ProjectArcturis

No one else has the scale to even begin negotiating with the Pfizers of the world. Try to buy something from them directly, and the price would be "Fuck you".

Several years ago, I worked as a data scientist for Express Scripts, before it was acquired by Cigna. I can't much speak to the macroeconomics of PBMs, but I can say that they were the worst technical organization I ever worked for. They were built out of mergers on top of mergers on top of acquisitions, so their IT systems were what you get when you duct tape a dozen legacy systems together.

I worked in the "Innovation Lab", which had been designed to look like an ad agency's idea of Innovation -- brushed metal, Edison lightbulbs, that kind of thing. They'd bring clients through on tours to show off how much Innovation was going on. Meanwhile, I didn't really have that much to do, and no one seemed very concerned about that. Soon I realized I was also part of the decoration - a genuine Data Scientist, hard at work Innovating.

Our group produced approximately nothing. Our boss's boss was evaluated mostly based on how much he was able to sell people's medical data for.

The point of actively managed funds is not so much to "beat the market", it's to provide diversified returns via strategies that are uncorrelated with the market.

On average, the S&P500 has returned about 7% annually. If I had a strategy that returned 5% on average but was totally uncorrelated with the S&P, then you'd get the best overall long-term returns (maximize the geometric average of annual returns) by investing in a combination of my strategy and the S&P.

Ah, I didn't realize Litvinenko had been naturalized.

Russia has famously poisoned or otherwise assassinated countless dissidents and political opponents. But as far as I can tell, they have never openly killed a NATO citizen in a NATO country. That would be a substantial bridge to cross, and I expect there would need to be retribution for such an act.

We seem to have been able to build GPS without this breakthrough.

Very impressive work. Just in mice so far but this looks very promising for Huntington's in particular.

You'd need a source of chemical energy. Here it's hydrogen sulfide. Mars has very little volcanism, so it's unlikely these microbes would find the energy they need.

Adding more decimal points to various physical constants. Any everyday applications are far away.

Nothing seemed very important. I had no motivation to work on anything. I was also in grad school at the time, which certainly contributed.

I did this too. After I got through the withdrawal, I basically found myself sitting at my desk and waiting to die.

Whiskey was also one of the inputs.

Yeah, but if you're, say, Home Depot, or any other large non-tech company, your company culture will simply not be smart or flexible enough to innovate. The VPs of Data in those kinds of places are the true marks in the game. They have money to spend and no ability to determine real AI from bullshit. They also will be rewarded in the short term for buying something shiny but useless. Which if they're smart they'll leverage into a CTO position somewhere.

The very idea of a library would be mocked as a ridiculous utopian fantasy.

Seems like the hardest part would be getting rid of all the coins. I'm not surprised he got caught, given two appearances on America's Most Wanted, and his tendency to pay for hotel rooms with rolls of quarters.

As another comment noted, perhaps he should have opened a laundromat or video arcade to launder the stolen money.

Matt Levine is a treasure. He is tied only with Derek Lowe for being a fantastic writer who makes complex topics understandable, interesting, and enjoyable.

I mean, 99% of crypto transactions are either speculation or crime, but occasionally it can be used to pay for AI porn.

As a neuroscientist, I can say that calling it "neurotargeting" vastly, vastly overstates the contribution of "neuro" to these ads. Advertising has had more than a century of practical experience in manipulating the human mind. fMRI studies have added approximately zero to this understanding.

Really, fMRI studies are their own form of neurotargeting, though aimed at ad buyers rather than ad consumers. Pretty colorful pictures of the brain convey the idea "Gosh, we're so smart and advanced and differentiated from those other ad agencies", even if those pretty pictures provided basically no new insights into how to sell anything.

>And this isn’t just about the recent glut of capital raised, lagged returns and private equity exit blockages either.

Lagged returns is exactly the explanation here. They publish the table of investments and returns by year. Obviously you don't get your money back the same year you invest it. It's more often a ~7 year investment. But let's round down to 5 years.

Look at the chart and compare the investment in PE to the return 5 years later. It's a gain of 80-100% in that time.

This is simply how the math works when the investments into a strategy are growing exponentially, with a delayed return. It can continue indefinitely to produce 100% returns for investors, while a snapshot at any given time will make it look like the strategy has lost money overall.

What business are you in that predicting health data can make you millions?

Well, doing blinded studies will certainly be a challenge.

No surprise that US military planners have been planning for the wars we've actually been fighting, which are low-intensity counterinsurgency operations. The goal is to minimize US and civilian casualties in an environment where we have superiority in air, land, and sea.

And it makes sense to do that. If we ever have a shooting war against Russia or China, the nukes will start flying and a) many people will die, but also b) our existing weapons supplies and industry will be destroyed.

We're building weapons that are well suited for the environment they are meant to be used in.

Anecdotally, my mother was diagnosed with Stage 4 lung cancer (despite never smoking in her life). It had already metastasized to her brain by the time they caught it. I read the clinical literature, and she probably had about 3 months to live.

BUT: her cancer had a druggable mutation. She took Tarceva, then Tagrisso. She lived 7 years, before an aggressive lymphoma finally ended things.

The side effects were not pleasant. Her hair and skin got weird, and she had extensive digestive problems. But I am immensely grateful to the researchers who developed these drugs and allowed her to live long enough to see her first grandchild make it to kindergarten.

Sadly this doesn't seem to work anymore.

As I'm sure you know, there is a big difference between something a human can do when they pay attention to something, vs a completely automated procedure that can be done at scale.

>It should be trivial to do clique detection and downrank such citations.

How would you distinguish between a clique and a group of people who cite each other because they work on the same topic?

I've spent 15 years in academia and 10 in industry. Academia isn't worse on a per capita basis. Leaders' flaws are rawer, mainly because hardly anyone gets leadership training. And they often have more severe consequences, especially for grad students who are put in a subordinate, dependent position for an extended period of time. But IME you're at least equally likely to find psychopaths at the top of businesses.

No it's not. It's common to cite relevant papers from your collaborators (you are working on the same topic, after all). But it's an aberration to create a semi-formal system where you force your underlings to cite your papers in completely different fields.

Forcing your subordinates to cite 20 of your unrelated papers is clear and obvious fraud.