Or he is. Which is also scary.

I think that Berliner actually was way off in his timeline. NPR almost fully lost me back when Obama did the DACA executive order. Every story NPR did about it was some version of "listen to this sob story about a kid who isn't going to go to Harvard if DACA is struck down by the courts." Certainly those stories were real and heartbreaking in a way, but you never heard them run a story about the wife beaters and tax evaders who came here illegally with their parents as children (and of course I'm not saying those types of people are representative of the population either...just that they never even tried to be balanced). I thought their coverage was a great disservice to the country by not focusing on the legal merits but rather on the human interest. It was so blatantly biased that I see that as the point in my life where the liberal media bubble was popped for me. It (DACA) is probably one of the major starting points for Trump's political career, so it's ironic that Berliner sees their Trump coverage as the beginning of the end. NPR-level liberal bullshit is why he exists as a political player, IMO.

Nothing brings people out of the woodwork like self-reflection. At this point, I'm not sure it makes a lot of difference. The traffic has slowed to a crawl, so I'm not sure whatever choice you make is going to get a lot of support or pushback. It's basically a half dozen of us reverberating ideas into a large echo-y cave at this point.

In what sense? It’s up by like 30 something percent since its debut. I’m the first person to admit that I have 0 understanding of internet companies’ valuations.

I've flipped out on multiple family members, including my own wife, for posting pictures of my kids on whatever site (not for money or "influence", whateverthefuck that is, but just for sharing). I know it's de rigeuer in today's world, but I feel like we don't let kids get tattoos, because tattoos are forever. Well the internet is forever, too, right? It's really shitty to let a kid do that to themselves, let alone to make that choice for them.

Thanks for finally joining! It's slow moving around here, but there are good people with some really interesting and niche expertise.

Actually I don’t think that journalist captures the biggest flaw of all, which is the sample size. 20,000 people sounds kind of big, but it’s actually tiny. The problem is that the average age of the population was 49. Heart attacks among people below old age are relatively rare to begin with, and when you are sorting out people who self-report eating on a time-restricted schedule your numbers are going to be minuscule. “91%” is the relative risk, but I don’t see where they report the absolute risk or the absolute numbers. My guess is that a couple instances could be driving that seemingly large figure. Relative risk is meaningless in almost any context (of data reporting) without also understanding absolute risk.

I'm gonna push back on that idea a little. Firstly, this was an academic scandal, with an eye toward industry. But more to the point, I have worked in academia and private industry both fairly extensively, and I've found that private industry generally does more rigorous science (though often not as exciting). The caveat is that I work in biotech, so I don't know how that relates to physics. I would imagine it's not so different, though, because the incentive structures dictate everyone's behavior (but to be fair, biology experiments are notoriously opaque and hard to reproduce even when the hypothesis is rock solid, so there's a lot more room for obfuscation than in a harder experimental science).

In academia the financial incentives come from grants, which generally result from publications, which generally result from high impact discoveries. So the incentive boils down to "make high impact discovery."

In industry the incentive is to move product, and moving product doesn't happen if the product doesn't work. The product won't work if the science behind it is faulty. So the incentive is to weed out bad science, and only pursue the most reproducible work. This leads to a relative lack of risk taking, but generally more faith that what comes out of it is solid.

I can tell you from years of experience that the attitude in academia is "defend this at all costs" and in industry it's "kill this at all costs". Totally different mentalities. But again this is biotech. I realize fully that not all industries follow this trend, especially, say, venture-backed tech. I would imagine that the academic side of tech is way more upstanding than the industry side, but that's a hunch based on little-to-no first hand knowledge.

So here’s the problem they faced in a nut shell: They absolutely had no basis to countermand the fact finding of the lower courts. I.e., if the lower courts found it was an insurrection and Trump didn’t really even try to say he wasn’t. At least that wasn’t the main thrust of his argument. And they really didn’t want to entertain that bullshit about the president not being an officer. So where do they go? Clearly, Roberts and crew weren’t comfortable with where the libs were, which is “states don’t get to say, but someone does”. I goes they felt they had to invent something, and the something turned out to be section 5. Ok. It’s a stretch that you don’t have to be a legal scholar to see though, but the problem is that they were all just uncomfortable with the facts. They were too afraid to just apply the law, because the consequences were too much to face. So now we have a wack-ass precedent that could stand for 150 years. Insane.

Pretty devastating review of the ruling from David French.

The money quote is: "Through inaction alone, Congress can effectively erase part of the 14th Amendment." I've been following French's and some other writers' interpretation of section 3 in advance of this ruling, and while I've read pieces that have been persuasive in both directions, not a single one imagined this as the outcome. This may be a new low for this court, and they have a lot of hits on that chart.

No one in government is more deserving of impeachment that Thomas, not least because of his obvious perjury during his confirmation hearings. I just think no one wants to cross that Rubicon.

