peterbonney

The total open interest of all major prediction markets is barely over a billion dollars. (1) That's the total amount of money currently wagered on ALL contracts across all platforms.

There are roughly 2,500 individual companies listed in the US alone with a market cap that exceeds 1 billion dollars. Tesla's market cap is 1.4 trillion - about 1,000 times the size of all prediction markets put together.

In other words, to the extent this incentive is of societal concern, it is not because of prediction markets - that is the tail wagging the dog.

Edit: to be clear I think these markets are a scourge as they currently operate, but not for this reason.

(1) https://predik.io/en/blog/open-interest-mercados-prediccion-...

I had the incredible good fortune to take one of his classes in college, and I loved it so much I took another just to learn from him again. A tremendous intellect AND an incredibly engaging and talented instructor. It would be an exaggeration to say that I knew him, but nevertheless he had a great impact on my education and my life. He will be missed.

I’m not saying it is definitely a hoax. But I am saying my prior is that this is much more likely to be in the vein of a hoax (ie operator driven, either by explicit or standing instruction) than it is to be the emergent behavior that would warrant giving it this kind of attention.

Correct, I haven’t set it up that way. That’s my point: I’d have to set it up to behave in this way, which is a conscious operator decision, not an emergent behavior of the bot.

Well I lol’d :)

Yes, this is the only plausible “the bot acted in its own” scenario: that it had some standing instructions awaiting the right trigger.

And yes, it’s worrisome in its own way, but not in any of the ways that all of this attention and engagement is suggesting.

Of course it’s capable.

But observing my own Openclaw bot’s interactions with GitHub, it is very clear to me that it would never take an action like this unless I told it to do so. And it would never use language like this unless unless I prompted it to do so, either explicitly for the task or in its config files or in prior interactions.

This is obviously human-driven. Either because the operator gave it specific instructions in this specific case, or acted as the bot, or has given it general standing instructions to respond in this way should such a situation arise.

Whatever the actual process, it’s almost certainly a human puppeteer using the capabilities of AI to create a viral moment. To conclude otherwise carries a heavy burden of proof.

This whole situation is almost certainly driven by a human puppeteer. There is absolutely no evidence to disprove the strong prior that a human posted (or directed the posting of) the blog post, possibly using AI to draft it but also likely adding human touches and/or going through multiple revisions to make it maximally dramatic.

This whole thing reeks of engineered virality driven by the person behind the bot behind the PR, and I really wish we would stop giving so much attention to the situation.

Edit: “Hoax” is the word I was reaching for but couldn’t find as I was writing. I fear we’re primed to fall hard for the wave of AI hoaxes we’re starting to see.

My understanding is that the existence of the nuclear triad is entirely about maximizing the likelihood of maintaining a second strike capability in the event of a preemptive nuclear attack, thus providing mutually assured destruction even if the first strike succeeds.

Dark stuff.

I like this piece. Doing hard things for the simple reason that they're hard is good for the psyche. I truly believe that.

The devil is really in the details on how the orders were executed in the backtest, slippage, etc. Instead of comparing to the S&P 500 I'd love to see it benchmarked against a range of active strategies, including common non-AI approaches (e.g. mean reversion, momentum, basic value focus, basic growth focus, etc.) and some simple predictive (non-generative) AI models. This would help shake out whether there is selection alpha coming out of the models, or whether there is execution alpha coming out of the backtest.

Having selected a 401k provider for a small (<15 person) company and also for a larger (>100 person) one, I can say that the big names make it prohibitively expensive for small companies to use them. And that expense ultimately comes out of peoples’ retirement funds in the form of fees. They frankly don’t want the business - too much compliance overhead for a small asset pool.

Believe me, I would prefer to have my own 401k at Fidelity too.

I have no dog in this fight, I just know from experience that setting up a 401k for your company is vastly different from setting up a brokerage account, and the reason a lot of small companies end up with off-the-run vendors is because those are the ones that will take the business.

"weird, overconfident interns" -> exactly the mental model I try to get people to use when thinking about LLM capabilities in ALL domains, not just coding.

