llamaLord

I'm pretty sure there was a weird bug in the original C&C generals where the game engine 'tick rate' was somehow tied to CPU cycles and assumed single-threading or something.

I remember going back and trying to play it many years after it came out, and it was basically broken because the entire game moved at like... 2-3x the maximum speed it was ever meant to run at

Sure there were some buzz words in there... But the actual core of the post wasn't patronising at all. Might need to take a step back and ask why your response to that was so strong.

My experience observing commercial LLM's since the release of GPT-4 is actually the opposite of this.

Sure, they've gotten much cheaper on a per-token basis, but that cost reduction has come with a non-trivial accuracy/reliability cost.

The problem is, tokens that are 10x cheaper are still useless if what they say is straight up wrong.

Actually I disagree, see previous comment re: the Fridman/zuch interview.

The genuinely cool thing about that though... They aren't sending the video.

Instead, they convert both people's faces into skeletal meshes and only send the transform data over the wire, recreating the other person at the other end as a 3D model that is being animated using the transmitted animation data.

The bandwidth requirements are reduced by like... 100x

You can turn off self-view by default.

The day I realised this, was the day 75% of my zoom fatigue disappeared.

Oh well that settles it then, because TWO medical professionals could never get something wrong, could they?

Thank you! This is EXACTLY the issue I've been failing to articulate.

Both examples were killed by the state... That's my point, the safeguards failed and the people were euthanized...

I get your point, but there are already MULTIPLE examples of people combining the mental-health component + some externality that is fundamentally a societal failing and coming to the conclusion that MAID is the optimal outcome.

"An expert committee reviewing euthanasia deaths in Canada’s most populous province has identified several cases in which patients asked [AND WERE APPROVED] to be killed in part for social reasons such as isolation and fears of homelessness, raising concerns over approvals for vulnerable people in the country’s assisted dying system."

-

"Another case focused on Ms B, a woman in her 50s suffering from multiple chemical sensitivity syndrome, who had a history of mental illness including suicidality and post-traumatic stress disorder. She was socially isolated and asked to die largely because she could not get proper housing, according to the report."

https://www.theguardian.com/world/2024/oct/17/canada-nonterm...

These are all excellent examples of how MAID should be used, I think the concerning thing is the widening of the eligibility criteria - especially when things like mental illness start to make you eligible.

Like... If your illness makes you suicidal... Is offering assisted suicide REALLY the best we can do? That starts to feel a lot like "eghh you're too hard for society to care about... We'll just let you die".

As horrible as this might sound, often it's failed suicide attempts that actually are the catalysts for people being able to get their lives back on track... What happens in a society where the government helps facilitate suicide and there's never any "attempts" anymore... Just successes...

> overly self-critical

IMO this relates heavily to Rejection Sensitive Dysphoria and ND people's propensity to maximise negative self-perception and minimise positive.

> still think of myself as a lazy cunt.

Unrelated, but "spot the Australian"

Maybe it's just me, but I've never been able to get a complex type schema to work properly with JSON schema.

The moment you have types referencing other types in a way that can become recursive in ANY way, the whole thing seems to explode.

Hate Jira all you want... but this is actually funny.

IMO a healthy relationship with LSD is one where, after a certain point (usually single-digit experiences), you can very confidently wish the substance a heartfelt goodbye, and thank it for it's service.

After my third experience, I woke up the next day knowing I'd learned an enormous amount about myself, and learned to think in ways I couldn't previously.

At the same time though, I just... "Knew"... I'd learned everything LSD had to teach me and that any further experiences with it would be like trying to over-optimise my brain, causing more harm than good.

That was ~4 years ago, and I've had literally zero desire to even consider touching the stuff since then.

I'm glad I did (3-4 times), but I'm VERY confident I'll never do it again.

Hey!!! We let you put a picture of your cat as the background for your board...

I find it fascinating how many things like this just completely gloss over the fact that some people have side-effects to these drugs that make them completely non-useable, and often it's some of the people who could most benefit.

I have ADHD, and have tried all the current generation of these drugs and absolutely CANNOT take them, they basically prevent my ADHD medication from working and I end up depressed and wanting to kill myself.

This idea of these drugs being a "cure-all" that we'll all just be taking one day proactively scares the shit out of me, because I worry it will mean that medical science will stall (as far as it relates to these types of conditions) for those of us who simply can't take these drugs.

There's probably a really interesting, bit potentially controversial topic to be discussed here.

As a Non-Indian, I'm quite aware/familiar with this kind of culture you speak of within Indian culture in the workplace and, to be blunt, if I were applying for a role as say a high level product manager (my current gig) and there was a native-born/raised Indian person as the hiring manager, I would at the very least be very cautious about this risk.

