I would argue against “entire”. As an academic I (and I believe many of my colleagues also) take much pride in what we write - both the content and the prose itself.
If you’re saying that being able to legally bring cannabis into a country is a test for whether the IETF can host their meetings there… I don’t know if that is accurate.
Sure, Singapore has draconian laws when it comes to narcotics. But surely everyone attending will be aware of this? It’s been widely reported over the decades how foreign nationals have gotten life sentences or even the death penalty for drug running. What I’m saying is that Singapore are up front about it and it’s not enforced arbitrarily. Leaving your personal stash at home and abstaining for a few days should hopefully not be too difficult for the attendant engineers.
I am aware of that, but the fact that it was still on the books matters. For one, it has psychological impact, and for another if some police officer in a bad mood doesn't like your face geometry or number of thumbs, things like this can become 'power trip utilities' even if they're thrown out a few hours later.
Those are very fair points. I too am glad that the law was repealed.
I frequently get the impression that the policymakers in Singapore are more progressive than they reveal, but are extremely cautious about loosening up because they don't want to antagonise certain voter groups (e.g. people of certain religious persuasions). It is quite telling that the law was repealed in its entirety only in 2023, one year before the former Prime Minister Lee Hsien Loong stepped down from his position.
It's true that since 2007 charges have been laid by the police in a few cases, these have been challenged, and variously overturned or thrown out from court.
So, enforcement was certainly attempted and people were certainly detained for periods of time and forced to defend against charges that were laid.
It's not readily clear how often charges under Section 377 (1860-) or Section 377A (1938-) were laid in Singapore prior to 2007 (or of the charges laid how many cases came to trial and how many convictions occurred).
I think we’re getting into kludge territory there. Might as well have an explicit property that can be autocompleted/stumbled upon via spec or docs than an implicit behavior born of a little trick.
I find the distinction you draw between weights and a program interesting - partially the idea that one is a “static file” and the other isn’t.
What makes a file non-static (dynamic?) other than +x?
Both are instructions about how to perform a computation. Both require other software/hardware/microcode to run. In general, the stack is tall!
Even so, I do agree that “a bunch of matrices” feels different to “a bunch of instructions” - although arguably the former may be closer in architecture to the greatest computing machine we know (the brain) than the latter.
Arguably the distinction between a .guff file and a .guff file with a llama.cpp runner slapped in front of it is negligible. But it does raise an interesting point the article glosses over:
There is a lot happening between a model file sitting on a disk and serving it in an API with attached playground, billing, abuse handling, etc, handling the load of thousands or millions of users calling these incredibly demanding programs. A lot of clever software, good hardware, even down to acquiring buildings and dealing with the order backlog for backup diesel generators.
Improvements in that layer were a large part of what OpenAI to go from the relative obscurity of GPT3.5 to generating massive hype with a ChatGPT anyone could try at a whim. As a more recent example x.ai seems to be struggling with that layer a lot right now. Grok3 is pretty good, but has almost daily partial outages. The 1M context model is promised but never rolls out, instead on some days the served context size is even less than the usual 64k. And they haven't even started making it available on the API.
All of this will be easy when we reach the point where everyone can run powerful LLMs on their own device, but for now just having a 400B parameter model sitting on your hard drive doesn't get your business very far
Yeah, "static" may not be the correct term, and sure, everything is a file. Yet +x makes a big difference. You can't chmod a list of weights and have it "do" anything.
From what I can see, there’s reasonable competitors for CF’s offerings, but extremely limited parallels to their free tier. The free tier is the killer.
Exactly this. If I’m doing something big enough to pay for it, I would almost never choose Cloudflare. But as much as I dislike them, for my small projects there just isn’t an option better than their free tier.
I think the majority of contributors to that thread are being very reasonable and measured. CF should not be in a position to determine which UAs are viable on the Internet.
> CF should not be in a position to determine which UAs are viable on the Internet
Site operators explicitly subscribe to their DDoS protection service. They're not intercepting random traffic they don't terminate.
How do you propose we solve the bot problem? The Internet is a digital Mos Eisley cantina. The "come one, come all" approach isn't viable in 2025.
What if a server operator decided to block Pale Moon user agents? Is that his right? What about blocking bot user agents? How is that any different?
They are fighting a losing battle. Like running your own SMTP server in 2025 and expecting the big dogs (Google and Microsoft) to accept your mail and play by anyone's rules but their own.
reply