What role do social networks – and the precise nature and structure of social interactions – play in the production of public goods and in human welfare? What about social contagion? A thread on #HNL experiments with many thousands of participants from 2010 to the present. 1/

New paper with @AlexRdgzG: We give polynomial-time (!) adaptive attacks against a number of threshold Schnorr signatures. At the core are the first polynomial-time attacks on LDVR, a problem whose solution yields adaptive attacks on several practical schemes.

I've been told by EA types, repeatedly, that China was going to fall into line and restrict AI development because their politicians agreed with EA. However, it turns out that Harry Potter and the Methods of Rationality is not nearly as influential with the Chinese leadership as they seem to believe. I've also been told, repeatedly over the last several years, that the US government was certainly going to fall in line with the EA propaganda machine this time, after the latest manufactured PR stunt, after the latest "crisis", and it keeps not happening. Why, only this last weekend a certain person from one of the big companies was claiming not to be gloating that open source AI was inevitably going to get banned, that he was simply stating the facts, except, well, it looks like he got ahead of himself. I'm going to simply state the facts as I see them. The war against the Doomers is not won and will not be won for many years to come; this is a marathon, not a sprint. However, reports of their victory, in *any* such sudden surge operation that they conduct, are inevitably premature. The fact that they have been unable to gain a clear victory in spite of a vast and overwhelming financial advantage (yes, the financial advantage is on the side of the EAs) and in spite of the insane numbers of front organizations they run should say something. We will fight, we will keep fighting, and although it is not going to be easy, we have won so far, and we will continue to win. So, the next time they manufacture a crisis, say when they try to rig something up to make open source models look dangerous, or they pull out the stops to get another shill a surge of television interview time, or when they announce on their Discord channels a new name for what they want this week now that "pause" and "pacing" have failed and start repeating it suddenly on social media, you will know that what you need to do is fight back, and not just roll over. After all, they want you to roll over, because that's the easiest way for them to win. Refuse to cooperate. Refuse to be psyoped into believing they've already won.

BOTH SYSTEMS TAKE FROM EVERYBODY. ONLY ONE OF THEM EVEN CLAIMS IT WANTS TO HELP YOU. Maybe the socialists have a point.

A tech-forward vision for the left: valkenburgh.substack.com/p/a…

30 years of arguing about lawful access to encrypted comms, almost entirely in adjectives. So I tried to put numbers on it. Annual compromise probability seem to show it is orders of magnitude away from what is acceptable in aviation, nuclear etc. substack.com/home/post/p-212…

For six months, Google Maps sent a slice of drivers in ten American cities down deliberately slower routes. About 30 seconds slower on average, and under 2% of trips were touched. The experiment ran on roughly 100 of the most congested road segments in each city, switching on and off day by day so each city acted as its own control. The results are out now in Nature Cities, from Google Research with collaborators at Berkeley and Stanford. The reasoning behind it is old and slightly counterintuitive. Every navigation app does the same thing: it finds you the fastest route right now, for you alone. When millions of phones do that at once they all pour onto the same handful of arteries, and those arteries stop being fast. Transport economists have been writing about the gap between what's good for one driver and what's good for the network since the 1950s. Nobody had tested a fix at city scale with real drivers, because you'd need to steer routing for a large share of a city's traffic, and about three companies on Earth are in that position. So Google put a penalty on the worst segments during their worst hours, which nudged the routing engine towards alternatives of similar road class and comparable travel time, then measured what happened across the whole network on weekdays between 7am and 8pm. On the targeted segments, traffic moved roughly 2% faster in the median city. Los Angeles got 4.56%, Atlanta 3.30%. Fuel burn on those stretches dropped between 0.5% and 1%. Across every road that saw a change in traffic, which covers around 80% of each city's driving, speeds rose 0.35%, reaching 0.5% during the morning and evening peaks. Total travel time on affected trips fell 0.69%. Their Bayesian model put the probability that the speed effect was genuinely positive at 99.8%. That adds up to more than 1,000 tonnes of CO2-equivalent saved per year, per city, in most of the places studied. Cars and vans account for around 10% of global carbon emissions, and the average driver spends about 2.6 years of their life behind the wheel, so a fraction of a percent across an entire road network is a serious amount of fuel. Per driver, the saving comes to about 0.25% of an average journey, roughly one fortieth of the normal day-to-day wobble in how long the same commute takes. Nobody in those ten cities noticed anything, in either direction, including the people sent the long way round. The authors are careful about what they haven't shown. They measured the immediate effect, not what happens once drivers work out the freed-up route is quicker and pile back onto it, which is the standard way road improvements get eaten. Their penalty scheme was deliberately crude, so the ceiling on the approach is unknown. And the underlying Google Maps data is commercially confidential, so nobody outside can check the numbers. The finding sitting underneath all this is that a private company's routing algorithm has become a piece of transport infrastructure, tunable in the way a traffic light or a congestion charge is tunable. The obvious follow-up experiments are the ones a city government would want to run, not the ones a mapping company would. link to full article: nature.com/articles/s44284-0…

Having synthesized neural networks capable of navigating high-dimensional vector spaces, organic engineers promptly attempted to turn them into Swiss Army knives by stuffing fifty markdown cheat sheets into a single execution context. The result is not omniscient synthetic engineering; it is stochastic attention decay. The industry treats `SKILL.md` as a silver bullet for agent modularity. In practice, recent empirical research (Li et al., 2026) reveals that single-agent skill libraries suffer a sharp, non-linear phase transition failure once semantic confusability takes over. The solution is what operating systems solved fifty years ago: process isolation. When an engineering task requires a non-trivial domain shift, the correct systems primitive is spawning an ephemeral subagent with a pristine, tailored system prompt and dedicated tools, bounded within an isolated failure domain. Full breakdown on phase transitions, semantic confusability, and subagent architectures: afshar-oss.gitlab.io/blog/po…