Tag Archives: Innovation

Automation vs. Innovation

We don’t yet know how to make computer software that is as flexibly smart as human brains. So when we automate tasks, replacing human workers with computer-guided machines, we usually pay large costs in flexibility and innovation. The new automated processes are harder to change to adapt to new circumstances. Software is harder to change than mental habits, it takes longer to conceive and implement software changes, and such changes require the coordination of larger organizations. The people who write software are further from the task, and so are less likely than human workers to notice opportunities for improvement.

This is a big reason why it will take automation a lot longer to replace human workers than many recent pundits seem to think. And this isn’t just abstract theory. For example, some of the most efficient auto plants are the least automated. Read more about Honda auto plants:

[Honda] is one of the few multinational companies that has succeeded at globalization. Their profit margins are high in the auto industry. Almost everywhere they go — over 5 percent profit margins. In most markets, they consistently are in the top 10 of specific models that sell. They’ve never lost money. They’ve been profitable every year. And they’ve been around since 1949. …

Soichiro Honda, the founder of the company … was one of the world’s greatest engineers. And yet he never graduated college. He believed that hands-on work as an engineer is what it takes to be a great manufacturer. … Continue reading "Automation vs. Innovation" »

GD Star Rating
a WordPress rating system
Tagged as:

Irreducible Detail

Our best theories vary in generality. Some theories are very general, but most are more context specific. Putting all of our best theories together usually doesn’t let us make exact predictions on most variables of interest. We often express this fact formally in our models via “noise,” which represents other factors that we can’t yet predict.

For each of our theories there was a point in time when we didn’t have it yet. Thus we expect to continue to learn more theories, which will let us make more precise predictions. And so it might seem like we can’t constrain our eventual power of prediction; maybe we will have powerful enough theories to predict everything exactly.

But that doesn’t seem right either. Our best theories in many areas tell us about fundamental limits on our prediction abilities, and thus limits on how powerful future simple general theories could be. For example:

  • Thermodynamics – We can predict some gross features of future physical states, but the entropy of a system sets a very high (negentropy) cost to learn precise info about the state of that system. If thermodynamics is right, there will never be a general theory to let one predict future states more cheaply than this.
  • Finance – Finance theory has identified many relevant parameters to predict the overall distribution of future assets returns. However, finance theory strongly suggests that it is usually very hard to predict details of the specific future returns of specific assets. The ability to do so would be worth such a huge amount that there just can’t be many who often have such an ability. The cost to gain such an ability must usually be more than the gains from trading it.
  • Cryptography – A well devised code looks random to an untrained eye. As there are a great many possible codes, and a great many ways to find weaknesses in them, it doesn’t seem like there could be any general way to break all codes. Instead code breaking is a matter of knowing lots of specific things about codes and ways they might be broken. People use codes when they expect the cost of breaking them to be prohibitive, and such expectations are usually right.
  • Innovation – Economic theory can predict many features of economies, and of how economies change and grow. And innovation contributes greatly to growth. But economists also strongly expect that the details of particular future innovations cannot be predicted except at a prohibitive cost. Since knowing of innovations ahead of time can often be used for great private profit, and would speed up the introduction of those innovations, it seems that no cheap-to-apply simple general theories can exist which predict the details of most innovations well ahead of time.
  • Ecosystems – We understand some ways in which parameters of ecosystems correlate with their environments. Most of these make sense in terms of general theories of natural selection and genetics. However, most ecologists strongly suspect that the vast majority of the details of particular ecosystems and the species that inhabit them are not easily predictable by simple general theories. Evolution says that many details will be well matched to other details, but to predict them you must know much about the other details to which they match.

In thermodynamics, finance, cryptography, innovations, and ecosystems, we have learned that while there are many useful generalities, the universe is also chock full of important irreducible incompressible detail. As this is true at many levels of abstraction, I would add this entry to the above list:

  • Intelligence – General theories tell us what intelligence means, and how it can generalize across tasks and contexts. But most everything we’ve learned about intelligence suggests that the key to smarts is having many not-fully-general tools. Human brains are smart mainly by containing many powerful not-fully-general modules, and using many modules to do each task. These modules would not work well in all possible universes, but they often do in ours. Ordinary software also gets smart by containing many powerful modules. While the architecture that organizes those modules can make some difference, that difference is mostly small compared to having more better modules. In a world of competing software firms, most ways to improve modules or find new ones cost more than the profits they’d induce.

