age is really cool.

everybody listening, go check out the pictures. so, so what is that and why was it important? So again, I think it was important because what it led to obviously was a, was a, a better steel and, steel that you could make. Again, as Mike has pointed out, you could [00:31:00] make,   the steel significantly faster than the existing processes.

and what it came down to was was, was a recognition that actually to remove the impurities from the steel, you, you could blow air through the steel. That that would cause a reaction that would cause the steel to heat up. Whereas if you think about blowing edge generally, if you blow on things, it makes things colder.

So this idea that you would blow air through something to make it hotter was was, was obviously a, you know, something you do in bellows and had been at. Had been thought about in terms of bellows, but actually literally blowing the air through the steel was, was not something that had been done and, and combined with that idea was also this idea that by removing all the impurities and making essentially something that was, that was pure.

And then adding back dosing back impurities after you've purified. So that you had control over the composition instead of attempting to stop right at the moment when you had exactly the [00:32:00] right amount of carbon, for example, in the steel, that, that was then another powerful idea that came about. So, so the Bessemer process really.

Had a profound impact, both in terms of, again, how much steel you could make in a given amount of time, because it increased the rate by this heating, and then also the control of quality by this site, this very counterintuitive idea of removing all the impurities and then adding something back in order to get to the, to the final product that you wanted.

That led then to, to much stronger steels than had been capable of being produced previously and much higher quality control too. I mean, that was a key piece of that. And so actually on that point, you, you, you, you note that the, the best word process led to, three order magnitude, three orders of magnitude increase in, in steel production.

And, I'm not, this is something that I, I always wonder about with the, these process innovations that both make it cheaper and [00:33:00] increase the quality, Do you have a sense of whether the order of magnitude increase was primarily due to sort of like moving down the supply demand curve, where there was just like people, you know, because the see was cheaper, they would consume more of it or was it primarily driven by, by new applications of the higher quality steel?

obviously it was both, but it's interesting to think about like, which of those. Ends up being, I think the high quality in this case was a, was a very critical factor in the, in the, in the equation poly, because one of the things that opened up was is it opened up the idea of making steel rights, as opposed to what was made from iron rails and steel rails were able to bear a huge, a significant amount, more weight.

And because of the fact that they could bear more weight. Now, suddenly again, you could increase the distances and volumes of which trade could happen. And so this, this was one of the reasons why, for example, you could spread [00:34:00] all the way across the United States because you could connect the resource rich West to the, population rich East.

with, you know, now a much more powerful, communications network driven by, you know, the steel rails that you were able to produce. So I think that a lot of it was, was, was, you know, bound up with this idea that suddenly now this new application came, came about, that you could do much as the steam engine sort of.

When you were able to move the steam engine with its fuel, you now actually could even start that whole process going. So, so again, it's this knock on effect, here, follow up on that and just make the connection for everyone that the efficiency threshold we talked about with watt is very similar to the strength threshold Matthew's talking about with steel.

Right. And cross that threshold to a new material, a new strength threshold, but then it was really this driving up production, driving down costs by orders of magnitude. And yeah, we, we got better [00:35:00] higher stress, but you're not going to change the strength of something by a million times. Right. Right. So again, it's, it's kind of these two columns, the efficiency or performance column, and then the manufacturing scale column.

Right. And, and going on to the next process in the, in that, in that, in our list, the calorie cracking process, again, you have that same, juxtaposition. You have the fact that by factoring the catalyst regeneration from the production of the fuel, you enabled yourself now to have a continuous process. which enabled you to increase the throughput in terms of the barrels of oil that you could, you could bring through this process, you enabled it to be increased significantly, but also this innovation was happening at a, at a time period where aviation in war was a significant factor and the quality of the fuel that you actually produced.

Out of the, out of the catalytic cracking process was higher than the quality of fuel you [00:36:00] produced just by distilling off a certain fraction of the, of the crude oil. And so what you were able to do essentially was, was have a higher performance aircraft engine that was quite significant in terms of its power to, to wait.

A ratio in terms of what it could deliver. And so that gave a, you know, allied aircraft, actually a significant boost in performance by having this fuel available to them. And again, provided a significant driving force to scale up the process, which again, went up by a factor of at least a thousand, over the course of, two or three years.

Yeah. It's these numbers like whatever, would it be? Say these numbers it's still sort of crazy because it feels like. So many things, focus on like getting, you know, like 10% more efficiency. whereas like, like truly getting to a thousand thousand Xs is like mind boggling. So, I believe this was the case for catalog cracking, and I know that it's the case for many process innovations, [00:37:00] where, at first the, the innovation actually makes the process less efficient like wall while you sort of are figuring out how to get everything working.

