Thanks for the fantastic explanation!

Would it be more efficient to calculate some kind of per-model or per-layer mean, and then only specify standard deviations, maybe by fp8 or smaller?

> they never push back on nonsense questions or stupid requirements

I was reminded of your comment this morning when I asked ChatGPT how to create a path mask in Rhino Grasshopper:

Me: what is a path mask that will get 1;1;0;0;* and also anything lower (like 1;0;5;10 or 0;20;1;15} ?

ChatGpt: Short answer: No single path mask can do that. Here's why: (very long answer)

Me: are you sure I can't use greater than, less than in the masks?

ChatGpt: Yes — **I am absolutely sure:** **Grasshopper path masks do *NOT* support greater-than or less-than comparisons.** Official sources and detailed confirmation: (sources and stuff)

...so I think your priors may need to be updated, at least as far as "never". And I especially like that ChatGpt hit me with not just bold, not just italics, but bold italics on that NOT. Seems like a fairly assertive disagreement to me.

Well if you’re going to cite unnamed posts where unnamed people cited some LLM to argue against you, obviously you are 100% correct.

Fair, I actually misremembered and it wasn't my post they responded to with LLM slop but here is a HN user named "ranger_danger" in a thread with me in it, doing exactly what I described, LLM generated "citations" of US court cases that never really happened.

https://news.ycombinator.com/item?id=43509544

It's hard to know whether it's worse if "ranger_danger" did this on purpose and thought it's OK, or whether they didn't realise how the LLM works and thought this list was real.

While I agree that’s beyond the pale, I do not agree it necessarily means they’re wrong. It just means they are noise adding no value, and a lot of irritation.

As annoying as misinformed people and hallucinating LLMs are, it is a mistake to believe they are always the exact opposite of the truth.

People buy porn; spouses see credit card bills; people are performatively outraged and say it must be fraud.

Any actual data you can share?

My understanding from friends in the payment industry is that the “high fraud rates” are just people lying to their credit card companies when their spouse / whoever questions the charge.

If someone has the ability to process fraudulent credit card charges, why in the world would they waste that opportunity buying $20 of digital porn rather than a physical good that can be sold?

I believe GP that the issue is just pressure from evangelicals resulting in a de facto boycott of the woke porn virus.

I mean from the processor's POV it hardly matters _why_ there is high fraud, just that there _is_ high fraud.

Sure, and that's why rates are high. But the person I was responding to was asserting it in the context of "all porn is a massive criminal operation."

This isn’t true — LLMs can generalize and synthesize information not in the corpus. You can ask one to create a new written language and get a grammar and vocabulary that is nowhere in the corpus.

+1 — and it only saves characters when someone really knows what they’re doing. Most of is end up with `if (not result) or (result == “”)` and worse.

Are you saying that the problems of spam and impersonation are so insurmountable that there’s no point trying to mitigate them?

They definitely did not say that and what is this constant need for people on the internet to respond to someone saying "maybe this isn't the right way to do something" with "Oh well then you're saying that something can't be done at all and it's pointless and why even try!!!11"

Your absolute conviction is misplaced. Support is expensive to provide, especially on hardware that’s expensive to ship around.

This may be a bad move, and you’re certainly right that Synology expects to make more profit with this policy than without it, but it’s a more complex system than you understand. Irate customers calling support and review-bombing for their own mistakes are a real cost.

I don’t blame Synology for wanting to sell fewer units at higher prices to more professional customers. Hobbyists are man attractive market but, well, waves hands at the comments in this thread.

Synology is in that "prosumer" space, though, where maybe you don't really want to sell to hobbyists - but the bulk of your hobbyists in this space are tech-savvy and will recommend your products at work if you don't alienate them.

The thing is, that more professional market would never make the mistake of putting SMR drives in a RAID array anyway and they are also (I hope) good enough at doing their own research to filter out reviews from uneducated retail consumers. So, again, we’re left with trying to find a justification for this move other than Synology’s profits.

And when this issue happened with WD drives, I don’t remember a backlash against Synology at all. WD, on the other hand, deserved and received plenty of blame.

I know a photographer who needs tech support for really anything, and who has bought drives and upgraded is NAS himself. I don’t think that’s unusual, but of course n=1.