thousandautumns

Last November, DeepMind released the results of a generative AI model that created theoretical molecular structures for over 2 million undiscovered synthetic materials. Within days, materials science researchers were able to confirm 700 of them. The sheer number of these new potential materials discovered is greater than has been created in the rest of human history combined. These are materials that can be used in manufacturing, energy production, and other objectives that are critical not just for advancing human society, but avoiding the impending crisis we are already facing.

Similar AI endeavors have been underway for medicine and human health.

The author is making extremely shallow, flawed arguments that hinge on an ignorant (or possibly, deliberately narrow-minded) understanding of what generative AI is, how it is already being used, and the magnitude of what is already being achieved with it.

> We urgently need the expertise of social scientists to be able to make much-needed collective decisions about the future of generative AI that we want; we can’t leave it to business, markets or technologists.

> Kean Birch is director of the Institute for Technoscience & Society at York University.

Academic sociologist argues that AI should be controlled by academic sociologists. Color me surprised.

Southern Indiana has lots of hills and forested areas, including the Hoosier National forest. The Great American Rail-Trail mostly goes through the upper third of the state, however, so you miss that and are mostly in the corn belt section of the state. Its very flat and mostly farm land. Which is likely why the railways were built on that route anyways.

I personally can find flat farmland quite beautiful, but I can imagine biking through it for days might get dull.

Must vary widely, because I went in August and there was 0 line at all. A lot of places are on Spring Break in the US right now, so that may contribute to your experience.

I'm not sure millenials or Gen-Xers are any less materially obsessed, I think we just generally have less money to spend on the things Boomers did. It isn't like we aren't spending thousands on tech, clothing, etc. I hesitate to condemn Boomers too harshly for their lifestyle choices given the benefit of hindsight when I'm sure my generation will also be pilloried for the aspects of lifestyle we are living that will later be seen as moronic.

More like lol at assuming your rent doesn't cover the other expenses of owning a house. You think your landlord is just taking the loss? When you rent, you are paying, if not directly, for every expense associated with the house, plus extra for the landlords profit margin.

Landlords absolutely incorporate home repair costs into rent. I'm shocked at the number of people who assume that landlords just eat the cost of maintenance. Just because you aren't forking over the money directly to the repairman doesn't mean you aren't paying.

Its Brazilian Ju Jitsu, for those wondering.

> I’m not sure if this is just an attempt to down play their results or if it’s more academic jealousy because the funding goes to the “cool stuff” like AI/ML in the CS dept. and the Stats dept. is seen as old and boring.

No one is trying to downplay the legitimately impressive results of AI/ML. Deep learning, convolutional neural networks and GANs have had incredible success in fields like computer vision, and image/speech recognition. But outside of those areas the "results" for the current fads in AI/ML learning have been grossly overstated. You have academic computer scientists like Judea Pearl decry the "backward thinking" of statistics and who are championing a "causal revolution", despite not actually doing anything revolutionary. You have modern machine learning touted ad nauseam as a panacea to any predictive problem, only for systematic reviews to show they don't actually out perform traditional statistical methods [1]. And you have industry giants like IBM and countless consulting companies promise AI solutions to every business problem that turn out to be more style than substance, and "machine learning" algorithms that are just regression.

There's a reason why AI research has gone through multiple winters, and why another is looming. Those in AI/ML seem to be more prone and/or willing to overpromise and underdeliver.

[1] https://www.sciencedirect.com/science/article/pii/S089543561...

God I hate this paper. Perhaps it was relevant at its time. But that was 18 years ago. The described dichotomy between the "two cultures" isn't nearly as pronounced, if it even exists, today. There are few statisticians today who adhere entirely to the "data modeling culture" as described by Breiman.

I'm surprised how often this paper continues to get trotted out. In my experience it seems to be a favorite of non-statisticians who use it as evidence that statistics is a dying dinosaur of a field to be superseded by X (usually machine learning). Perhaps they think if its repeated enough it will be spoken into existence?

> I am not an expert and am still reading thru the article, but why is it such a strong dichotomy?

There isn't. This paper is nearly 2 decades old and isn't nearly as relevant as you would think given how many times it gets trotted out. Any statistician under the age of 45 would not recognize the two cultures as they are described here.

> The standard best practices in STATS 101 is to compute R^2 coefficient (based on data of the sample)

> Null Statistical Hypothesis Testing (NHST), another one of the pillars of STATS 101

Best practices of statistics and what is taught in Stat 101 are not remotely the same thing. The problems with R^2 and NHST have been well documented they've been argued against for decades by the statistics community. But actual statisticians make up a small proportion of statistics practitioners, as statistics is the backbone of nearly all modern science. What gets taught in Stats 101 is not the "foundations of statistical best practices" so much as "a system of guidelines and rules of thumb that have been simplified greatly for the sake of the lowest common denominator". To make things worse, non-statisticians seem to over-estimate their statistical knowledge and prowess after having taken a handful of introductory stats courses more than any applied field I know of.

Only now after the magnitude and pervasiveness of the replication crisis has begun to be recognized by the broad scientific community are people starting to realize what many statisticians have been pointing out for years.

