derangedHorse

Ironically I thought of the notification systems of social media when I think of encalming. Like seeing the blue dot above the notification bell on X and not having to engage with it immediately.

I think we’ve been able to integrate these concepts well into modern software but maybe we need to tackle new problems towards working towards products that encalm. For example, establishing general design rules to make the center have a clearer distinction between things we care to action, and things we don’t. Maybe something small decision models can help with.

> Was the problem proven by using a collection of everyone's work, or was it due to a brand-new insight?

Loaded question. A "brand-new insight" is still built off the work of others. A possibly better way to frame it would be in how many subjectively unintuitive logical leaps have been made from prior work.

> AI will eclipse humans in every possible domain, leaving everyone without direct control over the most powerful models completely powerless, economically redundant, and at the mercy of the new AI hegemony

I disagree. I believe access to AI will shift the power balances of the world, but I wouldn't say people will be left completely powerless. But maybe our definition of powerless differs.

What is your definition of powerless? Unable to influence? Physically outmatched?

um...what is going on here? bots? lol

Adapting to the times, nicely done!

I like both. From an economic perspective I'm a little sad my programming skills give me less of an edge in the job market, but I still hand program here and there for fun.

> If you genuinely, genuinely, believed artificial superintelligence was coming, why would you want to work at the company that would be in the best position to replace its employees?

Just because the job becomes obsolete doesn't mean the employee has to. Some employees have adapted to different jobs and roles many times in their careers.

> I think I could have been a pretty formidable startup founder had I been born five or ten years earlier

I used to think this around 10 years ago. I eventually learned this kind of thinking is defeatist and figured if others are able to succeed in this environment then I can.

> there are only "AI startups", where you build a company that somebody else can write a prompt for in 30 seconds.

Just like how people think ideas are a dime a dozen, software will be. The value will be in polish, ongoing maintenance, feature development, and the incorporation of customer feedback.

> I can't think of a single piece of software I use that has gotten better

Users are notoriously bad at noticing small improvements over time. A more interesting thing way to see the impact of AI is to ask employees at big companies what it has allowed them to work on and release in a previously unimaginable time frame.

> infinite growth in MMR

Dating apps are a race to the bottom. They're collecting a tax on the lonely and desperate of dwindling populations. Ironically it's not sustainable over the long term.

I think it's unlikely for your worst case scenario to be realized. Underspecified objectives can cause unexpected results, but it can also lead to desired results. Fortunately experimentation is relatively cheap and tuning optimization metrics is the norm for most apps.

This is false [1].

[1] https://www.usaspending.gov/explorer/budget_function

> You are talking about someone who gets zero matches at all

No, I'm not.

> People are far more willing to pay to upgrade a good experience than they are to pay to try to improve a bad one

This is an assumption and one I believe is false. My belief is that people are more risk-averse rather than they are risk-taking [1]. People don't care to upgrade when they're winning.

> Further, an unappealing profile that gets zero likes will continue to get zero likes after they upgrade to a paid subscription.

Not matching with someone does not mean they have 0 likes. The platform just isn't incentivized to show you who likes you for free. Concretely we can take Tinder as an example. They will show you that someone likes you even when they won't show you who. Swiping enough to find the person who likes you usually takes countless hours.

> This user is far more likely to get frustrated and uninstall after a few minutes than if they were to download the app and be shown profiles that are appealing

If you're extending this to anything past the first profiles shown, this is simply untrue. You just need to dangle a carrot in the form of "someone likes you," even when the person liking you is a bot account.

[1] https://en.wikipedia.org/wiki/Prospect_theory

> they would rather be single than pair with any of the options that are realistically available to them

I don't know if the matching problem has been fully explored enough to ascertain this is the case.

In the western world, people will go out with someone they find undesirable with the intention of reaping the benefits of short term relationships. I think this sometimes sets unrealistic expectations on those undesirable people and further muddies the water. I wonder if there's a way to impose a cost on those who are not in it for the long term.

The sentiment of the linked post is different than the one you describe. The post is explaining why it's ok to have a non-zero value for a metric seen as objectively bad (like fraud) if there's high confidence that the non-zero value can be traded for a disproportionately larger metric somewhere beneficial (user acquisition).

In this case the undesirable metric is divorce. If the parent comment's feature for measuring marriage lengths and estimating likelihood of divorce were implemented, what disproportionate benefit would allowing more matches with a nonzero likelihood of divorce bring? The app wouldn't be enforcing whether divorces happen, it would just use the estimation of it happening as a matchmaking metric.

That's before they were acquired by Match Group.

> but in order to avoid serving you the best matches for a long-term relationship, they'd have to know what those are

No? With modern dating apps they can just serve you people who have already swiped left on you. Even if they made it a random presentation of candidates, they will never be incentivized to matchmake for free. If you're not paying then you'll get shown people in the area and that's it. That's all they need for their ad-revenue.

> given the social cost of divorce

I think they're more concerned about their fertility rate and labor force. A 0.87 total fertility rate for 2025 [1] is pretty bad.

[1] https://www.abc.net.au/news/2026-03-12/singapore-low-fertili...

> Unless it's properly filtered, returned water will be contaminated by material in the cooling system.

Any wastewater is regulated and typically doesn't need to be "filtered." For regular operations, water is treated but the evaporative process leaves most treatment chemicals behind.