A good intern is really valuable. An army of good interns is even more valuable. But interns are still interns, and you have to check their work. Carefully.

You're right, my view is way out of date. I didn't realize CBOE had grown so much in straight equities trading. IEX is the best of the rest, but it's NYSE, NASDAQ and (to my surprise) CBOE as the clear top 3.

I'd say IEX has done remarkably well - it's not likely to displace NASDAQ or NYSE but it has solidified its place as the #3 US exchange by any reasonable measure. If TXSE achieves comparable market share I'd call that a wild success.

You're not wrong to say that most participants don't care about what IEX offers, but enough do to make a meaningful dent in trading volume.

Exactly. Government entities have a funny habit of making their own decisions about what (and who) is and is not subject to their jurisdiction.

The more I learn about how AI companies trained their models, the more obvious it is that the rest of us are just suckers. We're out here assuming that laws matter, that we should never misrepresent or hide what we're doing for our work, that we should honor our own terms of use and the terms of use of other sites/products, that if we register for a website or piece of content we should always use our work email address so that the person or company on the other side of that exchange can make a reasonable decision about whether we can or should have access to it.

What we should have been doing all along is YOLO-ing everything. It's only illegal if you get caught. And if you get big enough before you get caught then the rules never have to apply to you anyway.

Suckers. All of us.

Thank you. The New York City subway has its issues, but most of them boil down to the fact that the system is really old and hard to modernize.

It also has unique strengths arising from the extreme population density and the inherent 24/7-ness of NYC (not to mention Manhattan's unique geography) but people don't talk about them as much as its flaws.

If your only reference point is transit in other US cities it's hard to grasp how different the NYC subway experience is, at least in Manhattan. Trains come every 5-10 minutes even at off-peak hours, and it's almost always busy. It's just not a very conducive environment for crime, unless you're riding in the middle of the night and/or at the tail ends of the system where density is lower.

When it comes to thinking about my own personal safety, I don't worry about crime on the subway, I worry about getting hit by a truck or e-bike rider.

That might be the case in other cities, but in NYC the socioeconomic dynamics are less clear. It’s mostly affluent-to-rich suburbanites that drive to work from outside the city, with rich and poor city residents primarily taking public transit (and to a lesser extent using taxis and car services). Almost no city residents - rich, poor or in between - drive to and from a 9-5 job in Manhattan.

Pricing is hard, no question about it.

One solution to the “not all users are equal” problem is to create different types of users and price accordingly. That’s what we’re doing, and it’s working out well so far. It depends on the product and use case, of course.

If you looked at how the average accountant spent their time before the arrival of the digital spreadsheet, you might have predicted that automated calculation would make the profession obsolete. But it didn't.

This time could be different, of course. But I'll need a lot more evidence before I start telling people to base their major life decisions on projected technological change.

That's before we even consider that only a very slim minority of the people who study math (or physics or statistics or biology or literature or...) go on to work in the field of math (or physics or statistics or biology or literature or...). AI could completely take over math research and still have next to impact on the value of the skills one acquires from studying math.

Or if you want to be more fatalistic about it: if AI is going to put everyone out of work then it doesn't really matter what you do now to prepare for it. Might as well follow your interests in the meantime.

Correct.

I used to have an orchard with c. 35 apple trees, 2 of which were honeycrisp. I can confirm that they are tricky trees to manage. I was theoretically in a good area for honeycrisps. But the trees were prone to all sorts of maladies that didn’t affect my other varietals, including antique varietals that are traditionally thought of as “difficult”. And when they did grow fruit it was usually small and misshapen.

Apples are interesting and this is a great example of the unexpected challenges you can face growing them. Every honeycrisp tree is a perfect clone of the very first one, but the environment of each is not a perfect clone of their original environment. And the interplay of genetics and growing conditions can have very unpredictable results.

Complicating this further is the fact that outcomes differ significantly across the U.S., with some leading hospitals able to keep twice as many 22-week-old babies alive as the national average, and occasionally able to keep babies born as early as 21 weeks alive.

