Patrick Collison on Innovation and Scientific Progress [audio]
econtalk.org
econtalk.org
Some things that define society, or at least peoples views, have also just been getting harder. You can't really do something as obviously meaningful in space as the moon landing. You can do many other things if you change your frame of reference, but we are not going interstellar anytime soon.
If there is any observation I would make it is that the way innovation is made seems to have changed. It used to be you had some sort of goal and then tried to get the best technology to do that. Today, it seems like companies, and societies, are sitting around waiting for technologies to exploit. One day self-driving cars, 3d printed houses and AI enabled health scanners are going to save us all. We aren't really sure how, why or if there is even much worth saving at that point, but it will surely happen?
Agreed -- I'm disappointed to see this from pc. This piece reeks of the "Andy Grove Fallacy", as described by to Derek Lowe ([1] and [2]).
It's one thing to be a mastercard middleman -- it's completely different to create new fundamental knowledge about the world.
[1] "Andy Grove: Rich, Famous, Smart and Wrong" http://blogs.sciencemag.org/pipeline/archives/2007/11/06/and...
[2] "Silicon Valley Sunglasses" http://blogs.sciencemag.org/pipeline/archives/2015/04/02/sil...
One thing they are at least implying, but I think they could have explored more is implementations. While you can argue whether there are less scientific breakthroughs or not, we certainly have less implementations. Were I assume I differ from their view is that I think this have to do with diminish returns for the implementations themselves.
To some extent it becomes meaningless to "cure cancer" when we have preventable diseases and situations killing people every day. So there can't realistically be a goal to do it. (Which isn't intended as an argument about the value of each). As I insinuated earlier I think a breakthrough is really in the eye of the beholder. Society these days doesn't really provide a platform for creating and implementing breakthroughs. The ambition get lost in the noise.
I think people assume when watching utopian science fiction that science resulted in society. Maybe in reality it is the other way around. Once you decide that everyone deserves e.g. housing the result becomes worth it and the science follows. That is certainly what happened in the ~60s in many parts of the world. If no one thinks that this is a worthy breakthrough why would anything happen in the first place today?
[What I really wanted to say was that I found the middle part of article surprisingly interesting].
If you'd ask yuval-noah harari, an historian, he'll say that with today's and tomorrow's innovation in AI and robotics , we're at unprecedented era in history.
On the other hand, when someone starts having public opinions on pretty much anything (from buddhism to AI), I tend to be very suspicious about their expertise.
Many, many, many. But it’s hard to pinpoint them.
Lower-friction communication between researchers and better access to published science is an incredible catalyst for every kind of advance (from trivial incremental improvements through to new paradigms) in every field.
I've reached a similar conclusion about fundamental physics. At the very least we are stuck with what we have for a very long time.
This may seem to be anti-scientific or anti-progress but who proved that scientific progress is unbounded?
This is also what what generally believed about physics in the 1890s, right before everything exploded with Relativity and Quantum Physics.
"it was generally accepted that all the important laws of physics had been discovered and that, henceforth, research would be concerned with clearing up minor problems and particularly with improvements of method and measurement"
https://en.wikipedia.org/wiki/History_of_physics#20th_centur...
I think being "mastercard middleman" is disingenuous. Firstly, because doing that well breaks down into getting good at many domains (software, finance, management, regulatory, sales, recruiting, fundraising, etc. etc. etc.), and secondly because Stripe is far more than that.
At the very least, his intuition on these issues is very likely to be much, much better than random chance.
Thanks for the book. It's great!
A deeper cause is mathematics taught as a trade to technicians, when Euclid and geometry stopped being central. Instead of a way to think, to prove, to investigate, we have tools for jobs.
Personally I would find it more surprising if there were not diminishing returns on investments on scientific inquiry, than (as presented in the article) if there are. Kuhn's paradigms offer the escape maybe, ROI diminishing within one paradigm, and then increasing rapidly within some new paradigm.
