I think the people that are saying that mean that the inference is massively profitable. From what I've seen, the margins are between 50 and 90% on an API call. It's the training that's very, very expensive!
As far as I understand, for both OpenAI and Anthropic, currently the majority of their GPUs are being used for training, with only a smaller portion being used for actual customer serving.
Inference requires training. That's a disingenuous point.
It's similar to people talking about drug companies charging hundreds of dollars for a medication that costs dollars to manufacture. Nevermind that it took them 15 years of research and trials to discover it and get it through trials and the approval process. Nevermind again all the drugs that failed at one of those steps.
I’ve not heard anyone say that the whole thing is profitable. Companies like Anthropic have said that inference is profitable, but I assume that’s only on a steady state basis, which nobody in the industry has ever reached at this point. As far as I know, everyone is still burning cash with data center buildouts and overall training costs.
There is almost always a next big thing. What happens when it ends is people move to the next thing. Practically everything needs some kind of construction, management, software, hardware, etc.
I'd be rich if I knew what the next big thing was but the economy has been running on the assumption that there will be a next big thing for many, many decades. Each cycle has the same rhymes of "what happens when it ends" questions but there has always something.
I don't disagree and I'm not a mastodon user but, as far as I know, if you local gov hosted an instance and your state gov did too, you could, essentially, join both and see them together?
RSS aggregation seems like the solution to me but I think we've got to admit defeat on that one because it's been around forever and people don't use it.
Pretty much every podcast in the world and all of their listeners are using RSS. WordPress powers something like 30% of the Web, and most of the operators behind those sites are publishing RSS whether they know it or not.
I thought I was more clear than I was being about rss in the context of the conversation about text based news.
How many people consuming text via rss the same way people consume audio via podcasts?
We're talking about text here. We all know how this sentence ends, "You should start a ..." and it isn't rss feed. Everyone used to have a blog and nobody really cared. Now everyone has a podcast and (more) people do. People don't know what an rss feed is.
> We all know how this sentence ends, "You should start a ..." and it isn't rss feed.
I don't know what this means. I'm guessing the correct answer is supposed to be "podcast"? Even so, weird false dichotomy.
That looking at tweets and listening to podcasts is more popular than reading long-form blog posts via RSS has a lot more to do with people's revealed preference for looking at tweets and listening to podcasts versus reading long-form blogs than it has to do with RSS.
(The actual context was delivery of notices by public services through non-Twitter feeds, anyway. Not blogging.)
> People don't know what an rss feed is.
I don't know why you think this is relevant here. It seems that when you're referring to using "RSS", what you mean is for people to care about RSS—in the way that Gemini, Secure Scuttlebutt, etc. have always been about caring about (and almost exclusively using them to talk about) Gemini, Secure Scuttlebutt, etc. than they ever were actually expected to be used organically and non-self-referentially.
The fact is that people (already/still) use RSS today. It hasn't been defeated. And Twitter could turn into an RSS-based feed reader tomorrow that works basically the same as it works today, and then suddenly everyone would be using RSS but no one would drop out because of it. Because people don't know (read: care) what an RSS feed is.
You can follow them both from your own mastodon account on a third server.
It's a technically bad solution though, because mastodon is push. The government server has to maintain a list of everyone who's subscribed and make an HTTP request to all of those random domains, and have some timeout and removal policy because sites die all the time. Pull-based, like RSS, is way cleaner.
We need to stop focusing on the specific protocol when it's not relevant and focus more on the patterns. Wanting RSS, specifically, made sense back when everyone had an RSS reader, the same way it makes sense to have an HTML website today because everyone has a web browser. But since very few people have RSS readers now, there's infinite flexibility in the format you use to provide your feed. The most useful is probably a phone app that vibrates your phone with a notification when your train is delayed. RSS can be provided as an option.
https://en.wikipedia.org/wiki/Consensus_decision-making alternative political movements care a lot about how decisions are taken. For instance, authors like David Graeber consider that majority voting only makes sense in the context if backed by violence.
So what percent of people are required for "consensus" because it seems like 50%+1 to 100%. There seems to also be an implication that democracy is 50+1 when it isn't. It can require any percentage to reach a decision.
Those two words refer to the same "type" of system where the members make a decision by ascent or descent. Both systems require an agreed upon barrier of acceptance.
They are the same. The US congress reaches decision by consensus.
Consensus is strictly 100% agreement (or at least 0% disagreement). Granted, it's not always possible. But in some setups and given some objectives, it's sometimes possible to have a collective that finds compromises that satisfy everyone (consensus).
IETF for instance, makes the distinction because they do not use a fixed vote threshold, but use a judgement of someone with good judgement, taking all facts into account such as the strength of disagreement.
And of course if that person turns out to have bad judgement they won't listen to him next time.
It is certainly possible for this to become unstable and split into two or more groups. This may work for the IETF which frequently creates and destroys working groups.
If the platform is filled with the same category of political speech, it directly flies in the face of the narrative that this type of political speech is being banned and that there's a chilling effect. You're taking a single anecdote that requires ignorance of all all surrounding evidence to prove a point that you want to prove.
You appear to have an overwhelming desire to see this obvious outlier as indicative of a larger pattern that clearly does not exist. A conclusion this article clearly alludes to and wants you to come to but doesn't do on it's on because it's so provably incorrect.
Just because this administration has done something in the past doesn't mean that every time that thing happens, this administration is responsible for it.
I didn't blame the administration at all. I don't even think this is specifically about political speech. You're ascribing a lot of beliefs to me that I haven't expressed, and assuming we have a similar experience of social media which flies in the face of everything we know about filter-bubbles and algorithmic content.
I think the parent's point - and what I more or less agree with - is that the problem is the hardware.
Human arms and hands are incredibly intricate. Reproducing their facility with hardware requires a large number of actuators and finicky fine parts. This isn't a software problem. Industry solves it with maintenance schedules.
There's probably nothing in your house that has as many moving parts as a robot needs. Your car maybe, and pretty much all it does is rotate wheels.
It's 175k per person including all benefits. Which is probably like 140-150k per year before the myriad of employer-paid payroll taxes and before the employer-side pay for insurance plans.
