Funny, today I was forced to deal with a phone tree system that wanted to hang up on me at any whiff of a plausible path to hang up.

Point being that even without LLM magic, they have already been making it supremely hard to get things done. The old standby of hitting zero or saying representative over and over again would not budge this system. I thought for sure when I got it to prompt for payment information and if I screwed that up, *surely* that would escalate to a human, surely they want my money. Nope, hung up when I failed to provide the payment info in a timely fashion either.

If they went 150% yesterday, and another 150% today, then within a month they will have stockpiled all the RAM in the world.

I'm saying they may not be given the same offers that the suppliers were formerly giving them. If a supplier sees that nVidia will absolutely buy a huge supply of memory, then they will demand a comparable commitment from other customers. They will divert capacity to the customers that are willing to make the biggest and most certain commitments.

So Lenovo may have had to commit to bigger orders, or just be left out of getting enough to keep shipping their systems at all. If supply is constrained *someone's* orders are getting delayed, and the bigger orders get priority.

To the extent it might be a gamble, it could be a very short term gamble. Companies try to adhere to 'just in time' supply chain and carry very little advance supply, since investors heavily penalize carrying any sort of inventory over time. So as one example put it, maybe they extended stock from 30 days to 45 days, assuming that the memory market won't get better for at least a couple of months, which may be what the suppliers are forecasting.

The point is to explicitly extend android to be more 'laptop/desktopy'. So it's still a linux distribution, and looks like it won't be *really* distinct from Android....

They might not have had a choice. The memory vendors getting sweetheart deals from AI supply chain might require other markets to increase their commitment or get nothing.

So the choices might be either stockpile or not have any supply at all for their mainstream product. It's worth a risk of overpaying for memory when you have no other viable option.

Oh, for dialog it would suck. I'm thinking more about commanding 'sidekicks' to do certain things. Like voice command saying: "Bob, get up to that ledge (while pointing your crosshairs indicating the ledge) and provide cover with your sniper rifle". Today you can't direct non-human 'squad mates' with that level of specificity, so they do their specific scripted things or vaguely adjust their behavior in accordance to your vague command based on a press of a directional button. Natural language command of NPCs could open up possibilities to fix long-standing annoyances/limitations with NPCs trying to actively contribute to these situations.

For dialog, you lose the ability to be confident that the correct information has been conveyed to the player. So you can use it for background dialog for NPCs with no actionable info, but that dialog is going to be pretty pointless and particularly painful if it's hard to tell if an NPC is just background or has actual information for you.

Directing NPCs using natural language could enhance single player experience, where games have long sought to have NPC "sidekicks" and at best had to settle for very basic inputs in a real-time scenario "focus on my target, pick your target, form up, spread out" and even then it is generally making the input "too busy". These NPCs are a common source of frustration today, and if the gaming industry can't seem to give up on them, this could at least make them less infuriating... maybe...

Eventually? We are kind of already there. I recall some question on one of these going viral, attracting a lot of actual humans to write up why they felt the AIs struggled with it including answering in their writeups. So then their writeups made their way into the RAG inputs into LLMs and also into training material. The AIs suddenly got better at that question, what a surprise...

Just like most specific examples of LLM screwups get self-corrected in short order, automatically as the mocking ironically shapes the RAG component to avoid the specific behavior. Suddenly the LLMs got really good at counting the number of 'r's in strawberry, even as they couldn't actually count letters, but the internet now said how many rs were in strawberry just a whole bunch of times...

Exactly this, the "spec" is almost always a very rough draft that is largely written and consumed by people that want to feel like they contribute to the project even though they don't understand the customer or the developer situation that well. You might reference it a bit in your first offering and then ignore it as the stakeholder sees what the spec produces and realizes the spec wasn't really what they wanted when they see it live.

Once upon a time more weight was given to design, but the industry largely realized that all that very careful effort just became a liability of sunk cost fallacy when they realized the resultant output was not desired, but so much work had gone into the spec we don't want to change.

Nowadays it's a way that PMP minded folks feel like they are core technical contributors without learning to code. This is of course the target audience. Spoke to an executive that sincerely believes the only thing of irreplaceable value is his 'insight' and over 90% of his employees are going to be dismissed since he can just do it all himself. In practice his is the *first* job that could go to LLM, as all he ever says is either obvious stuff or just confidently wrong and his business decisions amount to "all we need is more customers and for them to pay more for it and we will be profitable"... Genius.

Fun fact, there has been two Linux distributions officially certified as "UNIX". Inspur and Huawei for whatever reason bothered to get them officially certified.

On the flip side, there's an odd sentence in the XDG specification that explicitly qualifies the wording around filesystem feature requirements to apply only to Unix-like platforms. Clearly they had Unix in mind, but they explicitly bothered to give an implicit pass to any hypothetical non-Unix, non-Unix-like platforms.

It can be a viable strategy to lean into bad news perhaps even more than is warranted. When you proclaim in 3-4 months time that you've overcome that disadvantage, people find that marginally more credible, even if that is wrong.

When you speak *exclusively* in CEO optimism speak, at some point people just stop believing anything positive you say.

If that CEO thinks the behaviors of the LLMs are "super smart", then I really wonder about his level of intelligence...

IT's certainly novel and different and can handle sorts of things that were formerly essentially out of reach of computers, but they are very much not "smart".

Processing that is dumb but with more human-like flexibility can certainly be useful, but don't expect people to be in awe of some super intelligence when they deal with something that seems to get basic things incorrect, asserts such incorrect things confidently, and doubles down on the same mistakes after being steered toward admitting the mistakes by interaction. I know, I also described how executives work too, but most of us aren't convinced that executives have human intelligence either.

Broadly speaking, a lot of the 'cloud native' stuff are complex solutions to potentially complex problems that fit within the parameters that those approaches can handle.

If you don't have those complex problems, then it's a premature optimization that is painful. If your use case is not the sort of use case resembling the bread and butter of the applications that instigated these approaches, then it's all pain, no gain.

There was a team that maintained a project that was broadly panned for not being good enough. The developers decided that the cure for what ails them would be changing to 'cloud native' approach, despite the complaints really being about limited functionality, not even about performance or scaling issues. Now on top of having the same functionality complaints *now* they have performance and reliability complaints too, and they have no idea what they are doing, they just arbitrarily carved their single fixed instance of software into a couple of dozen fixed instances of services (they can't figure out how to scale arbitrarily, so they still have exactly one instance of every component). Not one of them is capable of debugging the convoluted network situation they've created, and the logging information is just a mess.

You already had your turn in the late 2010s after big data and before AI. You don't get to come back to the hype line for seconds.

We need BYD in canada. Fuck american cars, unloyal american car companies and all their stupid fascist BS, government and culture. Produce BYD cars instead. I want my 9k-19k ev please. Would be finally worth ditching the beater for.