HN user
Omitted the Betteridge's law part of the headline, since we all know the answer.
This sounds like only a drop in the bucket.
But nevertheless it is a hopeful sign that Congress is at least able to agree on the the right thing to put in the right bucket.
Perhaps we should provide feedback to encourage pouring more in...
Why should a retail investor never buy derivatives? spreads?
Correct archive link: https://archive.is/2026.07.11-040454/https://www.economist.c...
Oops. Here is a good one: https://archive.is/2026.07.11-040454/https://www.economist.c...
archive: https://archive.is/mai5w
Impressed by the progress. Also interesting how much stronger depression is as a risk factor, vs. all others.
This is why men don't ask for directions :P
Sister shared this with me. Actually looks cool.
They try to verify you are taking a live picture, so you get real cats.
more info: https://cheezburger.com/45806853/pokmon-go-but-for-cats-free...
archive: https://archive.is/zx8Ba
Here's the full research article, from the researcher's website: https://hu-my.sharepoint.com/personal/rdiamond_fas_harvard_e...
Single women’s marriage/cohabitation rates rise by 29 percentage points and employment among baseline nonemployed women rises 27 percentage points after six or more quarters.
These effect sizes are huge. That and the drama of the findings provokes general skepticism. So I welcome some experienced folks picking this apart.
SIgh, just seeing the comments here. I get emails for replies to my comments, but not to postings, so never saw these.
The link was one provided by the researcher, but HN stripped out al the params that allowed it to be viewed. I need to remember it does that...
or... HN could be improved a tad, ;-P
I found another article that covers this without a paywall, and will share that.
full article via preprint: https://osf.io/preprints/psyarxiv/7fjpr_v2
essentially negativity progresses since: >it is easier to differentiate oneself through negative comments than through positive comments
I find this interesting in two respects:
1) I agree with the finding about social media
2) In my experience this effect is broadly true beyond the internet and before it:
Journalism skews negative because it is easier to capture your attention with the bad vs. the good (in part due to human loss aversion).
This is my experience in the domain of art as well. Most high-status books and stories deal with negative themes and deeply flawed characters. There is a basic assumption that somehow the negative has more artistic value than the positive, and perhaps the positive is necessarily bland and uniform.
e.g. "All happy families are alike; each unhappy family is unhappy in its own way."
In my experience this is wrong. It is however true that its easier to come up with specific negative art.
apparently this is captured in a principle taken from the quote above: https://en.wikipedia.org/wiki/Anna_Karenina_principle
Positive states have lower entropy: there are fewer of them. There are many more ways to fail. Thus it's easier to come up with novel negative things vs. positive.
I think therefore the negative slant to our art represents a culture of laziness in our institutions. Part of this is the result of selection: unhappy artists succeed more easily, rise to levels of authority, and assume that anyone who does not share their negative view and aesthetics are bland.
It's sad though. I think they believe all happy families are alike because they have never truly understood even one happy person.
The implications are big: a society led by such sad people becomes a self fulfilling prophecy.
OK, these effect sizes are huge. That and the drama of the findings provokes general skepticism. So I welcome some experienced folks picking this apart.
Single women’s marriage/cohabitation rates rise by 29 percentage points and employment among baseline nonemployed women rises 27 percentage points after six or more quarters.
Gift link: https://www.nytimes.com/2026/06/24/science/giant-fruit-fly-s...
This seminal discovery is nuts. And interesting way beyond the puns.
Each sperm is almost as long as the fly itself. They seem to function collectively like an organ.
Even after reading this I don't understand at all how they avoid tangles:
sperm don’t become tangled because they’re constantly moving, sliding past one another
Yeah, I've tried this with my charging cords and it doesn't work.
Microbiology truly is alien nanotechnology.
We found converging evidence that AI's advantage stemmed from rapidly deploying larger quantities of information
This jibes with previous results in LLMs effectiveness at combatting disinformation: https://mitsloan.mit.edu/ideas-made-to-matter/mit-study-ai-c...
in a series of four preregistered experiments
Love seeing preregistered research!
archive https://archive.is/K5s2A#selection-1457.0-1457.12
OK, it's temping for me to make this comment about Vinod and perceptions of his past actions. But he's not important in the larger picture, and this issue definitely is.
Overall, we absolutely need to devote a lot of energy and discussion to this question now, since there are a lot of options, and building consensus and implementing anything smoothly will be slow. If we wait too long, we tempt disaster due to factors like unrest or missing the window of control. For example, the world is getting populist leaders now because we failed to address the inequality of growth in the last 30 years. So the more intelligent, powerful people raise this issue now, the better.
I'm not sure about each of these options yet. Definitely the change to capital gains makes sense (cough, OK I must gesture here at the carried interest rule for a moment...).
