notnullorvoid

People pleasing is a component for some, some also like the technical challenge. The biggest factor though is that as a software engineer "No" is basically never an accepted answer (even from management with an engineering background). They'll either counter with "we need to do X, so find a way", or they'll move along the line to the next person until they get an answer they were looking for.

Edit: There are some rare competent managers who will stop the enquiry before even asking the engineers.

So their survival rests on the presumption that AI progress will fall between these two extremes.

That feels like a very generous framing. There's very little opportunity between the extremes that would paint a convincing outlook of survival for either company.

Perhaps if they were able to scale down their spending drastically they could survive, but that requires acknowledging their current valuations are BS. Doing so is a major risk, that will piss off all share holders. There's also the employees they would need to fire or reduce salary. The shift of focus internally to sustainability would be a major challenge.

I dislike their practices, but the main motivation for hoping they'll crash is that I think their immense overvaluation posses too much economic risk.

Always nice to see more open-weights in the heavy model class. I can only hope this trend continues, causing OpenAI and Anthropic to crash and burn.

I wouldn't be worried so much about those examples. One could take the open weights and fine tune them to either fix the poisoning or omission of obvious topics.

It's the subtle topics that we should be concerned about, and double so with closed models where even if oddities are identified they are harder to research further and impossible to fix.

I mostly agree with the prediction though maybe a bit more pessimistic about the timeline. Also I'm not sure our current usage of parameter count would make sense in this scenario, such a feat would require compressing current parameters in a manner much different then something producing a bit count per parameter. A hypothetical example would be using a single seed parameter per layer which then passed into a noise function produces the functional weights for that layer, able to reduce per weight size to sub bit levels (256 bit seed, producing 16K weights).

The Google Play store did (maybe still does?) allow installing some PWAs advertised as light versions of the apps.

On desktop or on mobile?

On mobile it's rare for apps other than web browsers to allow multiple tabs/windows and the ones that do (Ex: Gmail allows a draft to be a second window) have many multi window related bugs.

No defense for not writing or reading code if want to call yourself a software engineer or programmer.

However I do think there is reason to use Java or Python (as much as I loathe both) they have GC, and it'd be a lot easier for AI to fuck up memory safety in something like assembly or C.

Neither is mutually exclusive.

They get lock-in, and through that lock-in are more effectively able to inflate token usage.

If China is able to undercut Nvidia on high performance local AI hardware, they will pop the AI bubble in a matter of days.

They wouldn't even need to make something equivalent to the latest hardware. A Chinese RTX 3090 equivalent would be enough.

I think all 3 are pretty important for using TS in online sandboxes and IDEs.

Performance of tsc wasn't an issue for small projects, and for larger projects it could be fixed by using incremental build option, and/or TS project references. Most either didn't care enough about the perf or were too lazy to set it up. TS7's perf boost will give people less of a reason to use these options.

Boolean attributes are less about types and more about syntax convenience IMO, even HTML treats them as empty strings. <tag attr> is a shorthand for <tag attr="">, which then in your program you can decide if some attributes should handle empty strings as a truethy value. Some cases like <person name> the program would still make sense to treat it as an empty string.

If HTML 5 is indeed not even used directly, that makes the argument for the _stricter_ format like XML-based XHTML even more attractive

True however those who control the HTML spec are unlikely to let it go. Since XHTML has been treated as effectively deprecated some feature like incremental document streaming do not work in XHTML, this makes it unlikely for libraries that abstract away HTML to adopt XHTML as a target.

I'm going to choose to look at this in a positive light. In the long term this talent will feed into more indie games and studios, and perhaps studios will be less inclined to get acquired by Microsoft and other big orgs going forward.

Much of the gaming industry outside the indie space has stagnated, making sequels that don't offer much outside of slight increases to graphical fidelity and the odd thematic switch.

Meanwhile valid XHTML like having custom elements as table children, in HTML gets silently "corrected" to an malformed document.