I'm in total agreement with Marcus on this one. The authors' misapplied statistics notwithstanding, I am a qualified expert on designing biologic assays, and I have used GPT-4 very successfully to help design a couple assays. My experience has been that it is a really good crutch for that sort of work, insofar as it can point you in the right direction. It can't point you to papers, but it can help you refine search terms which can then be brought to PubMed or another database as a refiner. All that said, the things that GPT has helped me with are things I could have done without it, but it sped the process up dramatically. It stands to reason that similar would be true for any arbitrary assay design.

As for the abused stats in the paper, my guess is that it was intentional, and that Marcus is giving them too much benefit of the doubt by saying it was probably someone who doesn't know what they're doing. If it were biologists I'd buy it. But these are math nerds who likely know very well what a Bonferroni correction is and how it's meant to be used. Bonferroni is one of the strictest post hoc correction methods. It's meant to weed out all but the most robust results. There are other less severe methods that are just as common, so the choice to use a strict methods has the air of deceitfulness. As in, it would make a lot of sense to use the strictest statistical methods if what you were doing was trying to prove a point and not just following the data.

Looking forward to an adventurous 2024. I have a day job, but what I'm really interested in is a drug discovery company that I started on the side. In 2023 I hired a postdoc, and we've been working tirelessly to optimize a drug screening assay. On the downside, I was recently told by NSF that they're not going to fund a grant I applied for. It was disappointing news but not devastating. On the upside, our assay is running really well, and we're generating data every day at this point, which is wild. Wasn't 100% sure we'd get there with all the trouble it put us through. But that's science. Solve problem 1. Solve problem 2. Keep going until problem 300 is solved and you're in business. Big shoutout to kleinbl00 for recommending just trawling Indeed for resumes, as that's where I found my postdoc and I would be exactly nowhere without her. Money will quickly become an issue later this year, but I'm hopeful I can convince some investors that what we're doing is worthwhile. Wish me luck. It's quixotic, but the payoff in terms of how many people I think I can help keeps me excited to do the work every day.

Not really worth a post of its own, but Bolton out here railing against a second Trump term is the best broken-clock-right-twice argument I've seen in a long time.

Polling has really wide margins of error at this early stage. I think people are likelier to say challenger over incumbent the farther from the vote you actually are. And in this case, it's probably more pronounced, because you know Uncle Joe is going to hammer J6 footage on TV ads until our eyes bleed. Mitt Romney had a lead in MI in early polls in 2012 and lost by like 9 points. Trump will fade. He may be more senile than Biden, and that will become more clear the more people have to actually watch him. He can't even keep Nikki Haley and Nancy Pelosi straight anymore. The Big Mac gunk is degrading his synapses by the day, and I don't think 50% of people are that stupid no matter how little they think of Biden. Call me an optimist. I've been wrong a lot.

An incumbent candidate whom the majority of his party views as the victim of a stolen election mustered 51% of primary votes in mutherfucking Iowa.

I think it's worse than that. There were like 110,000 voters out of 700,000 registered republicans in the state. So that 110,000 represents the most die hard republicans. Caucuses do not represent a valid statistical sample. So if anything, I think that his numbers are worse than the early indications. This is a very far cry from 2020 when they literally made their entire platform "MAGA, bitches!" The NH result was probably even worse, given that Haley won over 60% of independents. Biden may be unpopular generally, but there's no way those independents split more than 50% for Trump in the general. I think the pollsters say he needs 90+% of GOP voters and more than half of independents to have a realistic shot of winning. IA and NH seem to be really dim for the Trump campaign, despite all the catastrophizing from NYT today.

He's trying to buy back WeWork, so your dreams may. come true.

My 4 year old son asked me just this morning at what age kids usually get phones. I was like, um, a lot older than you. Kids want to be like older kids a lot more than they want to be like authority figures (even though all the authority figures have phone addictions, too, so no difference, I guess), so it's natural for them to emulate. I'm terrified of that day when we relent and get them phones (early teenager? Tween? Don't know exactly).

Here in the US you'll hear many parents say their kids need phones in school because of the school shooter phenomenon. Could have a flip, I suppose.

Hubski is the only social media (are we social media?) I use, so I'm not immersed in that world. Although I see the allure and have been sucked into Twitter vortices at times in the past, I see the whole endeavor as a cancer on society, and I hope eventually we will treat TikTok, FB and the rest just like we did PhilipMorris and RJ Reynolds. You make a product that is a giant net negative to humanity while making billions and defending it against all evidence, then you should be sued out of existence. Period. I do not care one whit about how much wealth would be destroyed by that action. Took 40 years in the case of Big Tobacco. Social media is destroying the world a lot faster, so I'm guessing less time here.

It's above my head, but at least they propose a practical (if not currently practicable) experiment, which is better than any string theorist has ever done.

My supposition is that there isn't a true association, and that a larger sample size would get closer to the true mean. I think this is less a case like you're describing and more a case of distorted findings based on a few outliers driving the average.