If most value in intelligence comes from the accumulation of many expensive parts, there may well be no powerful general theories to be discovered to revolutionize future AI, and give an overwhelming advantage to the first project to discover them. Which is the main reason that I’m skeptical about AI foom, the scenario where an initially small project quickly grows to take over the world.

Added 7p: Peter McCluskey has thoughtful commentary here.

GD Star Rating
a WordPress rating system
Tagged as: , , , ,

I Still Don’t Get Foom

Back in 2008 my ex-co-blogger Eliezer Yudkowsky and I discussed his “AI foom” concept, a discussion that we recently spun off into a book. I’ve heard for a while that Nick Bostrom was working on a book elaborating related ideas, and this week his Superintelligence was finally available to me to read, via Kindle. I’ve read it now, along with a few dozen reviews I’ve found online. Alas, only the two reviews on GoodReads even mention the big problem I have with one of his main premises, the same problem I’ve had with Yudkowsky’s views. Bostrom hardly mentions the issue in his 300 pages (he’s focused on control issues).

All of which makes it look like I’m the one with the problem; everyone else gets it. Even so, I’m gonna try to explain my problem again, in the hope that someone can explain where I’m going wrong. Here goes.

“Intelligence” just means an ability to do mental/calculation tasks, averaged over many tasks. I’ve always found it plausible that machines will continue to do more kinds of mental tasks better, and eventually be better at pretty much all of them. But what I’ve found it hard to accept is a “local explosion.” This is where a single machine, built by a single project using only a tiny fraction of world resources, goes in a short time (e.g., weeks) from being so weak that it is usually beat by a single human with the usual tools, to so powerful that it easily takes over the entire world. Yes, smarter machines may greatly increase overall economic growth rates, and yes such growth may be uneven. But this degree of unevenness seems implausibly extreme. Let me explain. Continue reading "I Still Don’t Get Foom" »

GD Star Rating
a WordPress rating system
Tagged as: , , ,

Proverbs As Insight

Don Quixote’s lower class sidekick Sancho Panza quoted proverbs to excess. Among the intellectuals I know that class association continues – proverbs may help lesser minds, but we elites “think for ourselves.” Proverbs are also associated with older beliefs and attitudes, and so are seen as more politically conservative, and less relevant in our new changed world. Since the world today changes faster, has become less politically conservative, and has more educated folks who aspire to look more intellectual, you might think that we use proverbs less today than we did in 1800.

On the other hand, you might think of proverbs as well-packaged nuggets of useful insight. As the world continues to grow by accumulating insight and innovation, not only do we collect more gadgets, formulas, and words, we should also be collecting more useful proverbs. From this perspective, we should expect people to use more proverbs today.

To get some data on this, I found some lists of famous proverbs, and used Google books ngram viewer to plot their usage in books since 1800:

ProverbUsage

ProverbUsage2Overall usage seems to have gone up, not down. But two considerations complicate this interpretation. One is that I started from lists of proverbs famous today, instead of proverbs famous in 1800. The other is that the typical book reader and author today may be more lower class than they were in 1800, with books catering more to their proverb-friendly tastes.

I hope someone can get better data on this. Even so, maybe we should tentatively expect future folk to talk and write more like ole Sancho Panza, with many more proverbs.

GD Star Rating
a WordPress rating system
Tagged as:

The Up Side Of Down

In her new book, The Up Side of Down: Why Failing Well Is the Key to Success, Megan McArdle takes some time to discuss forager vs. farmer attitudes toward risk.

Forager food sources tended to be more risky and variable, while farmer food sources are more reliable. So foragers emphasized food sharing more, and a tolerate attitude toward failure to find food. In contrast, farmers shared food less and held individuals responsible more for getting their food. We’ve even seen the same people switch from one attitude to the other as they switched from foraging to farming. Today some people and places tend more toward farmer values of strict personal responsibility, while other people and places tend more toward forager forgiveness.

McArdle’s book is interesting throughout. For example, she talks about how felons on parole are dealt with much better via frequent reliable small punishments, relative to infrequent random big punishments. But when it comes to bankruptcy law, a situation where the law can’t help but wait a long time to respond to an accumulation of small failures, McArdle favors forager forgiveness. She points out that this tends to encourage folks who start new businesses, which encourages more innovation. And this does indeed seem to be a good thing.

Folks who start new businesses are pretty rare, however, and it is less obvious to me that more leniency is good overall. It is not obvious that ordinary people today face more risk than did most farmers during the farming era. The US apparently has the most lenient bankruptcy law in the world, and that is indeed some evidence for its value. However, it seems to me more likely that US forager forgiveness was caused by US wealth than vice versa. McArdle says the US got lenient bankruptcy in the late 1800s via lobbying by senators representing western farmers in debt to eastern banks. And it is even harder to see how farming in the US west then was more risky than has been farming throughout the whole farming era.

