The thing is that R1 is being compared to gpt4 or in some cases gpt4o. That model cost OpenAI something like $80M to train, so saying it has roughly equivalent performance for an order of magnitude less cost is not for nothing. DeepSeek also says the model is much cheaper to run for inferencing as well, though I can’t find any figures on that.
This series of articles and replies has really made me think about the structure of the Fediverse (as a casual user, though in the software biz), and for that I am very thankful. It makes me think, though, that just as open source developers got around the “free as in speech vs. free as in beer” issue by using the word libre, if the Fediverse needs another term – or even just call it “capital-F federation” – to distinguish the kind of first-class federation that Christine and ActivityPub represent vs. the definition that ATProto has suggested (and even vs what lots of regular software/service companies mean when they describe a system of microservices as federated).
Algorithm is just a fancy word for rules to sort by. “New” is an algorithm that says “sort by the timestamp of the submissions”. That one is pretty innocuous, I think. Likewise “Active” which just says “sort by the last time someone commented” (or whatever). “Hot” and “Scaled”, though, involve business logic – rules that don’t have one technically correct solution, but involve decisions and preferences made by people to accomplish a certain aim. Again in Lemmy’s case I don’t think either the “Hot” or “Scaled” algorithms should be too controversial – and if they are, you can review the source code, make comments or a PR for changes, or stand up your own Lemmy instance that does it the way you want to. For walled-garden SM sites like TikTok, Facebook and Twitter/X, though, we don’t know what the logic behind the algorithm says. We can speculate that it’s optimized to keep people using the service for longer, or encouraging them to come back more frequently, but for all intents and purposes those algorithms are black boxes and we have to assume that they’re working only for the benefits of the companies, and not the users.
ActivityPub is not a great fit for realtime chat. As others have noted, the Matrix and XMPP protocols are federated and designed for exactly this use case.
Algorithms can be useful - and at a certain scale they’re necessary. Just look at Lemmy - even as small as it is there’s already some utility in algorithms like “Active”, “Hot” and “Scaled”, and as the number of communities and instances grows they’ll be even more useful. The trouble starts when there are perverse incentives to drive users toward one type of content or another, which I think is one of the fediverse’s key strengths.
Drat, a missed opportunity!
Protip: Adding “: a study” to virtually any title instantly makes it sound more informed and scientific.
Check out Mullvad Browser, which is based on FF. It has always showed up as giving a non-unique fingerprint for me on the EFF site.
Yup. I linked both the original press release about melanoma as well as the one from October about the similar trial for lung cancer.
Bonus: the photo was taken with a film camera and the graininess is from all the radiation.
It might be old and slow, but I love RS-232. It works on every platform, you can write a client or server in just about any programming language in a handful of lines (and understand what they all do). I’ve literally made working RS-232 connections with paperclips and scotch tape. After the corpo wars when we’re all computing on salvaged tech you’ll come to appreciate it.
Thank you for your work! A native Jellyfin client is pretty much the only reason I still have a Roku.
Hurts So Good
The Lemmy Gentle Whisper
Mostly you either went without knowing stuff, or you had to go to the library. I knew a couple of wealthy folks who had their own sets of encyclopedia at home which could cost thousands of dollars.
During lockdown I was jogging my usual route and passed someone walking in full plague doctor getup. Thinking “that’s kinda odd” I turned the corner and almost ran into a lady rollerskating backward entirely in the nude. I live in Florida and see weird shit somewhat frequently but that particular run stands out in my mind.
Toothbrush. In one hand it scrubs food and gunk away and helps distribute fluoride toothpaste around. On the other it’s made of tiny plastic bristles that are probably disintegrating when in your mouth and growing a fun ecosystem when out of it.
Public speaking. I’ve seen surveys where more people are afraid of speaking in front of an audience than they are of dying, which is utterly insane. For the vast, vast majority of scenarios where you might find yourself speaking to a group of people, the risk level is very low. Likewise, in the vast majority of cases, few people are likely to remember much about your performance. It’s just talking.
I wonder if this matrix app was just a honeypot that was named to trick people into thinking they were using the “real” matrix.
The original gpt4 is just an LLM though, not multimodal, and the training cost for that is still estimated to be over 10x R1’s if you believe the numbers. I think where R 1 is compared to 4o is in so-called reasoning, where you can see the chain of though or internal prompt paths that the model uses to (expensively) produce an output.