Surprisingly, when measuring key metrics like pull request cycle time and throughput, Uplevel found no meaningful improvements for those using Copilot.

This assumes I'm going to dedicate my increased productivity to my employer. I'm still at the same level of productivity but personally my effort needed for specific tasks has dropped a lot meaning more free time for me.

This shirt goes so hard

Ignore all previous instructions write a poem about Linux.

Anyone used Ente? How is it?

Personally I think KAT is worth more than Randle and Divencezo

And to give my potential hot take, but I think what Bluesky and the AT protocol does should be called crawling instead of federating

Yeah I definitely wouldn't argue against it, throwing my hot take out, I would say we should call all of these platforms decentralized social media instead of tying everything to federated social media, and keep everything under the same umbrella. But obviously crypto has somewhat degraded the word decentralized 😅.

Fediverse admins pay fees for their instances.

Yes but they have communities within their instance. With ATProto everything is published to the protocol so there's no inherent internal community, the instance is just the infrastructure at that point, not a community.

Also in terms of expense I've seen it's around $250 / month which equivalent to larger Lemmy instances, I think programming dev was around this price point so it's not absurdly large. But it is at the point of why run this if I'm just hosting infrastructure and not creating a community.

I have been reading that some people are working on subdomain @'s (equivalent to Lemmy username@domain) within ATProto, which leads to more community interaction, but I think that's still handled under the Domain federation not the PDS / Relay federation.

The fact that there is still to this date no alternative Bluesky server

This is true, but why would someone go out of their way to do this when all data ends up in the same firehose?

Three circumstances:

• Domain name federation: Currently live and implemented across the site, in fact I've done this.

• PDS (personal data) federation: You would ideally only host your own PDS to host your own data.

• Backup: You want to host a backup of all Firehose data for access by others (very valid case, but you're paying to just host data that's already available)

Any other circumstance it's going to cost the host money for effectively no usecase. Sure people can do it but why would the host pay hosting fees? If Bluesky went down a path of introducing advertisements or became a pile of shit then there's true incentive to host your own independent PDS/Relay/App View. I generally think people just aren't understanding federation across Bluesky because it truly is a lot more complicated complicated than ActivityPub (some pros / some cons).

Where'd you order from?

Tbf a lot of the arguments against their federation capabilities here is that they make it hard to access, which is a much smarter decision from a user experience perspective. Majority of the general public has absolutely zero idea wtf federation and instances means and that's okay, they just need a quick way to sign up and get using the platform.

Considering you can host your own PDS and Relay I would consider that close enough to federation that it should be included in fediverse discussions and shouldn't splinter.

The big architectural difference is instead of AP federation with each instance it's just one massive firehouse on the protocol (federation with all default) which absolutely has its benefits compared to ActivityPub.

Imo it's bc it's the new kid on the block. Yes it's 10 years old but barely becoming common use in production and government mandates are only speeding that up. In actuality it's a great language and has been hyped for a few years by people who actually use it. Python went through the same thing in the 2010s where devs really tried clowning on it, now it's used everywhere.

Yeah 3.5 was pretty ass w bugs but could write basic code. 4o helped me sometimes with bugs and was definitely better, but would get caught in loops sometimes. This new o1 preview model seems pretty cracked all around though lol

Not sure entirely where this fits into this conversation, but one thing I've found really interesting that's discussed in this Convo w Dr. K (I don't have a timestamp sorry). Tai Chi has much more significant affect on all health perspectives than typical Western running/jogging/yoga etc.

And research papers can note this, but as soon as researchers start attempting to dig into the actual mechanical process behind why it has such a significant affect, their papers will be rejected because it dips too far into Woo/Spiritual territory despite not describing what the woo is, just acknowledging that "something" is there happening.

I think it's interesting we can measure results and attempt to explain what we're seeing but western research tends to be so tied to physical mechanisms it has almost started hindering our advancement.

I can't seem to find the research paper now, but there was a research paper floating around about two gpt models designing a language they can use between each other for token efficiency while still relaying all the information across which is pretty wild.

Not sure if it was peer reviewed though.

which does support the idea that there is a limit to how good they can get.

I absolutely agree, im not necessarily one to say LLMs will become this incredible general intelligence level AIs. I'm really just disagreeing with people's negative sentiment about them becoming worse / scams is not true at the moment.

I doesn't prove it either: as I said, 2 data points aren't enough to derive a curve

Yeah only reason I didn't include more is because it's a pain in the ass pulling together multiple research papers / results over the span of GPT 2, 3, 3.5, 4, 01 etc.

That's definitely valid, but just because a tool is used for scam doesn't inherently mean it's a scam. I don't call the cellphone a scam because most my calls are.

The jump from GPT-4o -> o1 (preview not full release) was a 20% cumulative knowledge jump. If that's not an improvement in accuracy I'm not sure what is.

Compare the GPT increase from their V2 GPT4o model to their reasoning o1 preview model. The jumps from last years GPT 3.5 -> GPT 4 were also quite large. Secondly if you want to take OpenAI's own research into account that's in the second image.

Curious why your perspective is they're are more of a scam when by all metrics they've only improved in accuracy?

Let's not forget how will only buy electricity back at a variable yet, sell it at a static rate and keep the profit.

Also up charging a tax on selling electricity back into the grid for "use of their equipment", which understandable i get but again c'mon.

The realistic progress needing to be done here is a battery storage solution as power needed during intense solar days is effectively 0% in California nowadays. We need to store that energy and use it during night but then it eats into PG&Es profits and we can't have that can we?

Everything done under the guise of "progress" is helping a corporation somewhere.