And then, once you do that, then it makes the whole thing skyrocket. and so I, I guess, The question is like, do you have a sense of how people sort of got past got out of these, like these local equilibria where, you know, if you went to someone you're like, Hey, I want to think less efficiently so that eventually it will become more efficient.

so like how, how these, these things even got through.

I'm not sure I have any great answers except perseverance. I mean, I think a lot of this stuff comes down to, to the inventor, really, you know, from their experience from their early work on, innovation recognizing in themselves and in their work, that there is the potential, even if right now it's not quite there.

you know, [00:38:00] Bessemer was the same thing where, you know, you first,   licensed the patent to people and they could reproduce what he did. So the separation of full separation of impurities came later, so that people could reproduce it. So that was a reproducibility problem in the beginning, not so much a strength problem.

and, yeah, I don't know. I think a lot of this just comes down to the person, saying I see it just like any of today's, you know, visionaries we talk about in the innovation space and then just keep hammering on it. Yeah, right. I mean, there's counterfactuals, right? So sorry, Matthew. I mean, it was just, we can't, we don't know the ones where the person didn't hammer on it and it never came to fruition.

So it's hard to know. Right. I'm going to string together, you know, a few thousand laptop batteries and stick them on the bottom of, of a, of a car. And that is going to create a company called Tesla. Right. so, so, so the answer is, is, is it's very hard to predict, obviously a and B the T's about a lot of it is about [00:39:00] perseverance and certainly Elon Musk will we'll talk at length about the fact that he, he.

He's thinks his quality is perseverance. And that it's, that that's, that's very important in this context or I'm going to have a rocket that goes up into the air and then eventually pirouettes and lands on a, on a platform floating in the middle of the seat. so these, these are, these are, you know, innovations where, where certainly the, the individual involved has plays a pretty signature.

If you can, too, to the perseverance necessary to get it to that stage. But, but it's also important to recognize, right. That it's not perseverance along the existing trajectory. Right. It's stepping aside trying to establish a brand new trajectory and pushing on that. And I think sometimes those, those two are missed a lot.

When you use the word perseverance people, miss that. It's it's, it's also this stepping outside of the existing trajectory. Yeah. I I'm, I'm particularly interested in whether we can like. Create Mehta innovations in sort of [00:40:00] roadmapping out what that stepping aside looks like. So instead of just, I'm saying like, okay, we're gonna go this other way.

Like really sort of saying, we'd go this other way. And like, this is what it will take to get this too. Do that, that thousand X to hopefully make it easier for, these individuals too. So just convince other people that they're not crazy, when, when they don't maybe have a couple of million dollars to go off and like blow up rockets on an Island.

Yeah. It's I think it's, it's, it's hard to figure out. I mean, look at, look at the bottleneck that emerged that Matthew was talking about and continuous paper manufacturing. I, you know, I think I'm pretty sure when they started, developing that process, they didn't expect that to be the next roadblock.

Right. but it was, and so, so again, this comes back to the perseverance thing. I think, I think you can try and outline it stuff, but there's going to be roadblocks. And you probably should. Right? Don't just, this is not just serendipitous. I think there's a certain kind of [00:41:00] force that comes with these things that people push on the innovations.

but you know, recognizing that there's going to be one new bottlenecks that emerge, but not to let those discourage you and that, you know, this, they think of them as, you know, motivating new science and engineering and, and that's how I view a lot of this stuff. And, and yeah, that's what I would say, Matthew.

Yeah. And, and actually on the note of sort of unexpected bottlenecks, I think that that's another key point is that, like so much science and engineering does come out of trying to implement things and then running into bottlenecks that you can't even expect. Right. Like, instead of trying to like, imagine everything through,   cool.

So just in it for the sake of time, let's talk about the, the planar process for integrated circuitry, which like arguably, has been the driving force of at least the second half of the 20th century. [00:42:00] Yeah, and I think it's often a missed, right. We talk about the integrated circuit and information technology, and miss the fact that there's this process underlying it, that has enabled us to interconnect.

I mean, it's in certain settings, it's hundreds of billions of transistors now. Right. And so, in the early days, everything was discreet. just like everything else, everything was modular and discrete components. Yeah, transistors were all sold as single tracks. I would tell them that way. Yeah, exactly.

No, no. Yeah. I'll, I'll take three. And, they, P people have the idea of interconnecting them. We, we were building computers. We recognized how hard it was to take these modular components with the technology of the time and integrate them. the other thing that was happening at the same time was some science.

