It feels nearly impossible to have any rigorous approach when choosing a particular model and price point for a task and more like blindly picking one. The time period needed to actually get familiar with various models to a degree you can intuitively choose appropriate ones for a task is moot when it will likely be superseded faster than the needed time.

I guess if companies are footing the bills most employees just opt for whatever the most expensive model they can get away with. Even then choosing between the various leading models is the same kind of frustrating task. Every release every company has the same random collection of graphs and charts claiming the best performance on X, Y, and Z.

I think a lot of it is the old "perfect is the enemy of good" with Apple trying multiple times now to announce this big basket of all these AI features supposedly coming all at once instead of just regularly shipping new useful AI integrations every month. There was so much easy useful shit that was immediately apparent as soon at OpenAI dropped that first big voice mode years ago coupled with basic app integrations. Particularly in the context of the AI labs that are operating in that lane almost too much where it seems a new model or mode comes out every two weeks.

When RAM prices are increasing like a crypto currency we have a real societal problem.

At this point there is hardly anything left and I think it leads to some pretty dark scenarios when we have a society where we have somehow decided: fuck it, almost everything gets worse for almost all of you every single year.

The markup on concessions has always been a thing but it really is just insane to think the unit economics on 2 sodas and a popcorn must be like 50 cents and selling it from $26 (in my area). Clearly they must make the most money this way but it is just crazy that anyone outside of significant disposable income even considers buying concessions. It's priced in such a way where anyone outside of the top 5% income brackets should just laugh at the price and view it as an extreme luxury good and not ever even consider buying anything.

There was a lot in the new digg that I was concerned or at least not optimistic about but come on - are we even going to try anymore?

And you are saying that's a bad thing?

Basically all discussion platforms are broken for any sort of long term meaningful discussion which I think is at least part of the problem. Even this thread and this comment just the fact that the thread is now 4 hours old the amount of views and chance of getting many responses drops precipitously. On most platforms unless you are someone with a large following you basically have to think like a marketer and post often and early on posts to stand a chance of getting a discussion going. It's always so ephemeral too. Even though posts on a platform like HN on reddit still exist and you can comment on old threads probably 99% of the activity happens in the first 6 or so hours and then it largely ends.

It makes me miss forums where at least you had long lived threads with simple time based post order and a good chance of replies. This doesn't seem to exist on only platforms now and forums have largely faded away.

The fragmentation of discussion has also messed things up. For example yesterday I was listening to a HardFork podcast episode which is a fairly popular pod, topping the charts in at least the tech category, and after listening I wanted to check for discussion around the episode and probably leave a comment or two. I assume this episode had to have gotten at least in the low tens of thousands of plays though perhaps that is way off. I went searching for discussion and basically found a largely dead subreddit for the podcast with no threads being regularly created for the episode and an empty comment section on nytimes which any site comment section is a useless place for discussion anyways. The pod is also posted on youtube which the youtube comment section had the most activity of anything I found but the youtube comment section and the way it is structured/operates is perhaps the most useless of all the platforms for trying to have any discussion. I just don't understand how if at least say ten thousand people listened to the episode surely at least 1% would be interested in discussing it and 100+ people going back and forth would be a large, active, healthy discussion somewhere.

Even threads that seem "active" on sites like HN or Reddit in the context of the actual audience sizes are shockingly small and confuse me. For example The Pitt season 2 just premiered and posted 5.4 million viewers, the subreddit post ep discussion currently has 5.3k comments which is quite high for a show. That is a joke of a percentage though, 0.1%! and even worse in the context of people that are posting probably post more than once in the thread. I understand many and even most people not wanting to post to discuss a show they just watched but how the hell is less than 1 in a thousand!

This post got long which also damages the chance of any engagement due to TLDR culture.

https://chromewebstore.google.com/detail/tabsio/kfgpkekbcoae... (I have done zero marketing on this hence why basically no users)

Comments URL: https://news.ycombinator.com/item?id=46472581

Points: 3

# Comments: 2

On a separate note also not mentioned llm's are really good at generating ffmpeg commands. Just discuss with chatGPT your source file and goals for a video and you can typically oneshot a targeted command even if you aren't familiar with ffmpeg cli.

On the age verification thing the only reasonable proposition i've heard would be a feature that allows parents to set some setting that gives a device users age or age range for mobiles and tablets. I think this covers a reasonable percentage of use cases if your goal is actually protecting kids and not just using that as deceptive cover to sneak in widespread surveillance laws. A simple setting that says for example this ipad user is 10-13yrs is privacy preserving enough and would not negatively impact adults and because it would be coming from the device itself would actually be harder to get around vs VPN's or spoofing IDs, etc.

The idea of trying to address all devices in all scenarios is absolutely preposterous in my opinion.