omalleyt

Inflation dynamics are not like stock-market dynamics.

Widespread stock-market optimism is a self-denying belief. Investors reach peak optimism, fully invest, and then the market can only go down.

Widespread inflation expectations is a self-fulfilling belief. People dump dollars for physical goods, increasing the price of those goods. This causes more inflation, leading to further dumping of the dollar, and further price increases.

This is why runwaway inflation / hyperinflation occurs.

Gold.

That’s putting a lot of faith on the gov’t to correctly report CPI, when they have every incentive to underreport it.

In fact, they’ve changed the CPI methodology twice in order to report lower average inflation (under the premise that the old measures “overstated” inflation).

There’s a level of dysfunction that causes a company to acknowledge the issue and hire consultants. And then there’s a level of dysfunction that causes the company to not acknowledge the issue and not hire consultants. The latter level is higher

Every essay is about how and why Paul Graham had a hard time in high school

> And, while there's certainly some intrinsic value in an S&P 500 listing - as an advertising device, if nothing else - it's hard to imagine that it's worth $50bn

The value of being in an index is that 80% of the market is passive investing, most of which tracks indices. If TSLA were added to the S&P500, all of these index-tracking funds would have to buy it in order to continue to track the index

Unfortunately, your intelligent agents qualify as optimization algorithms and therefore the No Free Lunch Theorem applies:

https://ti.arc.nasa.gov/m/profile/dhw/papers/78.pdf

I.e. across the space of all possible environments, all agents perform equally well

Citizens would be the shareholders

What was the treatment?

That would take extra calories, that's where the balance is

Industrial engineer by training here, I honed my chops on Kaggle competitions, eventually winning one. I was recently brought into a FAANG in a ML/AI engineering capacity.

I can’t reccomend Kaggle enough for those who are looking to prove their abilities in the field.

Look at it this way: human ability to act is (for elite gamers) 300 actions per minute. That's 5fps. So with 30fps the AI could theoretically already have 1/6th the latency of the most-responsive human drivers

I think there was some inevitable loss of nuance as that got translated into layman's terms.

For all practical purposes, you can write causality like

p(x|y) = 1 and time(y) < time(x)

I.e. causality is just when one event always happens after another event. Any additional requirements for causality are basically philosophy.

But typical ML systems don't construct networks of causal relations, is basically what he's getting at from my reading of it

Well, "appeal to consequences of belief" is a logical fallacy. But yes, of course it's possible that inaccurate views of the world can actually bring about better outcomes than accurate ones

It's showing him Wawas he's been to before. That's what the clocks mean.

For the search "wawa" it doesn't make a lot of sense to prioritize history over distance, because Wawas are nearly identical. But for "farmer's markets" this would probably be a useful feature.

Figuring out how to differentiate between those two examples algorithmically is super non-trivial though

Societies compete with each other like organisms. Successful societies propagate. Primary dimensions of competition are demographics, economics, and military. Thus the history of social organization is necessarily a history of extracting ever-more value from the labor of society's members.

Keynes and others seem to have labored under the delusion that the history of social organization is a history of ever-increasing happiness for society's members.

Google doesn't run those images through a pixel-level ConvNet, Google's algo uses a combo of the HTML info surrounding the image and the image's popularity. Running every image on the web through a CNN would be computationally infeasible

ConvNets for image classification (like what the article is about) often achieve better-than-average-human performance

I think there's a simple explanation for this.

Increased head size == greater risk of death in childbirth

Female hips have already evolved to be as large as possible while still being load-bearing.

It would be interesting to see what the wide-spread use of Caesarean sections would do to the human brain, if we weren't all going to be transhuman within a few generations anyway

As far as whether there are diminishing returns to increased intelligence within one individual, I'd say try taking two people with half of Einstein's IQ and asking them to explain General Relativity

If I was a celebrity, I could see this being huge. I could only give out my virtual number even to friends and family. And then I could have my virtual number leaked but it wouldn't matter, bc I only whitelist my friends and family on that number (if I'm understanding this product correctly)

I'm really hoping they actually mean an adaption of some period from The Simarillion. Maybe Feanor's story? Something like that would be epic

I had this problem all my life. Now I take 500 micrograms of melatonin every night an hour before bed. I'm out like a light.