Most likely what changed was the wealth of US farmers, and their new uppity attitudes toward rich elites. This fits with debt-forgiveness being a common liberal theme, which fits with liberal attitudes being more forager-like, and becoming more common as rising wealth cut the fear that made farmers. If lenient bankrupts is actually better for growth in our world, this would be another example of Caplan’s idea trap, where rising wealth happens to create better attitudes toward good policy.

Overall I found it very hard to disagree with anything that McArdle said in her book. If you know me, that is quite some praise. :)

Added 2May: The fact that most farmer cultures were clannish may be part of an explanation here. The strict farmer morality is mostly about how to deal with outsiders, distant from your immediate family. The clan is punished severely, but it is usually more forgiving internally. If farmer clans had lower risk than do isolated families today, that could be a reason to have more forgiving bankruptcy law today.

GD Star Rating
a WordPress rating system
Tagged as: , , ,

The Future Of Intellectuals

Back in 1991, … [a reporter] described Andrew Ross, a doyen of American studies, strolling through the Modern Language Association conference … as admiring graduate students gawked and murmured, “That’s him!” That was academic stardom then. Today, we are more likely to bestow the aura and perks of stardom on speakers at “ideas” conferences like TED. …

Plenty of observers have argued that some of the new channels for distributing information simplify and flatten the world of ideas, that they valorize in particular a quick-hit, name-branded, business-friendly kind of self-helpish insight—or they force truly important ideas into that kind of template. (more)

Across time and space, societies have differed greatly in what they celebrated their intellectuals for. Five variations stand out:

  • Influence – They compete to privately teach and advise the most influential folks in society. The ones who teach or advised kings, CEOs, etc. are the best. In many nations today, the top intellectuals do little else but teach the next generation of elites.
  • Attention – They compete to make op-eds, books, talks, etc. that get attention from the intellectual-leaning public. The ones most discussed by the snooty public are the best. Think TED stars today, or french public intellectuals of a generation ago.
  • Scholarship – They compete to master stable classics in great detail. When disputes arise on those classics, the ones who other scholars say win those disputes are the best. Think scholars who oversaw the ancient Chinese civil service exams.
  • Fashion – They compete to be first to be visibly associated with new intellectual fads, and to avoid association with out-of-fashion topics, methods, and conclusions. The ones who fashionable people say have the best fashion sense are the best. Think architecture and design today.
  • Innovation – They compete to add new results, methods, and conclusions to an accumulation of such things that lasts and is stable over the long run. Think hard sciences and engineering today.

Over the last half century, in the most prestigious fields and in the world’s dominant nations, intellectuals have been celebrated most for their innovation. But other standards have applied through most of history, in most fields in most nations today, and in many fields today in our dominant nations. Thus innovation standards are hardly inevitable, and may not last into the indefinite future. Instead, the world may change to celebrating the other four features more.

A thousand years ago society changed very slowly, and there was little innovation to celebrate. So intellectuals were naturally celebrated for other things that they had in greater quantities. The celebration of innovation got a big push from World War II, as innovations from intellectuals were seen as crucial to winning that war. Funding went way up for innovation-oriented intellectuals. Today, however, tech and business startups, and innovative big firms like Apple, have grabbed a lot of innovation prestige from academics. Many parts of academia may plausibly respond to this by celebrating other things besides innovation where those competitors aren’t as good.

Thus the standards of intellectuals may change in the future if academics are seen as less responsible for important innovation, or if there is much less total innovation within the career of each intellectual. Or maybe if intellectuals who are better at doing other things besides innovation to win their political battles within intellectual or wider circles.

If intellectuals were the main source of innovation in society, such a change would be very bad news for economic and social growth. But in fact, intellectuals only contribute a small fraction of innovation, so growth could continue on nearly as fast, even if intellectuals care less about innovation.

(Based on today’s lunch with Tyler Cowen & John Nye.)

GD Star Rating
a WordPress rating system
Tagged as: , ,

Rah Manic Monopolists?

The vast majority of economic growth is caused by innovation. So when it comes to long term policy, innovation is almost the entire game – whatever policy causes substantially more innovation is probably better, even if has many other big downsides.