But in this situation, given the power dynamic is in favour of the other party, does this make me "biased"?... Or just "careful"?

I get your point, but using the best-sellers list as a proving point isn't exactly a slam-dunk.

What's that saying? "Nobody ever went broke overestimating the poor taste of the average person"

IMO we've had the EXACT same economic system since we were living in caves, it's called "supply and demand".

What do you do when "demand" for human labour drops to zero and "supply" stays at >8 billion.

No account of tinkering at the edges is going to fix that. We're in a much deeper fundamental problem than you might seem to think we are.

IMO it seems almost epistemologically impossible that LLM's following anything even resembling the current techniques will ever be able to comfortably out-perform humans at genuinely creative endeavours because they, almost by definition, cannot be "exceptional".

If you think about how an LLM works, it's effectively going "given a certain input, what is the statistically average output that I should provide, given my training corpus".

The thing is, humans are remarkably shit at understanding just how exception someone needs to be to be genuinely creative in a way that most humans would consider "artistic"... You're talking 1/1000 people AT best.

This creates a kind of devils bargain for LLMs where you have to start trading training set size for training set quality, because there's a remarkably small amount of genuinely GREAT quality content to feed this things.

I DO believe that the current field of LLM/LXM's will get much better at a lot of stuff, and my god anyone below the top 10-15% of their particular field is going to be in a LOT of trouble, but unless you can train models SOLELY on the input of exceptionally high performing people (which I fundamentally believe there is simply not enough content in existence to do), the models almost by definition will not be able to outperform those high performing people.

Will they be able to do the intellectual work of the average person? Yeah absolutely. Will they be able to do it probably 100/1000x faster than any human (no matter how exceptional)?... Yeah probably... But I don't believe they'll be able to do it better than the truly exceptional people.

Yeah but saying Jira is like Notion makes fundamentally no sense when Confluence is sitting like... Right there...

He's talking about the economic consequences.

AGI in an internet connected world is capitalism end-game. Once you have AGI, labour (both physical and intellectual) becomes redundant, humans have a "value to the system" approaching zero.

Our economic system is built on a series of assumptions that fundamentally cannot survive AGI, and nobody is really even trying to grapple with that fact.

This may have been a valid response ten years ago, but these days it's absolutely not.

The phrase "degrowth" is thrown around ALL the time in environmentalist circles these days, I've seen it printed on posters atleast three times in the last few months alone.

And climate activism has ABSOLUTELY grown into a subculture, that's why so many people now are uncomfortable openly associating with it, because it's been hijacked by radicals that demand you agree with them on ALL their socio-cultural ideas.

The amount of bugs in Azure that break genuinely critical parts of the workflow (like... We're talking, the ability to run a deployment on their K8's PaaS at all) which they seem to be totally fine with not fixing for long periods of time genuinely amazed me when we were trying to use them.

To be clear, I'm NOT advocating for premature abstraction, as a PM I live and die by "rule of three" when making decisions re: "let's just quickly solve this one bespoke need", vs "let's think about this more systematically".

But customers will never tell you all their requirements up front, and it's a PM's job when hearing a need from a customer to take the time to figure out if it's a single unique stand-alone problem, or simply one instance of an entire category, the other 3687 of which your customers are going to come asking for the moment they realise it's a category of problem your product is able to help with at all.

A lot of this comes down to poor product management.

A shitty PM sees problems like "when a guest at a hotel room completed {specific action X}, they want to receive a notification saying {specific message Y}".

A half decent PM will stop, step back, and ask "is there a meta-pattern here that I'm just not seeing yet", and realise the actual requirement is something like.

"When a guest at a hotel is the subject of any meaningful business event related to their booking, the management of the hotel want to be able to send them some kind of configurable, context-bound notification containing information about that event (ideally with some simple conditional logic to skip scenarios where the notification isn't relevant).

As an engineer, you're going to look at those two requirements fundamentally differently, and sure, the second one is maybe 3x harder to implement initially... But once you're 20x different "bespoke notification events" deep, you're gonna bloody wish you built the second option from day one.

Src: am Snr PM who sees teams get screwed by this type of bad PM thinking all the time.

You'd be surprised actually. Once you factor in network costs, running your own on AWS is only ~50% cheaper than paying for SaaS.

Factor in the opportunity cost you pay by having to have a person manage a self-hosted CI/CD cluster rather than outsourcing it, and SaaS CI/CD is around 3x better value than self-hosted.

Happy to share more detailed numbers if people are curious...

As the person who runs one of the largest cloud CI/CD services around, I can tell you exactly how they're making money... Actions.

GitHub would be making a VERY healthy margin on every minute of build time run on their platform, especially considering they don't have to pay a cloud platform for the compute like Gitlab or Bitbucket Pipelines do.