> Most of the smartest people I know lean on these verbal crutches
Yes, “most” smart people never spend much deliberate effort on improving verbal communication. (Just like “most” smart people never put deliberate effort into their cooking skills, or their running form, or their mental calculation speed, or their tree climbing, or their drawing, ....)
There are a lot of skills people might choose to work on, and only so much time, and if you pick any particular one (unless it is something unavoidable for most adults like eating or reading), you will find most people to be unpracticed at it.
But that doesn’t mean intellectuals shouldn’t work on verbal communication skills.
I never suggested otherwise.
I listened to this podcast and rarely read transcripts. When listening to podcasts, at least for me, the "umms" and "awws" allow me time to think about what is being said.
If you prefer reading. Read the article that the interview is based upon: https://www.theatlantic.com/science/archive/2018/11/diminish...
And Stubborn Attachments by Tyler Cowen: https://www.amazon.com/Stubborn-Attachments-Prosperous-Respo...
I did read the article when it came out. The interesting part here is the way the presenter might challenge some of the contents of the article.
Polish of delivery is just a nice to have feature, substance matters most of all.
I've attended public speaking courses, which tend to emphasize polish, but seeing some of the most effective speakers using fillers like 'ums' 'ahs' and 'you know' just goes to show polish doesn't matter all that much. Reality supersedes theory.
There is no Income-As-A-Service available on the internet or in industrialized society.
Real technology liberates people from labor and minutia. What we have today could be thought of as phantom technology (similar to phantom wealth) in that it gives the appearance of providing labor-saving devices and software without actually providing such things:
http://neweconomy-wg.thenextsystem.org/visions/living-wealth...
Under our current capitalist system, no amount of income frees someone from the daily obligations and distractions that slowly drain away the time that each of us is allotted. So we end up with millionaires that have less time than ever, or that feel undignified that they can't get out of The Matrix, who then impress their toxic worldview on things like our political system, perpetuating servitude on the masses and future generations.
The vast majority of people working in technology are underemployed. University researchers spend too much time raising funding. Internet startups can raise millions and even billions of dollars if they find creative ways to turn dollars into profit, but there seems to be no funding to permanently solve problems like energy, hunger, disease, etc etc.
People have tried to create their own off-the-grid societies, but these have so far been mostly unsuccessful. I'm not aware of any similar online groups that one could join and obtain a stipend just by being a member. Open source or crowd funding could possibly go this direction someday, feeding profits back into the groups instead of siphoning them off to private shareholders.
IMHO the way to solve this is to get back to a grant-based system. Give researchers, hackers and makers the funding they need to solve the real problems facing the world, and we'd see outsized returns on investment almost overnight. That's why I'm a huge proponent of UBI. And also why I think the entrenched forces of the status quo will never let it happen.
I am not of the "startups do everything better" variety, but I do agree that a ton of effort goes into upgrading yesterdays ideas. Without going to political I really think it has to do with the ownership model. It is the owners that hold the position in the market. Today we aren't seeing that many spinoffs anymore either from what I can tell. In fact most acquisitions are the opposite. And I know many people think startups are a solutions, but it is really a big barrier to overcome and especially to go on to become an independent company.
More importantly, though, the article I was being interviewed about is one I cowrote with Michael Nielsen, who does have a lot of first-hand experience: https://en.wikipedia.org/wiki/Michael_Nielsen. (And, as part of writing it, we spent time discussing the questions with many practicing scientists across a lot of different fields.)
There are plenty of ways to improve research productivity on the margin, but I don't think that's the point of the argument here. The argument isn't that research productivity is getting worse because of X, and to solve it we need to do Y. The argument is research productivity is getting worse because of X, and because X is 'the laws of physics', this slowdown is a feature of the universe, not a feature of our poor research methodologies. It's inherently unavoidable, to some extent.