> the gross margins right now are very positive. [...] So the company loses money. Each model makes money, but the company loses money.

Amodei mentioned the reason for the big losses in an interview released on Feb 2026[1]. Assuming he was telling the truth, that's not a bad position to be in. From what I can tell from the replies to simonw, a lot of people are cynical about Anthropic's growth potential. I'd like to encourage the cynics to bet against Anthropic wish them luck. I will be doing the opposite.

[1] https://www.dwarkesh.com/p/dario-amodei-2

I think everyone having the same standard of living any time soon is an unrealistic expectation. I expect standards to be sustainably raised, not for people to have all their problems solved.

I believe he’s advocating for mathematicians to understand the novel concepts introduced with each breakthrough rather than individuals understanding data stemming from empirical observations.

> Everybody else is then "economically redundant", and the whole system collapses because people aren't going to sit around and starve because there are no longer jobs to be had

This is a non sequitur. People being economically redundant doesn’t necessitate a system collapse. It assumes our existing systems can’t handle an increase (assuming we have economically redundant people today) in those who are economically redundant.

Capitalism defines who owns the means of production scoped to the definition of ownership in a particular governance structure. In the U.S., for example, full ownership gives people the right to produce as much as they want with rules in place to seize the rights of ownership if broken.

One rule of ownership is that taxes must be levied on the positive return of capital. This today is what helps funds the welfare state. The idea that the system collapses implies we won’t be able to scale this welfare state up which seems unlikely to me.

There’s already a deficit caused by welfare spending. An increase in abundance (of work, resources, etc.) can reasonably alleviate this if we consider the possibilities in removing inefficiencies within the government itself and their biggest cost drivers like public health.

The distribution of goods will probably be maintained until others lower the gap over time.

Using the metric of “4 earths of resources” is confusing and easily misunderstood. What that metric means is that if everyone were to have the same consumption as the average U.S. person tomorrow, we wouldn’t have enough renewable resources to meet demand.

This is not an insurmountable problem. It just requires increasing production everywhere else. Your cited metric isn’t a stat on the theoretical limit of global production, it’s one that describes its current limits.

The stat also factors in carbon emissions, but this is also easily addressable in a world where labor and maintenance is cheap.

> "Who's gonna pay me, and for what ?". That's the question I want answered.

Unfortunately that’s a question you’ll have to answer for yourself. Also, people are going to risk going along whatever path gives them work as many today already do.

If the “hack” referenced by the latest announcement from Australia is the same described in this article, I’d hardly call it a hack. It seems the agent was tasked with obtaining data and reasonably guessed query parameters in an attempt to do so.

When it was unable to, it used cross site scripting as a way to check the capabilities accessible through the browser making the requests. In this case cross site scripting wouldn’t be a hack against the Australian website, it would be a hack against the urlquery site, if one could even call it that.

Finally, downloading public files from the public pre-production server also seems like a non-issue.

The sql injection attempts against the other sites are less ambiguous. Attempting to access non-public user passwords rather than reasonably tweaking the parameters for a site designed to serve public data are categorically different things.

As someone who uses both, Astra was 100% the better model. I have yet to give 5.5 a spin so maybe that’ll be the new top contender.

Intuition happens naturally and people are prone to inducting the wrong conclusions. Proofs provide a framework for rigorously analyzing drawn conclusions such that it can be used to build intuition in others. If math is about sharing the insights gained in a particular class of problems, proofs are the means to getting there.

Tail emissions and infinite divisibility are proposals to address this.

This would be a new publication with OpenAI's novel solution. Buckmaster and Alpöge did not arrive at the NS solution, they used an approach that also happened to be used in OpenAI's proof.

> there’s absolutely no reason to do it if you believe you independently arrived at the result using only public prior work

Unless the goal is to assuage the existential pain felt by many mathematicians with respect to what may feel like increasingly inconsequential efforts. It also seemed like a way to build good will amongst knowledge workers dealing with similar issues, with the goal of showing they are dedicated to easing the transition for them as society parades into the future. This has clearly backfired.

> The leaving out collaborator part is outright academic malpractice.

They would not be leaving a collaborator out. Again, this would be a new publication. Here's the quote from Buckmaster's post [1]:

"Two proposals were offered to me. The first was that we post our Euler result, and that OpenAI post its Navier-Stokes result the next day. The second was that, after posting Euler, I alone write a paper presenting the Navier-Stokes result, acknowledging that an internal OpenAI model had resolved it."

[1] https://cims.nyu.edu/~tristanb/statement.pdf

> AI maths makes no sense, like AI art makes no sense, because those are things that people enjoy and can do pretty damn well ourselves so there's no point to automate them away.

There are people who appreciate and have use for math and art, but don't or can't produce. Those same people are now given tools to produce works that they enjoy and can do pretty damn well at.

> Those are brave men. Let's go kill them!

This is a horrible perversion of the quote. Aside from it's questionable use in this context (as it can be construed as a violent call to action), Tyrion made the battle cry for self-defense against a threat to their lives. Generally, AI reducing the dependence of one human on another's work is incomparable to this.

> Only after learning the secret to cracking the problem did they send the first prompt.

Which quote in the announcement post provides evidence for the above quote?