And actually, this is one of the cool things about the planar process was that there was science going on. Where there was a recognition that embedding these electronic devices all the way inside a single crystal, Silicon wafer gave you much better performance. [00:43:00] And so it was kind of the realization that you could jam these things inside the top surface of a wafer.

There was also surface passivation, for those who are familiar with this process, that was key to making the devices good once they were embedded, but then once they were inside the wafer, the top surface remained flat. but they were embedded. Right. but the, the technology before that was what they used to call Mesa technology, where the transistors were kind of built on top, like mesas and Utah or Arizona, but putting them in, okay.

The wafer left the top surface flat and much easier to interconnect using this development of photo lithography. And then it went from there. and, and so that, that was the key innovation, was this extreme parallelization basically. of embedding, not just a single transistor, but thousands and then millions and billions of transistors.

And I want to also point out, you know, The, the, the trajectory that, that set us on as described by Moore's law, [00:44:00] this idea that we, decrease the size, increase the number at a, at a rate that's, gives us Moore's law and, and potentially that's slowing down. that's another one of the features of process innovations in many cases is that they, they eventually will run out of steam.

and, I, I think we're starting to see this with the planar process, where it's had a tremendous runway. but we're getting to the point where the underlying assumptions of it may no longer not, they're not going to go away, but that we may benefit from an alternative way of building circuitry.

Yeah. The, these processes they're, their effects tend to fall as you point out, tend to follow S-curves. Right. So that's, we're sort of, you see it when you start to like hit the top of that. S-curve that's when you need to think about like these fundamental process innovations. I think we've been at the top of the S curve for a long time, the processing, I mean the prediction of the [00:45:00] end of Moore's law.

And I say that in quotes, it has been around for decades and, always been able to get around it. and that's impressive. It's a Testament to the scientists and engineers that work in the industry. But, you know, you can only get so small. yeah, that was an interesting thing here about biases also that, the planar process biased us towards miniaturization, right.

biased us. But one of the central tenants of the planar process is perfection at every step. Once you put transistors in the solid wafer and you can't pull them out very easily, or really you can't, if they're defective, You're now in a world where every transistor up to these tens of billions, we're talking about better, be really close to.

Perfect. And, so what that drives you towards it incentivizes you to, not change too much about the process and find a trajectory that allows you to still increase performance. And that trajectory was just shrinking thing. Don't change the materials too much. Don't change the [00:46:00] processes by a large amount to shrink stuff.

And that was very synergistic, right? That's Moore's law and it's a tremendous success, but it did incentivize us down that pathway. And it's a bias that process innovation set up and that other innovations would set us up to go in a different direction. Yeah. Yeah. That's the, the counterfactuals are fascinating.

And, and, and another thing that I think is really interesting about the, the planet process. and, and it happens in other places where,   horny, who, who came up with it happened to have had experience with printing, if I remember correctly. And so you tend to see these, these situations where like someone who has experienced in like a completely different discipline.

Just so happens to be interacting with the process and say like, Oh, Hey, perhaps this thing from this other discipline can be applied in this process. and I wonder if there are that, like, do you have an incentive, like sort of better ways to get that to [00:47:00] happen?

well I do, which is to create a specific, discipline around, this. So, so I, you know, if I'm going to take a very strong position here, I would say we need, we need a discipline of process studies. where we do try to lead, you know, young minds because ours have too inflexible at this point, across these different kinds of examples and allow them to see the connections between the different processes in different technological domains.

And that may be, although that's not a, not a, a pedagogical, certainly that will be this opportunity. They will then connect these ideas in some other manufacturing domain, or even across. for example, service domains, I do see that there is this general principle around process innovation, manufacturing, so potentially, possibly founded on the schema that we've, that we've outlined that could enable people to see these [00:48:00] connections and start to use ideas from one process discipline in another.

And so factoring could be sunny appears as we've said, in, in services. And it could appear in other manufacturing domains as well. So, so I would advocate for a borough, sort of a discipline that's built around this, these ideas so that we could lead people to make this more efficient in terms of our discovery.

Wait, Mike's refraining. No. I, I, I agree. I think probably the things we're talking about or the discipline Matthew's talking about, I would liken it a lot to the role mathematics plays, right? Mathematics is its own discipline. it's separate, but all of the engineering and sciences use it. and so this is kind of similar and we were very careful, to pick out to process innovations that span the gamut.

