Looking at 5 tok/s after reading this comment made me think about why it felt slow and would be unacceptable for work. If you didn't plan or even sometimes despite planning, you have absolutely no idea if it is suddenly going to go off the rails in a wrong direction. Everyday, I'll look at the thinking and it seems pretty good until suddenly I have to slam the esc key because it decided to pursue a completely wrong direction. Much faster is better for skimming to make sure you don't have to throw everything away.
and if it goes fast enough, it can fail and you can prompt again before you need to worry about if it's going in the wrong direction. or, it will try multiple directions!
A lot of the cultural changes that were achieved are actively being fought against and are slowly being reversed. Seems like every few weeks there is something from the civil rights era being chipped away at.
This stance seems weird to me. Do you really want to be locked into paying subscriptions for everything and have to rent all your software?
You got burned by a crappy company and we desperately need laws about being made whole when a company shuts down it's servers, but demanding everything be a subscription seems like a bad approach in the mean time.
Jellyfin continues to be a noticeable downgrade from Plex, but it is chugging along to being as good. If you're already a Plex lifetime subscriber the only logical move is to make sure your library is setup in a way that is compatible with both. There is no reason yet to switch away from Plex, unless you've had that terrible UI forced on you on Roku. I can switch at any moment when ever some MBA decides to enshittify something important.
I have no idea who will ever buy a lifetime pass at 10x what I paid for Plex in 2019. I struggled with the decision to pay $75 back then. There were effectively zero competitors to their product then.
Having Claude calculate which beers are the best deal at the bar based on price to alcohol from a picture of the menu is currently a massive party trick.
Outside of programmers, almost no one has actually seen AI be useful for anything except do a barely acceptable job at a task they could have done better if they felt like it.
Not all programmers with AI mandates have seen this yet either.
"If they felt like it" is key here: in my experience, AI makes it very easy to not feel like anything and just trust whatever the AI comes up with. The barrier to diving deep into code, or researching any other topic by myself, has become higher.
At the same time the barrier to getting some results has become much lower, especially for complex topics I knew nothing about. So that seems great, but I keep running into cases where I do check sources and the AI turns out to have summarised it incorrectly.
It seems like that issue where LLMs are just flat wrong isn't going away, at least as long as they're trained on human data. Humans of course are fallible but it can be a cascading problem that isn't clear to anyone if all we're doing is looking at the solution and letting the LLMs check the work.
I remember being excited about LLMs initially because it seemed like a natural evolution of the information saturated world we live in, we now need help deciphering the signals from the noise. But that was assuming that most/all the noise was from mis/disinformation. Now it just looks like another tool of control for the megarich to take even more from the rest of us. Maybe there is some hope in open source models.
Why use Claude Code over something like Opencode then? From my limited usage of the tools over the past couple of months Claude Codes ergonomics feel strictly worse than Opencode, but I haven't deeply investigated either yet. I am using Claude models in both so I am getting a one to one comparison.
Can this be backed up with any numbers, especially in the US? Every company I've seen using an AI something has obviously been using the API of one of the bigger companies. If this is a valid approach with proof it's basically as good, it would be something I would recommend to my company
For most use cases, you don't actually need frontier performance either. Customization, cost, and data sovereignty are far bigger practical concerns. If you can run your own model on prem and tune it exactly what you need, then you're both saving money and getting better quality output.
It's also wroth noting that tooling can go a long way to improve the quality of output from the models as well, and this is very much an under explored area right now. For example, ATLAS agentic harness does a clever trick where it gets the model to generate multiple candidates then uses a second lightweight model as a heuristic to score them keeping the promising ones. And this drastically improves coding capability.
There's also a paper along similar lines discussing how using a harness to force a project structure also allows it to work on much larger projects successfully.
So, I don't think that raw power of the model is even the most important part at this point. We can squeeze a lot more juice out of smaller models we can run locally by using them more effectively.
We're basically in the mainframe era of this tech, but the pendulum always swings to tech getting more optimized and moving to edge devices over time. And I think we're already starting to see this happen with local models becoming good enough to do real work.
I assumed you were asking about capabilities since that is a question that can be answered. There isn't any comprehensive reporting by companies, so it's just the anecdotal reports the video I linked discusses.
Some managers are fine screwing power users when they feel they are big enough. I will never buy a Chamberlain garage door opener for their similar stance against the Home Assistant community
Same! I was burned by Chamberlain, twice. In my case I had bought a $40 retrofit device (it connected to the MyQ cloud and simply used RF to emulate a remote to open any brand of opener).
First, there was their debacle where they broke HA connectivity just for fun, meaning I couldn't use HA or Apple HomeKit anymore. Then, after a pretty routine reset of the opener (I needed to clear out some old remotes and re-learn them) I found that in some recent update of the 'app' or whatever, they'd deleted my brand of opener from their supported list, due to some IP dispute of some kind, leaving it unable to learn the same remote it had learned the year before. So, as peeved as I had been to have to use their ad-laden app, the myQ device itself was completely useless to me.
Never again.
Irony is I just moved to a house with a brand new MyQ cloud-connected door. I bought a RatGDO anyway and will never buy any of the devices in the myQ ecosystem, even though some look attractive. Closed system on purpose = never buy.
Though I'm not OP, I will say that it seems like there are two brands that mostly have the market cornered.
Chamberlain/Liftmaster/MyQ is all the same company; they are a gross company that hates the idea of giving you control over your device. Zero LAN control story, Zero Homekit story, zero Home Assistant and no possibility of any of these.
Genie - whose "app" thing is called Aladdin Connect is the other one. There is a HA integration[1] for it, though it's cloud-dependent, no LAN story so again your ability to control it is subject the company's cloud servers being available, and to any future whim they may have. The Github for the plugin has issues reported, but no idea how widespread they are.
Looking at places like Home Depot it seems there's a brand called SkyLink[2] but it seems cheap in the bad way, and while it has its own "app" there seems to be no HA story whatsoever, so I assume kinda the worst of all.
Deeply uncomfortably, I would have to grudgingly acknowledge the practicality of buying from the gross Chamberlain, never using its MyQ BS, and connecting a RatGDO to it instead, which would give the best experience, even though giving them any business deeply offends me.
Please read my comment again - I mentioned RatGDO and literally own one. My point was that there's no possibility Chamberlain would ever give you any of these things in the built-in Wi-Fi-connected hardware you've already paid them for.
It is unfortunate to reward those weasels for their bad behavior by buying their device, even if one substitutes their own "brain" and never uses MyQ. But yeah, that may be the only practical option.
reply