yunohn

> We were aware of this mismatch before launch. We expected the engine to return an error for any prompt it could not process, and our long-context tests returned successful responses, so we did not believe this to be an issue.

This is insane? They thought it would return errors for long context, which didn’t happen, and they proceeded assuming it was fine? Didn’t even investigate the known misconfiguration and issue before going to production?

Works on mobile Safari for me. I’ve seen that scrolling issues are often caused by adblockers hiding cookie prompts, that need to be accepted to interact with the webpage.

Well, DS shows the thinking stream so it feels that way, but I’ve realized that OpenAI hiding it just gives a false impression - the non thinking output is also very wordy for OpenAI models.

I have the same phone. And I also use it just as a camera - it truly is that good! Gotta try it yourself to believe it maybe.

> if you enable them in about:config (privacy.userContext.enabled and privacy.userContext.ui.enabled)

This would be my definition of not available as a native user-accessible feature versus what they are launching now.

While I sympathize with the general negative outrage towards this change, I truly believe that people here fail to empathize with the mainstream users of Android phones.

I personally have seen every single older relative and non-tech friend, end up installing bloateare, spyware, and malware inadvertently - because they have no idea how anything in the tech domain works. And given the widespread popularity of Android (globally 70% vs iOS at 30% market share) and even moreso in lower income demographics, it also leads to rampant piracy of obviously non-essential apps like games and streaming (eg Spotify). In fact, even here on HN, almost everyone who has given their parents an iPhone has extolled the virtues of a secured AppStore/device and the peace of mind it brings.

While there may someday be a way to support both the average user and the HN power user, we are not there yet. It’s hard for me to outright reject Google/Android attempts to secure people’s devices.

> a product sheet showing what each models strengths an weaknesses are

This presumes that the labs themselves know how well their models perform. But all they have are overtuned benchmarks and hype vibes.

> PG's response can be boiled down to "nuh uh"

Worse, it's just a long post trying to show that doing math with a calculator somehow disproves real-life ethics.

> a side project so there’s not much to lose other than some of my weekly limit that resets soon

Basically the entire token-maxxing AI hype train in a nutshell. Lovely!

Maybe I missed something, but this article felt more like an ad for their modern matchbox designs, versus any sort of gallery of older ones - save for a collage near the end.

This has clearly worked out well with giving guns to the entire US population…

I believe you've misread - the Nvidia article and your quote support my point. Only by disabling the fp optimizations, are the authors are able to stop the inaccuracies.

I initially thought the same, but apparently with the inaccuracies inherent to floating-point arithmetic and various other such accuracy leakage, it’s not true!

https://arxiv.org/html/2408.04667v5

> kernel version powering Ubuntu 26.04 *LTS*

Feels like a lot of words to avoid thinking about “black” money and favors in kind. For example, nobody would include Trump’s golden bar from Switzerland in such ann estimate - repeated ad nauseam for all lobbying corruption.

Truly tired of seeing yet-another spam machine. All of these hype machines are built to spam people about their /paid/ hype product, rinse-repeat. BS

> Previous HN discussions

You say this, and yet there are no real comments i.e. discussion in either of them? This must be the HN equivalent of Stack Overflow's infamous "closed as duplicate".

Almost all companies issuing stock to employees also ban them and their family members and fellow house residents from trading in the same stock to avoid insider style improprieties and the SEC has frequently prosecuted such cases. Wild that congress and WH staff have zero such restrictions even in 2026!

Here’s the FTC’s definition of an ad:

> Any message designed to promote or sell a product, service, or brand, where there is a material connection between the speaker and the advertiser.

Yes, a discount is an ad - sometimes by the brand/manufacturer to get you to buy their product instead of a competitor, or by the seller to sell that product over others (for even mundane reasons like stock clearing).

Yes, sponsored content is an ad. The content creator is reimbursed for their output that is used to convince viewers to perform some purchase activity, usually over alternatives.