Too many excellent excerpts in this article.

> “It’s such a strange service — who wants to be in the cold? You need to hear about it from someone you trust,” said Michael Garrett, the head of Reboot, a spa that offers cryotherapy around the Bay Area. (Cryotherapy is when you make yourself get cold.)

And? The arguments about marketing are completely non-sensical. People treat pharma companies marketing budgets being as large as they are as some indictment of the industry, but its not like they are just throwing money at marketing because they like to burn cash. Money spent on marketing is money that the company has determined will result in a greater return in sales and is therefore a calculated investment. If the company were to learn that their marketing was more costly than the money it generated, they would be eager to cut back on marketing. The revenue then available for the company to use for operating expenses, including R&D, is therefore greater than it would be if the company conducted no marketing.

There are plenty of arguments to be had about how the pharmaceutical industry operates and how/should they be regulated in ways that other industries aren't, but the idea that marketing budgets are the reason for high drug prices is moronic.

You also see this kind of argument against housing a lot, which also ignores that housing is one of the few highly leveraged investments an individual can/will make, and that even if you don't buy a house you will have to pay for housing, all of which goes to someone else's pocket.

Probably lots. Its a huge issue thats driving the reproducibility crisis in science and academia, so I imagine it would also be a problem for industry.

Well first of all, blockchain technology has become so overhyped that its bound to draw scoffs even if the business case were legitimate. "Blockchain" anything almost sounds like a joke at first pass these days.

But in this specific instance, it is even more ridiculous as one of the biggest concerns surrounding blockchain technology is that it is insanely inefficient in terms of energy usage, and concerns have been raised about how widespread usage of blockchain technology could exacerbate climate change.

I've never looked at Data Science in R, but Hadley Wickam's R for Data Science is great in my opinion. Really applicable, down to earth, and focuses much more on the meat of data science (data manipulation and munging, visualization, relational data, and efficient programing) more than the typical "fit a neural network to this idealized toy data set!"

Its also available for free online at https://r4ds.had.co.nz/

Why? There's no point. If the results come back indicating multiple authors, great. If the results come back indicating a single author, you could just make the argument that it is the result of the book having a single translator.

Academic pursuits should have more meaning than trying to dunk on other people's belief systems when those belief systems are fairly harmless.

> I could spend unhealthy amounts of my time in those 100 Yen stores (or whatever they are called, I might be misremembering their names)

Daiso

There are two types of people consuming these projections. I'll call them the sportsbook type and the newspaper reader. The sportsbook consumers want to know what the probabilities are for the actual event, when it occurs (election day). The newspaper consumers want to know what is going on currently and the country's political temperature.

538's forecasts, especially far from election day, are for the newspaper consumer. They provide insight into current political dynamics, and a quantitative look at how people view policy, the economy, where current fundraising is at, etc. in a single concise number that can serve as the basis for further discussion and analysis. They however aren't useful for the sportsbook consumer, because they don't account for the practically innumerable "unknown unknowns" that will occur between now and election day (scandals, terrorist attacks, fluctuations in the economy, flubs on the campaign trail, external activist efforts gaining traction). If they did account for all that uncertainty, then the predictions consistently be near 50-50 for the majority of the race, which not only is still uninteresting for the sportsbook types, but also means you no longer are providing value to the newspaper types.

Its important to point out that, over time, the predictions become applicable to both readers. As election day nears, the number of "unknown unknowns" decreases, and a prediction for "what would happen if an election were held today" and "what will happen on election day" converges. So up until the few months and weeks leading up to election day, the predictions should be taken more as commentary, and then perceived as serious indications of what will occur closer to election day.

This is not correct, or at least it isn't any longer. The current predictions by 538 are the "lite" (polls only), the "classic" (polls + other data driven indicators) and the "deluxe" (polls + other data driven indicators + expert opinion). All of these are predictions based on "what would be the outcome of the election be were it to be held today". They just incorporate different data and complexity in their models.

Previously, the reported predictions were the "nowcast", "polls-only", and "polls-plus". But despite their names, I believe those too, whatever 538 may have described them as, were "predictions for outcomes as if the event occurred today". In fact, were they not the case, then Taleb's argument would actually have merit, despite his obtuse attempts to explain it.

Essentially, his argument is that, if you are predicting a particular outcome to an event (say, Clinton winning the election) a year from now with a certainty of 90%, and 3 months from now, your prediction is 75%, and 3 months after that is 85%, and then right before the election, is 70%, is evidence of a flawed model (keep in mind, I'm making these numbers up).

If your prediction is 90% a year out, it should fluctuate much as time goes on, because 90% is very strong certainty. One may counter that as time goes on, public opinion may change, there are changes in news coverage, and external events like the economy or scandal can make a big impact, and all of that is absolutely true. But the point is that, if in light of new information coming forward, your model predictions repeatedly exhibit volatility, than your model failed to adequately account for the uncertainty as a result of future events, and you were never actually 90% sure in the first place. By necessity, a model predicting a distant event in which lots of unknown factors can cause it to fluctuate before it occurs will either be fairly stable as time goes on (meaning you indeed were correct that there was a high chance of a Clinton win), or should give much more hedged estimates of a win (meaning your probabilities a year out would be much closer to 50-50).