We really, we think, I think it's hard to argue that any of the eight we picked, were not really impactful. but they, they really [00:49:00] span a whole variety of, of disciplines kind of showing that it really is everywhere, but we don't recognize it as, so as pervasive as something like mathematics. and, I, I don't want to be heard as saying, well, we're as important as mathematics.

mathematics has been along around a long time, but it's something akin to that. Right? I think the one place that I think it's different and would need to be adjusted somehow is that there's there isn't a ton. I mean, there are some, but like there isn't a whole lot of feedback loops between. Matt and the, all the other disciplines that math, enables.

so the, so like occasionally you'll see like a mathematical problem. That's been inspired by a, a sort of more applied problem. whereas I imagine in some kind of, process innovation discipline, you really do need to have these, like these feedback [00:50:00] loops. Between, the, the discipline and the, and the sort of like the effective disciplines and sort of like setting up those, those feedback loops seems, important and harder.

Yeah. Discipline is hard. Yes, absolutely. And I think with mathematics, we may have been doing it for so long that we don't see it. Right. I think, I think, you know, if you think about astronomy, for example, astronomy uses uses mathematics falling objects, is one of the inspirations for a lot of, a lot of mathematics.

And so sometimes I think we know that mathematics has become the problems in mathematics have become so embedded with each other in some sense that we don't see that we need to create that, that, that feedback loop. Right. whereas, you know, geometry, for example, is another one, where, whereas in, in process, I agree with you.

It's still something that I think is despite us having, you know, used [00:51:00] processes since we were, you know, since we were time in Memorial, right. We haven't really set up that as a formal means of, of analyzing the way we, the way we do things, right? I mean, that's, that's, if you like, it's the science of the way we do things.

and that's what we need to, we need to think about and actually put that out. I'm going to argue against myself and, and there's, there's tons of examples of math, being inspired by, by applications where like, look at information theory, right? Like the whole reason that. We have information theories because they wanted to see how much information they could cram in a single copper wire.

So, so I will actually rescind that

really. Yes, I think so. And I, and I think the other thing there is, is look how impactful, what is the impactful mathematics? It is actually, I mean, in some sense, almost by default, but it is the sorts of things where now, you know, where information theory was obstructed away from the app, from the original idea.

And [00:52:00] now has come back to influence a whole range of. Of of applications beyond that. And that's, that's the, the value. And I think that's the same thing with process innovation, right? If we could abstract away find the, find the, the, the core of that as a discipline that could then come back and influence a whole range of, of the way that we do things.

Yeah. And, and so, so I do want to be respectful of both of your times. so, what I will do is encourage people, listening to go look, read, like, read the paper, to discover, the, the last three, fundamental process innovations. And the way I'd love to close is, sort of beyond reading this paper, like, how do you think that we could.

Get beyond, reading the paper and Vicky about a new discipline. Like what, what are ways to get more of more fundamental process innovations? Well, I think we, we, at least in some, [00:53:00] some amount of our innovation sequence, need to recognize that there are things that happen. Within the Valley of death.

So, you know, we talk a lot about the Valley of death as something to cross. first of all, Valley death is very manmade because we've split fundamental science and applied science and processes. An example where the splits are really bad thing. And instead of crossing it, we should look at at it as we want to go into it and hang out in it.

Yeah, right. I think this is one of the issues with it. This course is it's all about something bad versus no, it's actually where we need to be. for, for certain innovations. you know, I think you think about the Nobel prize from this last week for CRISPR like that, that is squarely in my mind, that is a discovery.

It's a fundamental discovery and it'll be translated that that's kind of the conventional view of things, but there we are not doing ourselves any favors by. By having the scale too [00:54:00] much on the fundamental side and that we should at least rebalance a little bit and force ourselves down into that Valley.

Just hang out. Yeah. Love it. Matthew, what do you think. Yes. I think the, the stepping away from some of the things that we take for granted, like electronics manufacturing, and, and considering Mike's question around what would make this a thousand X,

 

better in some dimension. Is is, is really the way that we can, that we can make progress.

And again, your point was very well taken, which is sometimes when we get better at something, we're going to get worse at something else. Right. And, and it could be that we're going to have to accept that we will not have circuitry that behaves as, as, as well, or as fast as it did previously. But now we may have gained in some other dimension.

So again, it's about taking the blinkers off and not saying, okay, we have to have these particular metrics [00:55:00] always be improving, but think about how through processes. We may take some other metric and now make that significant it'd be better than it was previously. And then. Hang out and see what happens as Mike said, because by doing so, we may in fact then lead ourselves to improve other areas as well.

And that, that could then lead to the kinds of scalings we saw with making steel, making paper or making energy. And so that's what we really need to think about.

Here are my key takeaways. Sometimes you need to go down, go back up. The interplay between processes and paradigms is absolutely fascinating. And we don't talk about it enough. And finally, we need to spend more time hanging out in the Valley of death.

[00:56:00]