Anecdotal but for some reason I had a pretty bad time with qwen3.5 locally for tool usage. I've been using GPT-OSS-120B successfully and switched to qwen so that I could process images as well (I'm using this for a discord chat bot).
Everything worked fine on GPT but Qwen as often as not preferred to pretend to call a tool and not actually call it. After much aggravation I wound up just setting my bot / llama swap to use gpt for chat and only load up qwen when someone posts an image and just process / respond to the image with qwen and pop back over to gpt when the next chat comes in.
I'm not worried about the danger of losing my job to an AI capable of performing it. I'm worried about the danger of losing my job because an executive wanted to be able to claim that AI has enhanced productivity to such a degree that they were able to eliminate redundancies with no regard for whether there was any truth to that statement or not.
They’ve been enshittifying for the better part of a decade. The model 3 launching without rain sensors and taking years to get any where near comparable with cameras comes to mind.
I made the move about a month ago to bazzite on my desktop with an nvidia graphics card. I still have my windows drive for when I need it but that's pretty rare. Bazzite isn't perfect but we've reached the point where the rough edges are less painful than the self sabotage microsoft has been inflicting on their users in recent versions of windows.
I moved to FSPTX a while back because it doesn't have TSLA. I'm not sure how long I'll stay there though, it has like a quarter of its holdings in NVDA now which has been great so far but it's going to hurt when the AI bubble pops.
It is wildly disingenuous to just copy paste that line from wikipedia and not the rest of the paragraph.
> In February 1998, Ericsson Radio Systems banned the in-house use of Erlang for new products, citing a preference for non-proprietary languages.[15] The ban caused Armstrong and others to make plans to leave Ericsson.[16] In March 1998 Ericsson announced the AXD301 switch,[8] containing over a million lines of Erlang and reported to achieve a high availability of nine "9"s.[17] In December 1998, the implementation of Erlang was open-sourced and most of the Erlang team resigned to form a new company, Bluetail AB.[8] Ericsson eventually relaxed the ban and re-hired Armstrong in 2004.
- edit, poster was quoting a quote in the article, not wikipedia, the article is the one omitting the context
If you find that you like something that most people don't it's generally best to just leave it alone and be happy you enjoyed it. If you try to delve into why so many people don't like the thing you like then either they will manage to convince you the thing you like is actually bad, which doesn't benefit you, or you'll start to question whether or not your taste in media is bad, which doesn't benefit you.
I think it's just different learning styles. My preference for learning e.g. a new programming language has always been to read a book cover to cover as the first step (if it's a language established enough to have a book anyway). Note that it's not important that I actually understand everything on this first pass. The cover to cover is mostly about getting the lay of the land so I know what exists, then, even years later when I have a problem that could be solved by using some bit of the language that I read about but didn't really understand but vaguely recall is a thing it springs to mind and I can do my deep dive on that aspect then.
I built a MUD in Elixir and I completely agree. After many iterations what I landed on as the best path was having rooms as actors. Mobs, characters, items etc are just data in the room process (actor) state (in memory that is, in the database they're all modeled more or less as you'd expect for a relational DB).
This gives me a large number of units of concurrency, while allowing the overwhelming majority of the code to be written sequentially since most of the complicated bits (combat etc) happen within a room so I don't have to think about concurrency at all for those. The only communication between processes is moving monster / character data from one room to another when those entities move around the map.
The ROG Ally seems like a really nice piece of hardware. Having said that, I cannot possibly stress enough how bad of a time you will have if it breaks and you have to deal with an RMA / ASUS customer support.
You'd probably save yourself a lot of time and frustration by just throwing it in the trash and buying a new one or a Steam Deck if something goes wrong with it.
D2:R was mostly Vicarious Visions (but it absolutely did turn out awesome).
WoW (retail) has had 3 of its last 4 expansions ultimately perceived as failures and is (justifiably, belatedly, finally) having its lunch eaten by the vastly superior Final Fantasy XIV.
Immensely is probably overstating Starcraft 2's popularity, but what popularity it still has is in spite of anything Blizzard has done for it recently rather than because of it. They've essentially abandoned the franchise to wither on the vine at this point.
This is the first acquisition, possibly ever, that I view as potentially a positive for the customers of the company being acquired, if only because Microsoft can't possibly mishandle Blizzard's IP and staff any worse than Activision and Blizzard already have.
Everything worked fine on GPT but Qwen as often as not preferred to pretend to call a tool and not actually call it. After much aggravation I wound up just setting my bot / llama swap to use gpt for chat and only load up qwen when someone posts an image and just process / respond to the image with qwen and pop back over to gpt when the next chat comes in.