Oh please, they deserve shaming. This is a great use of social media. Put the facts out there as you see them. No need to sue or have a call to action.

Good title

The idea that users will switch because a platform has ads has never played out in the past. The average user just doesn’t care.

Plenty of iPhone users are on Brave Browser because it blocks ads on Youtube. When you type in a search query in the address bar, it defaults to Brave Search. Don't underestimate the power of defaults.

There is a boiling point for user tolerance when it comes to ads. Google has been pushing that needle deep into the red for years now.

No meaningful amount of iPhone users use Brave.

Most platforms have a lot of cost and friction to moving. LLMs are all basically the same thing.

> The average user appears to care enough to go out of their way to use browsers and extensions and such that block ads

The average user isn't going out of their way to use ad block, at least globally. Last estimates I remember are still around only 30% of internet users using an ad blocker unfortunately. Ad blocker adoption is heavily skewed toward techies.

It's an even lower amount of people on mobile, where the 'average user' spends the majority of their time.

30% of the general population is a staggering number for basically anything voluntary. I don’t know how you can possibly trivialize that number.

If you can get 30% of the population to boycott anything it’s not only significant but forces a response out of the group on the other side. You’re talking almost 1 in 3 people. Those are literally historic numbers.

Do you have any idea how much companies spend to increase their conversions by single digits?

Ok but using adblock isn’t switching platforms. It doesn’t inconvenience the user at all, and there are no negative network effects.

> Anecdotally, the number of people who have asked me about VPNs over the last 12 to 18 months has skyrocketed. I probably get a message a month from people of all backgrounds.

That's probably more to do with the increasing popularity of China-style national firewalls/tracking (sorry, "age verification") than ads.

Negative. They all want to block ads and protect their identity insofar as they can. Most of them don’t even realize the age verification stuff is going on. My friends that do know about that already have enough technical literacy that they don’t need to be asking me these questions

It’s a cool idea, but is there no way to hear the actual notes?

I've thought about adding something that would vamp certain chords, say. But sounds like you mean something different... like, play the pitches in a scale?

I very much would like some way to preview what the sound and feel of certain combinations of chords and scales/tones is, but I haven't quite figured out how it might work.

Perhaps something like alphatab could work?

https://www.alphatab.net/

For now I just mean something simple like playing whichever note you press on the piano/fretboard.

Ok, I took a stab at this!

- All the notes on fingering charts can be clicked and they play a sound

- Chords get a little playback button

- Compose mode gets its own playback controls

The sound is very basic, I'll see if I can fix that next.

It’s not a 3B model, it has 3B active parameters. The full model is much larger.

That's true, I should have mentioned active. Actual params are closer to 12B-14B likely, given the 40GB VRAM usage.

It’s not a micro model at all, it requires 40gb of VRAM. The 3B is just the active parameters.

Have you considered that maybe the experts in a field are actually correct about that field?

have you considered that at any second all our existing knowledge could be rendered redundant on the frontier these experts work in?

Yeah it has a confusing clickbait title that gives the wrong impression of what the article is about. His point is that the bottleneck to making a complex app like photoshop is architectural rather than just writing basic code… and he argues the LLMs don’t magically make the architectural part easier.

The inference is absolutely not sold at a loss, at least not when paying API prices (the subscriptions are less clear). The reason frontier model companies aren’t profitable is because training the models is so costly, not inference.

I mean everyone knows they need more compute. That’s not a secret or up for debate at all. They are maybe the fastest growing company in history.