You’re really severely restricting the definition yourself by claiming an ad is “things that ublock origin” blocks. They can’t block physical banners and billboards or TV commercial breaks - does that now make them not ads? Whether you intended to buy something again doesn’t disqualify something from being an ad. In fact, that’s often when an ad is most effective - to buy the one they show you, instead of one you haven’t heard of or considered.

You seem to have defined ads as "obvious calls to action that end up in me buying it for sure". That's a pretty narrow view of marketing, but it does feel like you are aware that there may be other forms as you provide examples across the thread. It comes off as some form of elitism, where you deem the simplest ads as ineffective on yourself (but work on "average people") - but then go on to mention things like discounts and sponsorships, which to most are obvious marketing ploys too. No judgement, but maybe reflect on this?

> Yes, it does.

I find the way you operate - supporting and benefitting from Polymarket - to be equally disdainful from the same moralistic standpoint that gambling is banned from, but I guess even in orthodoxy one can bend the rules to their liking.

> Now you get basically the actual odds.

But that's the thing - insiders bet & trade at the very last minute, and thus are not supporting the /just cause/ of "information sharing", rather just plain, old front-running and racketeering. The odds you see when booking your flight are not the real odds - when the actual action happens just before event takes place.

https://ritchietorres.house.gov/posts/in-response-to-suspici...

> yeshiva

Doesn’t Orthodox Judaism (like all religions) look quite harshly upon all forms of gambling? How is Polymarket kosher?

To be clear, I didn’t question efficient market hypotheses - my stance has been pretty clear along the thread, questioning the value of the kind of information gambled upon in popular prediction markets.

Neither the blog nor Taalas' original post specify what speed to expect when using the SRAM in conjunction with the baked-in weights? To be taken seriously, that is really necessary to explain in detail, than a passing mention.

The world continues to evolve, in a way that requires flexibility - not more constraints. I just fail to see a future where we want less general purpose computers, and more hard-wired ones? Would be interesting to be proven wrong though!

> then yes.

I missed a link to any source for this claim?

> I'd also challenge you to outperform

I wasn't making a competition out of this - rather I'm questioning the fundamental basis of this.

IME llama et all require LoRA or fine-tuning to be usable. That's their real value vs closed source massive models, and their small size makes this possible, appealing, and doable on a recurring basis as things evolve. Again, rendering ASICs useless.

But the BTC mining algorithm has not and will not change. That’s the only reason ASICs atleast make a bit of sense for crypto.

AI being static weights is already challenged with the frequent model updates we already see - but may even be a relic once we find a new architecture.

> there is large value for some people in knowing when a country will be invaded

Are there any examples of people/companies trusting degenerate gamblers on prediction markets and making real life-changing decisions?

All the examples I’ve seen are exactly what I started in my original post - the insider circle opening a massive position on the right invasion date mere minutes/hours before they actually do it. This is useful to precisely nobody! And it happens because they are insiders, who want to avoid risk of exposure. Not to share their godly wisdom with the world for others benefit.

> it's clear prediction markets like Polymarket incentivize sharing information

I severely dislike these euphemisms used by prediction market enthusiasts. What exactly is the value of information like “most searched person on Google in year N”? Creating 10s of options to answer this question via gambling on Polymarket/Kalshi does not help anyone except their fellow degenerates. Heck, even events like “by N date the USA invade country X” also offer no real value, except for the insider circle to front run their own invasion and profit from it. Even worse, apparently they provide anonymity and cover to illegal participants (eg obviously US citizens) just like crypto exchanges like Binance did.

I truly question the sanity of those who believe that prediction markets are providing a positive force in this world.

What is the firm example that provides meaning to “claw”? I guess we don’t have any concrete analytics, but I would be willing to bet that the fraction of people who actually used openclaw is abysmally small, vs the hype. “Agent”s have been used by a disproportionately larger number of people. “Assistant” is also a great existing term (understood by everyone), that encompasses what the blogs hyping openclaw discussed using it for as well.