I'd much rather have the stricter syntax parsing of XML, vs the complex and outdated structural correction rules of HTML.

The one good feature of HTML 5 was the introduction of boolean attributes. It's a feature XML could and should adopt.

The whole handling of custom elements was fumbled beyond belief. The HTML spec is a disaster particularly it's parsing rules the complexity of which is used as excuse by HTML spec authors not to improve the language.

XHTML was a better path.

I think the reason we don't see too many people complaining about HTML 5 is because not many web programmers use it directly, most are using JSX.

It bothers me that these popular UI toolkits render copious amounts of divs. I went to look at Base UI and for nearly every component it's "Renders a <div> element", when there are native elements that can achieve the same thing like details/summary for accordion.

I'm far from a native web evangelist, I think there's a lot the web APIs especially HTML gets wrong, but div overuse is also wrong.

I've yet to update to Android 17, but I ended up disabling haptics on 16 because the annoying pop sound feels like a gimmick compared to what haptic feedback used to be.

I run podman for local dev containers or for isolated sandboxes mostly. Anything needing orchestration and I go straight to k8s, never liked docker compose.

I hear this argument all the time, but I've always found it lacking connection with the human behaviour I observe as someone growing up low-middle class, now middle-upper class.

It completely misses the mark on human behaviour of those living in scarcity. Inflation forces them to save whatever they can in the most stable and liquid medium (cash). As a result it creates a very strong force pushing low income individuals further down, it takes a lot of hard work and luck to get out.

Those with enough wealth don't need the same liquidity or stability, they have the luxury to invest and see their wealth grow and outpace inflation. As a result of this security they are more willing to spend on products and services.

Inflation causes scarcity for the poor and security for the wealthy.

The lower inflation is the less scarcity for the poor, and they will be more willing to spend and invest. Even in a environment with 0 inflation the wealthy still have incentive not to hoard cash. The incencentive to invest was never about the devaluation of cash, but rather the outpacing return of value that investment brings. Theoretically that still exists even in a deflationary environment, though I do suspect high enough deflation would have drastic negative impacts on the market to the point where returns are too low to justify the risk.

loudly stating the foreign labs have been distilling their models for a while now.

They would be stating this even if it weren't true, because it fits their marketing.

While I don't disbelieve the claim outright, I highly suspect Anthropic is misleading everyone about the severity.

It's going to be a lot worse if these companies IPO before the crash.

The US gov should dissolve Anthropic and OpenAI tomorrow. Failing that they should block them from going public.

As much as I'd like to see Sam and Dario fall from power, I don't think they should be the ones going to prison for this. Investors gave them money full well knowing there would be no return until they could steal from retail investors when the companies IPO.

in a rational market.

Unfortunately the market is often not rational in this way.

Hype within retail market means there are suckers willing to buy. Institutional market knows there are suckers when the hype is high. Both would drive the price up, and retail investors the ones left when it falls.

Until there's some sort of "community owned hardware"

Or until some bright people figure out drastically more efficient means of training.

Banning next gen open models would never happen globally, and would be a major disadvantage to any country that does.

If the USA continues to put barriers into the release of models, open and/or foreign models will start to out perform them.

If open models are competitive enough nothing will stop even US companies from running them locally.

My experience has been that unless you are using the basic graphics options of Bedrock it performs worse at the same render distance, and even with basic graphics the perf is not much better. With Java version getting Vulkan rendering I suspect it'll out perform bedrock even without mods.

Not to mention bedrocks "improved" graphics look like trash compared to the shaders available for modded Java.

My biggest problem with dumb phones and similar concepts like this digital detox phone is the comparitively crappy cameras. I don't take photos often, but when I do I want them to be good.

The company having it's product market restricted will negatively impact their financial investment.

Also if you have an agreement with a company for them to provide you with a service, and investing in them is part of that deal, reneging on the service part still isn't okay.