The plain fact is that everything we do is limited by (1) the laws of physics and (2) the space of possible inventions. In the early days of the industrial revolution, we didn't really know the laws of physics and we hadn't mapped out the space of possible inventions. But today is qualitatively different. Broadly, research productivity is hitting diminishing returns. The key questions here are (1) how do we know this to be true? and (2) why is this happening?
(1) is touched on by a number of books and papers. You can look at progress in transportation. The gains of today are pitiful compared to the gains of 1800-1950, where we went from horses to steam locomotives to airplanes to rockets. You can look at progress in medicine. We made progress earlier by discovering that germs existed and that we should wash our hands. Equivalent gains today now take billions in drug R&D. Really the only sector where we've been doing well over the past few decades is IT, and now even that's maturing. Moore's law is dead, the gold rush for the web is over, and smartphone penetration is approaching its natural limit of 100%.
The two key readings I would recommend here are Robert Gordon's 'The Rise and Fall of American Growth' and Nicholas Bloom et al.'s 'Are ideas getting harder to find?'
On (2), I think it's worth emphasizing that electromagnetism was mostly figured out in the late 1800s. Quantum mechanics was mostly figured out in the early 1900s. The essentials of nuclear energy were figured our in the early to mid 1900s. Even though we know gaps in our physics knowledge remain, these gaps are not going to overturn what we already do know. And what we already do know really constrains the efficiency of our designs. We're not gonna make engines that beat the Carnot limit. We're not gonna make vehicles that are simultaneously (1) fast, (2) energy efficient, and (3) move through the atmosphere. We're not gonna find chemical fuels that are denser than hydrocarbons. We're never going to beat entropy in improving the energy efficiency of ore smelting. Etc.
Now, I don't mean to establish a straw man that progress is over, that innovation is impossible, etc. But I really do believe that design space is finite and there are diminishing returns in its exploration. Before you believe it, it sounds crazy and defeatist, but after you believe it sounds banal and obvious.
1. I think you could to some extent William Gibson this though. "The future is already here — it’s just not evenly distributed". Maybe we have made less progress, but it also looks far less because it isn't being obviously distributed. (I guess in regards to Moore's law you can remove "not evenly" from the quote).
2. I agree. I wonder if this doesn't have a lot to do with misguidance. Basically a lot of our ideas of the future are old. Therefor we want self-driving cars, even though we know cars are limited in the sense that they have non-specific applications. So a lot of what people see as the future is a non-starter from the beginning.
I will have to read some of that material. I do think that science could be getting harder. On the other hand I am not sure we are even making the effort. Space exploration is something that get really hard really quick. But there are certainly a bunch of things you could at least try to explore in our solar system. But I guess that progress can be hard to quantify as well.
In addition, from my perspective, research isn't just about what we find out in an isolated field, but also the interplay of discoveries and knowledge between fields. This latter view, 'integration', is largely unexplored at scale, though lip service is constantly being given to it. I think there is a tremendous amount we can do here, and it's not just marginal. Overall, I am more optimistic I think, perhaps to an unwarranted degree :)
I also think that language is important, especially in the service of scientific understanding and progress. In this respect, I think category theory is poised to play a major role as a language for the sciences, and particularly integrative science in the 21st century.
I've been working on this in my spare time: https://www.eonias.org/, which I definitely hesitate to link to here because it's unfinished, but I figure I might as well start putting it out there. I think there are lots of ways to improve the way we go about research, having been in the grind for a few years now. There are tons of different opinions on this, but I just wanted to do something.
In short, I personally am not concerned whether 'research productivity is getting worse because of X, and because X is "the laws of physics"' is actually true or not (this is a complicated question, and the relevant question is whether 'productivity' can go up again--I think it can). I believe there's plenty left to do, and we need far more people doing it than we have engaged now. After all, should we move the money from the budget of the NSF into tax incentives for football stadiums? From my perspective, I think the NEH should have a much larger budget, which in and of itself is an interesting topic of debate. I mean, where have the humanities gone?? I'm f@#C!ing tired of AI and robots already. ;)