The hard truth that many don’t want to face about prematurity is that the odds of survival at 24 or even 23 weeks are actually quite high, IF the baby is lucky enough to be born in the right facility. The odds at 22 weeks and even 21 weeks are not a lock but actually much better than you’d think.

We don’t actually need new science to radically improve prematurity outcomes. We just need to invest money in equipping and training more NICUs.

My son was born at 26 weeks, luckily in New York City where the standard of care is excellent (level 3 on a scale of 1-4) at even the second-tier NICUs, and where the highest possible standard of care is never more than a short ambulance ride away.

To put it bluntly: in NYC, a 26-weeker is 90% likely to survive to term. In some areas of the country a 26-weeker is 90% likely to die. The averages cited here flatten out this reality and make the problem seem more scientific and less social than it actually is.

Plausible result, but considering that this HN post is appearing on the same day as one of the top posts is about scientific fraud… let’s not get ahead of ourselves.

I say it all the time: provocative experimental results should be considered fraudulent until the raw data is shared, and then spurious until replicated.

My son was born _very_ premature, which gave me an intensive crash course on interacting with the medical system. I know two (mildly contradictory) things to be true about that experience: - My son would not be alive and healthy without the incredible care and expertise of his doctors and nurses, along with decades of scientific advancements on the treatment of prematurity. - Around the margins, his health outcomes would have been worse if my wife and I had simply trusted that same expertise without ever questioning decisions and pushing back on recommendations.

It’s so hard to have a nuanced conversation about the blind spots of our medical system. People want to force you into one of two buckets: you believe in medicine, or you believe in mysticism.

But there _are_ blind spots, because doctors are human and fallible, they aren’t generally experts in every subject that is useful for medical diagnosis (usually including statistics and often including genetics), and it isn’t just doctors themselves that are involved but a whole medical bureaucracy whose incentives don’t always align with optimal patient outcomes.

I can't trust the answers it provides or the text it generates. It's not a replacement for search, it simply makes search worse.

Can you trust search results without AI summarization? They’re mostly SEO spam.

The images it generates are, at best, a polished regression to the mean. If you want custom art, pay an artist.

Most visual art that people consume is polished regression to the mean. It’s designed for mass appeal, not originality.

I want to talk to a person, not a chatbot. The chatbot wastes time while you wait for a person that can actually help.

Have you spoken to an outsourced call center in the last, oh, 20 years?

I don't want music recommendations from something that can't appreciate or understand music. Human recommendations will always be better.

I’m old enough to remember hearing the same popular song 10 times a day on the local radio station because their rotation was designed for broad appeal to maximize ad sales, not to uncover new and interesting music for eager listeners.

I don't want AI mediating social interactions that it cannot and does not understand (though it may appear to). If I'm weary of too much volume on any social platform or in any news feed, I'll cut back on what I'm following.

This one I agree with, but it’s been our reality for at least a decade now, not a new development.

If you're having AI attend a meeting for you, it probably wasn't that important.

True

If you're having AI write your email, it probably wasn't that important.

Even in important emails there is a lot of “copy”, and AI-generated copy is no worse than human-generated copy. It’s just copy.

If it's screening job candidates for you, you're missing quality candidates.

If the alternative is keyword screening, I’ll take AI screening, thank you.

Fentanyl is frequently laced into other drugs (without the buyer knowing) because it is cheaper to produce than e.g. heroin. Or it is used as the main ingredient in fake versions of “real” painkillers like Oxy.

But it’s so deadly at low levels that it’s very easy to end up with a lethal amount in a single dose through inconsistent manufacturing processes, unbeknownst to seller or buyer.

Everything related to the surge in opioid addiction is obviously a societal problem. But not the one you’re imagining.

If “allowing the execution of arbitrary database queries written by an LLM inside a SaaS application” is the answer, I’d love to know what the question is.

Reminds me of this incredible New Yorker article about illegal mining in abandoned South African gold mines: https://www.newyorker.com/magazine/2023/02/27/the-dystopian-...