ryan_b's Comments

Comprehensive COVID-19 Disinfection Protocol for Packages and Envelopes

Well that sucks. Take care of yourself and stay sane during isolation!

History's Biggest Natural Experiment

I feel like this is evidence for the natural experiment interpretation. This means we will get a steady stream of new findings as each maturation window approaches, for decades to come.

Are veterans more self-disciplined than non-veterans?

To be more exact, if you have a group, then the group provides social incentives; but social incentives do not imply a group. For example, if I were publicly humiliated in front of strangers, they might mock me if they saw me later in a restaurant. This is a social (dis)incentive, but the fact remains that we aren’t in a group.

What qualifies people as a group in the sense that I intend is at least twofold: they have to share the same set of incentives; this fact has to be common knowledge among them.

I do agree that if person trains successfully it would improve long-run discipline, but doing military training won’t meaningfully change the outcome from non-military training because the group context is what does the extra work. If that is not the focus, ie veterans are just an example disciplined population, then my comments are probably not relevant to the true concern.

Can crimes be discussed literally?

It seems to me precisely the opposite: my reading is that Benquo is driving exactly at how to talk about the problem of systemic falsification of information.

If the post is noncentral, what is the central thing instead?

Are veterans more self-disciplined than non-veterans?

I am a veteran, and my inside view suggests two things: one, the least disciplined members of the population are filtered out by the military (which is to say they are not accepted or kicked out early); two, the military experience pushes veterans towards the extremes.

Reasons to consider that veterans would be more productive than average:

  • Acclimated to long and/or strenuous work periods.
  • Better access to education through veterans programs and admission boosts.
  • Direct boost to employability in a variety of industries.

Reasons to consider that veterans would be less productive than average:

  • Higher rates of homelessness
  • Higher rates of mental illness and suicide
  • Higher rates of substance abuse
  • Etc.

My expectation is that the productivity advantage is highest when veterans enter a civilian industry that matches military tasks closely, like compliance with regulations or uncomfortable work environments. I also expect that the veterans who fail to re-adapt to civilian life suffer an almost complete collapse of productivity.

Turning to the question of discipline, I think we will benefit from a little context. Discipline in the military is very much a team phenomenon; Army training is focused overwhelmingly on establishing and maintaining a group identity. Most of the things people associate with military discipline require other people to make sense, like the chain of command, pulling security, and how tasks are divided. Even the individual things like physical fitness or memorizing trivia, are thoroughly steeped in the team environment because they are motivated by being able to help your buddy out and are how status is sorted in the group.

I believe your friend's statement:

If you can just train yourself like you're in the army, then you can become just as self disciplined as a soldier

is wrong as a consequence, because you can never train yourself like you are in the Army. That fundamentally needs a group, entirely separate from the question of social incentives and environment. Outside of the group context, discipline doesn't really mean anything more than habit formation.

LessWrong Coronavirus Agenda

If the same type of facility works for almost every kind of vaccine, do we think there would be interest in constructing the facilities as a speculative venture? Consider:

1. The economy is in chaos and may remain so, which I expect to produce unusually affordable access to design firms, construction crews, raw materials, and land.

2. There will be a strong incentive for regulators/inspectors to move with best speed, and the current administration at least in the US has a track record of being friendly to shortcuts.

3. If the facilities are already built, this allows a limit to the risk the companies producing the vaccines need to absorb in order to increase supply.

4. We could squeeze out unscrupulous opportunists.

Ways that China is surpassing the US

My model for this is that China is achieving success largely by ignoring externalities. Environmental pollution is a prime example, like in the case of their previous recycling policy and mining of rare earth minerals. It is actually against the law for the US to build as quickly or as cheaply as China, but this is reasonably motivated by trying to account for things like pollution and safety, and avoiding things like resettling entire towns.

Chinese success looks a lot like the WWII and postwar years in the US, and for much the same reasons.

Cortés, Pizarro, and Afonso as Precedents for Takeover
because why was it that the conquistadors were able to exploit the locals and not the other way around?