Sovereign wealth funds scare me. A fundamental approach where the government owns and directs capital sets up the wrong incentives. Over time, it risks privileging the elite, or at best the leviathan, over the median individual.
Rather, my gut is to pursue a broader distribution of wealth & power, e.g. by:
1) reducing the barriers to capital value, e.g. through promoting open source software, and eroding software IP
2) creating a systemic forcing function reducing inequality. The ideal one would be progressive wealth tax. Of course this is incredibly difficult in practice.
3) creating friction agains large corporations. this would need to be a totally new approach to antitrust. beyond eroding IP, I'm not sure of the best way to do this generally, even in ideal terms (perhaps progressive taxes on capital value appreciation?)
paper: Spontaneous problem-solving in bumble bees https://www.science.org/doi/10.1126/science.ady1618
Taking action to even the distribution of wealth & power anywhere in the world is a good idea. this has always been true. The Machine Learning Revolution is just making it more urgently true.
That said it's a bad idea to have the state:
- pick the winners (who will just end up being the ones who lobby the most)
- pick the industry (what even counts as "AI"? machine learning is applicable to to every domain)
archive: https://archive.is/ainRf
Having done my share of pipetting back in the day, I have no problem letting my ML overlords take that part of the job.
for good or ill, overall biology is so complex and hard to experiment with, ML progress will be bottlenecked by other factors at a level of productivity much lower relative to information technology. But it's still be a huge deal.
gift link: https://www.nytimes.com/2026/06/04/opinion/voting-gerrymande...
Every system has its tradeoffs. But I am sold on the idea that our current one is a major contributor to the degradation of our institutions. The options here would likely be better.
Sure it's hard to make any changes to our entrenched system. But at least these changes are non-partisan. Of course since it hurts the current system both parties will cooperate to defeat change.
Thus I call: "moderates of the US, unite!"
:P
Sure, I get that API's are an essential thing.
Nevertheless, I feel uncomfortable when we are entering an era when the software we design is more for use by machines than by humans.
Your phone can use tiny skin-color changes to measure your heart rate, measuring your face when you unlock it. good to within 5 beats per minute. it also estimates its confidence of each measurement, and throws them away when too low. impressive, and a bit scary what you can measure with neural nets.
good discussion in the podcast: https://www.nature.com/articles/d41586-026-01793-1
paper: https://doi.org/10.15184/aqy.2026.10359
also: >the material record offers indirect evidence for distributed authority. Indus seals, small stone stamps that likely facilitated exchange and credit, were found primarily in private residences at Mohenjo-daro rather than in temples or central administrative buildings.
Speculative, of course. But cool data & approach. And it doesn't have to prove anything, except that it's plausible there are other ways to structure societies, that can have different results.
Tremendous progress in a year.
While these foundation models aren't trying to be calculators, this kind of test previously provided a decent benchmark on their ability to scale composing iterative reasoning steps, and showed they were not that good at it.
At this point I'm tempted to conclude they are pretty good at it, since I don't see how such long calculations could really be considered "in distribution" from training or "memorized," except in the sense the model learned the algorithm correctly.
I still have doubts about how good present the present architecture & training is at learning to "generalize" effectively. e.g. see ARC3
But you can go a very long way, by memorizing everything, being able to compose steps well, being able to try many times, and being able verify as well as a human, even if you aren't so efficient in your "fluid intelligence."
The fraction of human cognition operating today that can be handled with that current approach seems pretty large.
paper: https://doi.org/10.15184/aqy.2026.10359
also: >the material record offers indirect evidence for distributed authority. Indus seals, small stone stamps that likely facilitated exchange and credit, were found primarily in private residences at Mohenjo-daro rather than in temples or central administrative buildings.
Speculative, of course. But cool data & approach. And it doesn't have to prove anything, except that it's plausible there are other ways to structure societies, that can have different results.
archive https://archive.is/9iUDH
what I found most interesting here is that it fleshed out for me the future where dedicated devices are designed to do everything on your behalf, in ways that are fundamentally non-transparent, and and thus hobble user agency. The analogy to me is consuming all your social media via an opaque algorithm determined by a large company, vs. selecting sources or feeds on your own.
and of course US companies would love to achieve this as much as the Chinese.
When your main device is an earbud and it advises you on who to vote for, you will have no open browser to track down its sources to humans (who would have written using the same device anyhow). So how are you going to effectively question it?
Interesting points in here.
e.g. that Godel didn't think this scrapped Hilbert's project totally:
Gödel believed that it was possible to redefine what we mean by a formal mathematical framework, or allow for alternative frameworks. He often discussed an infinite sequence of acceptable logical systems, each more powerful than the last. Every well-formulated mathematical question might be answerable within one of them.