Had an inflection point in my side business this week, as we started phase 2 of our drug discovery program (and actually started getting some hits!). I'm itching to get away from my day job, but that wholly depends on some additional funding coming through. Have some irons in that fire, but nothing solid yet. The biotech funding situation is shaky right now anyway, so there's no harm in biding my time while I can. Either way it's exciting, and I'm getting closer and closer of realizing what seemed like a pipe dream until very recently.

Seems to be so many instances where these tech guys watch/read sci-fi and identify completely with the bad guys. Fury Road in this case, I guess.

Though I don't actually know, my guess is that you're on the young side of adulthood, no? The older you get, the easier it is to keep things in perspective. Like for example, when my parents were young, the n-word wasn't controversial and a lot of white people used it as the default to discuss blacks. By the time I was a kid in the 80s that was unthinkable. But, 'fag' was the de facto insult for everyone, everywhere, all the time. Transgender may have been invented as a term in academic circles back then, but it sure wasn't known to anyone in the public. The fact that conservatives are making anti-trans legislation is only possible because trans people exist in public these days in a way that they did not until very recently. It may not be perfect, but it's very difficult to not see "positive change" as being possible to hope for. Do not let short periods of noise obscure the very loud signal.

To you, but have you always voted for the winning candidate?

Lol. I turned 18 just in time to vote for Gore. That was a nice welcome into presidential politics. Been shitty every since. I remember being so angry, disappointed, and disillusioned by the result that I didn't want to get out of bed for like 3 days. But I was 18. Things affect you more at that age. But on the other hand, Bush turned out to be the worst president in American history hands down, and I include Trump in that assessment. So I suppose I was right to be down.

I do not think Trump has much of a shot in the rematch, but I also think his chances are non-zero. However, I do not share your pessimism that it will lead to the dissolution of the USA. We survived it once intact. We will again. None of this "he knows how to do it next time" carries water, because he doesn't give a fuck about policy. He cares about being in the news, and he'll never cede power to anyone who upstages him. So nothing will get done in Trump 2.0 either.

And I also do not know anyone who like Harris. I almost didn't vote Biden because he picked her. But my Trump hate was stronger than my conviction that I'd never vote for her. And I think that's true of a lot of other people, too.

Really looking forward to the first debate in presidential history in which not a single complete sentence is uttered by either candidate.

“This” is whatever hypothesis you’ve tied your success to.

Some of the issues I’ve discussed above maybe are unique to pharma, where I work. And in no way did I mean to imply that academia isn’t a good career populated with almost all good people. Just that the incentive structure, which can affect good people too, is such that it encourages putting one’s best data forward, say. This is obviously a ton harder in a hugely collaborative gazillion dollar field such as the one you work in…Lot of small time scientists work sort of on an island.

Also, the bigger the claim, the easier it is to poke holes in. The superconductor thing reminds of this thing that happened maybe 10 years ago where a researcher in my field claimed she could make induced stem cells by just bathing them in acid in some specific way. Took like 5 minutes to rip to pieces. Some people really want that paper in Nature.

For real don’t mind me though. I’m just kind of a jaded asshole because of some experiences I’ve had when my (very well connected and basically bulletproof) data didn’t align with the big lab’s big money hypothesis. They’ll skewer you.

Without getting into the technical details of the work I do, I have a certain amount of expertise in sex hormones, at least as they relate to the brain. One curious aspect of SSRIs that is not paid nearly enough attention to is that one of the ways in which they work is to increase a certain hormone called allopregnaolone, whose primary function in the brain is to make receptors for the neurotransmitter GABA much more sensitive. Curiously, Prozac, e.g., causes this change at about 10x lower of a dose than what the typical starting clinical dose is. However, the typical response to drugs not working is to increase dosage. My opinion is that people are generally already far overdosed and that docs should start way lower than they normally do. I, obviously, have no idea what your history is and I wouldn’t in a million years pretend to tell you what you should or shouldn’t do. However, what I will say, and what I say to anyone on your shoes, is that if you ever choose to do SSRIs again, start small and give it time. Your doctor will push back. Don’t let them. Happy to share more if you’re interested and also happy to butt out.

Already got scabs taking over

I'm always shocked by the laziness that goes into made up data. Like here you have multiples of a number. In bio you often see stretched or inverted images that are portrayed as individual replicates, say. You'd think that if you were going to go to the level of just plain cheating that you'd put your back into it. There are statistical tests, e.g., that can tell you with pretty good precision whether numbers are random, say, or whether a large group of numbers is spaced in a way you'd expect a natural set of data to be spaced (e.g., the frequency of small numbers increases with the size of the number). I doubt it would be that hard to fake a set of data if you reverse engineered it to pass the standard battery of smell-test statistics. But you never see that. Or maybe only the dumb ones get caught?

Real milestone yesterday, as my company had its first employee start. It's a big investment for me, but one I think will pay off. I feel good about it. She was already giving suggestions on improving the assays I'm running. New experience for me navigating all of the legal and administrative side of things...which agencies to register with, how to do payroll, where the hell to buy health insurance. If I can raise more money, I'm definitely making an administrative assistant my next hire!