It might not be for everyone, but it's changed the way I feel about sleeping. And it's non-habituating

The funny thing is the author thinks having an 81% success rate of predicting which image out of a pair of images is of a gay person, is not as hard as predicting with 81% accuracy whether any given individual is gay or straight.

Meaning the author really doesn't understand the concept of unbalanced classes

Well, that's the problem with the status quo, that wouldn't be the problem if universities switched to an income sharing strategy as proposed. I was assuming we were talking about "unprofitable"-but-valuable majors in a world in which income sharing has been implemented

Sure, I'd be good with that. I also think that if there are some things that are non-profitable but still valuable to society (and no doubt, there are), then the problem is really that the agent isn't capturing enough of the value they are creating for society, and society ought to find an incentive mechanism that changes that. But since we live in a practical world and that is an impractical demand, I agree a hybrid model would be beneficial

Awesome, that's great to hear. And I completely agree with your last paragraph, my concern (without knowing the relevant laws) was always just what a judge would hear. Really glad to see someone doing this

Brilliant, I've long thought that all universities should run under this model (although when applied to the humanities I can hear in my head the 3000 years of the land owning, non-laboring intelligentsia protesting about the noble virtues of an education uncorrupted by the banalities of productive yada yada...) But anyway back to CS.

1) Risk is pooled in the institution rather than distributed amongst the students, which is the textbook way to deal with uncorrelated risk.

2) The incentive of the university and the student are aligned as much as possible

3) By putting costs and benefits into a form with equal time horizons, disadvantaged students no longer need to rely on the generosity of governments or private lenders for upfront cash.

The only thing I'd always questioned was whether such a scheme as described above could pass legal muster, as it bears a resemblance to involuntary servitude, as well as requiring access to income statements. I've never heard of anyone but the govt placing liens on income. I'm hoping a workaround has been found, because from a strictly incentive based analogy, this model has the potential to do to modern education what patent law did to manufacturing

Historical records did not begin 100 years ago, that much I don't think anyone can truly believe. We have detailed contemporary histories of the American Revolutionary War, for one, and they seem no more "biased" than histories of WWII.

There is a 500 year difference bt Herodotus and Tacitus. Even so, Herodotus did no more than accurately record what he was told, assiduously pointing out when he saw something first hand. Early ancient historians made up speeches, that much is known. But by Tacitus's day, the act of writing an objective history was not a novelty, and in fact he complains in the beginning of his text that he is undertaking the work because he thinks his peers, also ostensibly engaged in objective history, have not been objective enough out of fear or hatred when covering the reigns of Tiberius, Caligula, Claudius, and Nero. Once again, with no reason to doubt him, I ask: why not even consider taking one of the greatest historians whose works have been preserved for posterity at face value?

Is there anything in particular Tacitus says in any of his works that you doubt happened?

Well, it's not like I wholly disagree with what you're saying, but I think the paragraph above is a straw man of what I'm saying.

First, here is the element of Postmodern discourse I'm talking about:

https://en.m.wikipedia.org/wiki/Deconstruction

Second, my complaint is that it's not even considered as a possibility that Tacitus is telling the truth. In relation to The Annals (a public document concerning public affairs thousands of contemporaries would have had firsthand experiences of) do you really think it's a forgery? Or do you really think Tacitus is so biased he can't record history accurately? The idea that someone becomes so corrupted by their interests (especially race and class) that they cant be objective is a uniquely postmodern idea. The Enlightenment was premised on the idea that objective rationality was accessible to all.

Further, we can test your hypothesis about the Renaissance. The Renaissance writers read Tacitus, and when they mention him, they treat his works like a complete and unmitigated factual account. Also, Tacitus is not engaged in rhetoric. That's the point. He's writing history, not philosophy.

It's a symptom of postmodernism that nowhere in this text is it even suggested that Tacitus is maybe just, you know, relating the facts about Germancius as accurately as he can.

Instead we're sitting here quibbling over what literary fiction trope "Tacitus's Germanicus" fulfills in his "story"