Have you considered the possibility that it was a case of mutual exploitation? The Aztec allies of the conquistadors weren't there out of the goodness of their hearts; they had found a new angle that would help them defeat Tenochtitlan. They lost the post-victory power struggle, but it was always going to be someone.

Have epistemic conditions always been this bad?
You make good points here. Any ideas why those other shifts happened and how can we help reverse them or prevent them from happening elsewhere?

Mostly it looks to me like a series of unrelated changes built up over time, and the unintended consequences were mostly adverse.

An example is the War on Cancer and the changes that came with it to funding. It had long been the case that funding was mostly handed out on a project-by-project basis, but in order to get the funding dedicated to cancer research it was necessary to explain how cancer research would benefit. The obvious first-order impact is an increase in administrative overhead for getting the money.

Alongside this science sort of professionalized. I expect that when the sense of how important something is permeates, professionalization is viewed as a natural consequence, but it seems to have misfired here. Professionalization, like other forms of labor organization, isn't about maximizing anything but about ensuring a minimum. This means things like more metrics, which is why our civilization formally prefers a lot of crappy scientific papers to a few good ones, and doesn't want any kind of non-paper presentation of scientific progress at all. Science jobs become subject to Goodharting, because people start thinking that the right way to get more science is just to increase the number of scientists, on account of them all being interchangeable professionals with a reliable minimum output.

The university environment also got leaned on as a lever for progress; the student loan programs all grew over this same period, which seems to have driven a long period of competition for headcount. This shifted universities' priorities from executing their nominal mission towards signalling desirability among students/parents/etc. I am certain at least part of that came at the expense of faculty, even if only by increasing the administrative burden still further by yet more metrics.

On the fixing side, I am actually pretty optimistic. A few simple things would probably help a lot, two examples being funding and organization. Example: Bell Labs and Xerox PARC have been discussed here a lot. Both cases deviated significantly from the standard university/government system of funding individual projects case by case. Under the project/grant system being a scientist reduces to being able to successfully get funding for a series of projects over time. At Bell and at PARC, they rather made long-term investments on a person-by-person basis. I think this has wide-ranging effects, but not least among them is that there wasn't a lot of administrative overhead to a given investigation; rather they could all be picked up, put down, or adapted as needed. Another effect, maybe intentional but seemingly happenstance, is that they built a community of researchers in the colloquial sense. This is pretty different from the formal employee relationships that dominate now. Around 7 years ago I listened to a recruiting pitch from Sandia National Laboratories for engineering students, and asked how communication was between different groups in the lab. The representative said that she knew of a case where two labs right across the hall from each other were investigating the same thing for over a year before they realized it, because nobody talks.

This suggests to me that a university that was struggling financially, or maybe just needed to take a gamble on moving up in the world, could cheaply implement what appears to be a superior research-producing apparatus, just by shifting their methods of funding and tracking results.

ryan_b's Shortform

Are math proofs useful at all for writing better algorithms? I saw on Reddit recently that they proved Batchelor's Law in 3D, the core idea of which seems to be using stochastic assumptions to prove it cannot be violated. The Quanta article does not seem to contain a link to the paper, which is weird.

Batchelor's Law is the experimentally-observed fact that turbulence occurs at a specific ratio across scales, which is to say when you zoom in on a small chunk of the turbulence it looks remarkably like all of the turbulence, and so on. Something something fractals something.

Looking up the relationship between proofs and algorithms mostly goes to proofs about specific algorithms, and sometimes using algorithms as a form of proof; but what I am after is whether a pure-math proof like the above one can be mined for useful information about how to build an algorithm in the first place. I have read elsewhere that algorithmic efficiency is about problem information, and this makes intuitive sense to me; but what kind of information am I really getting out of mathematical proofs, assuming I can understand them?

I don't suppose there's a list somewhere that handily matches tricks for proving things in mathematics to tricks for constructing algorithms in computer science?

Load More