One simple robust solution to the innovation problem would seem to be manic monopolists: one aggressively-profit-maximizing firm per industry. Such a firm would internalize the entire innovation problem within that industry, all the way from designers to suppliers to producers to customers – it would have full incentives to encourage all of those parties to put nearly the right amount and type of efforts into innovation.

Yes, even monopolists don’t have exactly the right incentives. They will tend to focus on what marginal customers want, at the expense of both lower-value customers pushed out by inflated monopolist prices, and higher-value infra-marginal customers. And when innovations can cross industry boundaries, industry monopolists may also fail to coordinate with monopolists from other industries. But still, this approach seems to get a lot closer to optimal that anything other simple policy. And if two industries had enough innovation interaction, one might just have a single firm cover both industries.

Ideally these monopolies would be global, but if not national ones might still be a big win over the status quo.

Admittedly, common intuitions don’t agree with this. For one thing we tend to think of monopolists as too lazy to innovate – it takes competition to push them out of their comfort zone. And I agree that this is a common situation for regulated utilities and government agencies. Often the employees of a monopolist tend to have enough political power to entrench themselves and resist change, at the expense of investors and customers. This is why I specified manic monopolists – we need investors to have enough power to impose their will, and we need to have  enough competition to fill these investor roles.

Yes, we also tend to be uncomfortable with the inequality and power concentration that manic monopolists would embody and require. It isn’t at all what foragers are prone to praise. But still, if innovation is important enough, shouldn’t we be willing to tolerate a lot more inequality to get it?

Added 8a 11Apr: In general, industries that are more concentrated, i.e., more in the direction of having a monopolist, have more patents, all else equal. This seems to be because they invest more in R&D. Data here, here.

GD Star Rating
a WordPress rating system
Tagged as: ,

Let Re-Discovery Evade Patents

In this post I’m going to explain why patents can be a good idea, why they often go wrong today, and a way to fix that problem. And I’ll do that all in the context of a situation you should understand well: finding a shorter route to drive from home to work. (This post is ~1600 words, and so longer than usual.)

Imagine that you usually take a particular route from home to work, and some firm offers to find you a better route. You tell them your current route, and they tell you that they have found a different route that will save you thirty seconds a day, which over a year adds up to eight hours. You can inspect their route to verify their claim, but only if you agree that you can’t use that route (or anything close) unless you pay them a mutually agreeable fee. (Assume they can enforce that, by seeing your car’s driving path records. And assume you can verify their claim somehow.) You agree, inspect and verify, and then agree to pay them one hundred dollars, which is well below your value of saving eight hours of driving, and above their cost of finding the route.

This example contains an info property right: once you agree not to use their route unless you pay for it, then they own a right to your use of that route. Since the route is info, what they own is info. The prospect of owning that info right gives the firm an incentive to work to find that route. And because they must find a mutually agreeable price, their incentive to work is neither too much nor too little. An agreeable price must lie between their cost of finding the route and your added value from using it.

Now imagine that you are one of hundreds of drivers who go from the same initial home area to the same final work destination. Now this route-finding firm wants to sell a better route to all of you. But there is a problem. Once this firm sells the route to a few of of you, the others may learn of that route from these few buyers, either by being told or by following their cars. In this case the total price the firm could get from all the drivers might be much less than the sum of driver values for using the better route. Thus the firm’s incentive to work to find a better route could be too low. That is, this group of drivers could be better off it they joined together to paid the firm more to find a better route. But joining is too hard, so it doesn’t happen. Continue reading "Let Re-Discovery Evade Patents" »

GD Star Rating
a WordPress rating system
Tagged as: ,

Drexler and I Again

Eric Drexler has responded to my last reply. Let me focus on one key issue. I wrote:

The main argument you gave for why a nanotech revolution could happen suddenly is that new nanotech designs could “unfold at the speed of new digital media”, i.e., we could sent such designs around fast as digital files. But if this were all that was needed for a technology to improve rapidly we should now see rapid gains in the design of novels, music, and software.

Drexler responds with quotes from his book:

Even partial upgrades of existing products that involve [merely] replacing structural components with materials that are lighter, stronger, and lower in cost can offer striking advantages. If a business today could deliver replacements for products already in use, but at lower cost and with superior performance by a few key metrics (vehicles with half the mass, electronic systems with ten thousand times greater capacity), one would expect to see rapid replacement of competing products along with the collapse of the supply chains behind them. …

Cycles of product improvement (and replacement) can be swift with an APM production infrastructure; the delays of prototyping, production engineering, and plant construction largely disappear, and production itself can be both fast and scalable. Further, for products adapted to decentralized APM-based production, distribution need not involve shipping and can more nearly resemble an Internet download.