Taleb's point is that 538's models are too volatile to actually be accurate when they say there is an X% chance someone wins an election a year, or 6 months from now. However, his argument does not apply in the case where the prediction is that there is an X% chance someone wins an election, were the election held today. Instead he seems to dismiss that that type of prediction is worthless, which is simply wrong and foolish in my opinion. There is lots to be learned about the current political climate from those predictions.

So its entirely possible both parties are partially wrong and right. Taleb is correct that 538's models are too volatile to be accurate predictions far out from the election, but it assumes that the model is a prediction of what will happen on election day, rather than what would happen today. He also is throwing out the baby with the bathwater by insisting that that is the only kind of political prediction that has any value. On the other hand, 538 has not been very clear on if their forecasts are predictions of what happens on election day or what happens were election day today. If its the former, then their models have issues and Taleb has at least some validity to his argument. If its the latter, then Taleb's argument is no longer applicable.

You can't verify probabilities for any single given event, but you can look at the outcomes for all events for which predicted probabilities were given, and use that to estimate the accuracy of the methodology/models used.

So, for example, when 538 said that Trump had a 30% chance of winning, its hard to say how accurate that was given it was a one time event. But if you look over all of their election predictions, and look at those they predicted to win 30% of the time, about 30% of them should have been victories (with some margin for error, of course). If 50% were victories, or only 10%, then maybe that would indicate their models aren't doing so hot.

Renault is partially owned by the French Government I believe

It might possible that most of the people in SF who own property have owned it for a long time. So for someone buying a house now, they would be paying a ton of money. If you bought the house more than 10 years ago, when prices were much lower, your mortgage would be much more reasonable and the monthly costs you would be paying would be less than what people are renting for currently.

Hence complaints of NIMBYism. The lack of new development in SF drives housing and therefore rental prices upwards, and so if you owned property before the bubble, you can now charge rent at the modern, high rental costs while paying a mortgage based on the lower pre-housing crisis prices. You could easily be pulling in thousands of dollars a month this way with a single rental property. So lots of people are now have a big incentive to fight against new development.

> So instead of shunning white supremacists from media - tolerate them on and challenge them with arguments, explain how they are wrong, educate them, ridicule their nonsense; but never censor them.

This takes the mistaken position that social media is a good tool for discussion and debate, and we have plenty of evidence to the contrary at this point.

This is also the same thought process that leads to ideologies like climate change-denialism, evolution-denialism, anti-vaxxing, etc. to be given oxygen when they shouldn't. It assumes that by shining the light on these topics, the debate should result in people rejecting them, but in reality it simply gives them more credibility. We live in a post truth era, where people don't use the internet to debate and agree on what is write and wrong and rethink our opinions, but to form bubbles with people who agree with what we want to believe, which we leave occasionally to scream at each other, and then retreat to in order to insulate ourselves from reality.

The people who are driven to non-mainstream forums such as 8chan that comprise the scum of the internet are already fairly radical in their thinking. That's why they seek out those sites. But non-radical thinkers may be susceptible to radicalization by giving a platform on a mainstream site like Twitter or Youtube to idealogies like white supremacy, militant Islam, or violent misogyny.

It isn't censorship. Free speech does not equate to Twitter/Facebook/Google having an obligation to host any/all content someone might post on their site.

The flipside being that you are making the argument that private entities should not be allowed to choose what content they host. You worry about flat earthers censoring non-flat earth depictions, but a much more realistic threat is flat-earthers forcing their insanity to have a larger share of voice than they should by strong arming platforms and people that know better into giving them a microphone, because were they not to do so, they would be accused of "censorship".

No one is saying the NZ video should be purged from the internet. But it is foolish to equate censorship with private social media companies scrubbing it from their sites. You are welcome to search out other venues that host the video, or host it yourself. Your rights are not being infringed upon if Twitter and Facebook won't.

Japanese people like English a lot. They all learn it in school. Most aren't that great, but they are good enough that English words and phrases are appealing, even if many of those phrases end up nonsensical.

If you visit Japan for any length of time, you will notice English phrases used all over the place, from restaurants, to street signs, to advertising and television, to the clothing people wear. I lived there for a few years right around the time Obama won the election in '08, and I encountered kids probably 6 years old who would shout "Yes we can!" when they saw me pass by. They tend to be more culturally aware and interested in Western culture (particularly American culture) than we do of them (which is saying something, given that Japanese culture is fairly popular in the West).

Not problems that San Francisco isn't also having. People like to prop up SF as an example of inclusiveness, as if it isn't the pinnacle of gentrification. Things may feel inclusive if you are white/asian and working in tech. Ask a black or hispanic person if they feel like SF is a land of opportunity or inclusiveness.

And the rest of California is just as bad. LA somehow has a reputation for inclusiveness despite being home to some of the most serious racial conflicts and explosions of racial violence in the last century.

That California is somehow radically more inclusive than the rest of the country is a myth, but one they are exploiting very well.