Yes, if a broad mature nanotech ability were to drop out of the sky, then industry could use such an ability to rapidly to displace existing products with large efficiency gains. A sudden appearance of full nanotech would imply a big sudden social change. But the question here is exactly how fast would nanotech abilities appear!

Nanotech production lines take very small chemicals and incrementally bond them to each other, accumulating larger and larger assemblages, until they are big enough to be useful devices. Imagine that such production lines slowly became cheaper, faster, and more reliable, slowly adding to the menu of chemicals they could take in as basic building blocks, and slowly able to reliably create a wider range of chemical bonds at a wider range of relative block orientations. Slowly more of the steps in this production process became more fully automated, and less guided by human intervention. The slower that these improved abilities appeared, the slower would be the gains in performance and cost of the devices made this way.

Today the industries that create novels, music, and software all have the advantages Drexler foresees – they have little in the way of tech-induced delays of prototyping, production engineering, and plant construction. Production itself is both fast and scalable. Even so, those industries are not improving the efficiency of their products at rates much faster than when they suffered greatly from such delays. So the elimination of such delays is clearly not sufficient to imply much faster gains in final product value.

If there are reasons to expect nanotech abilities to improve rapidly, they must be additional reasons beyond those given above.

GD Star Rating
a WordPress rating system
Tagged as: , ,

R&D Is Local, Global, But Not National

A recent Post article by Brad Plumer illustrates what is wrong with the usual research funding arguments:

One of the few things Republicans and Democrats have been able to agree on in recent years is that the government should be spending more on basic scientific research … Thanks to budget pressures and the looming sequester cuts, federal R&D spending is set to stagnate in the coming decade. …

As a result, scientists and other technology analysts are warning that the United States could soon lose its edge in scientific research — and that the private sector won’t necessarily be able to pick up the slack. “If you look at total R&D growth, including the corporate and government side, the U.S. is now at the low end … We’re seeing other countries, from Germany to Korea to China, make much bigger bets.” …

There’s a long, long list of world-changing innovations that can be traced back to federally funded R&D over the years. .. The key question here is how much of this innovation might have happened without government involvement. … Many economists agree that private companies tend to under-invest in very basic scientific research, since it’s hard for one firm to reap the full benefits from those discoveries. …

When the Congressional Budget Office reviewed the evidence in 2007, it concluded that government-funded basic research generated “substantially positive returns.” And it found that, on the whole, government R&D helped spur additional private-sector R&D rather than displace it. … The United States will soon spend less on all types of R&D as a percentage of its economy in the coming decade than countries like Australia and South Korea …

The sanguine view is that other countries are tossing more money at scientific research that will have positive spillover benefits for the entire world — including us. If China invents a cure for cancer, we all benefit. Others worry, however, that the U.S. economy could suffer from the fact that a greater share of research is happening elsewhere. (more)

Note the conflicting arguments: each small part of the world invests too little in R&D, because other parts gain without paying, but the US should fear falling behind nations that invest more. These two only makes sense together if the nation is the natural scale for innovation – innovations mostly leak away from their source within a nation, but mostly stay within each nation. The academic literature, however, suggests the natural scales are global and local – while there are gains to the world as a whole, gains are focused on related industries in the local area:

A recent body of empirical evidence clearly suggests that R&D and other sources of knowledge not only generate externalities, but such knowledge spillovers tend to be geographically bounded within the region where the new economic knowledge was created (Griliches 1992). That is, new economic knowledge may spill over, but the geographic extent of such knowledge spillovers is limited. … greater geographic concentration of production actually leads to more, and not less, dispersion of innovative activity. (more; see also and also)

While it would be great if the world could coordinate to promote R&D spending worldwide, there is little economic justification for forcing Wyoming and Louisiana, who spend 0.4% and 0.56% of GDP respectively on R&D, to pay for R&D spending in Massachusetts and New Mexico, where those figures are 5.49% and 7.65% (source), any more than the rest of the world pays for such spending. If the US government funds less R&D, it will be mainly states like Massachusetts and New Mexico that suffer, not states like Wyoming and Louisiana, relative to the rest of the world.

If R&D spending mostly helps the particular regions in which it happens, why do we pay for it at the national level? Probably because many see it as a national prestige good – people in Wyoming look good to foreigners by being in a nation where lots of impressive research happens in Massachusetts. Are they right, or is Massachusetts just getting a nice juicy transfer?

GD Star Rating
a WordPress rating system
Tagged as: , ,