you can't deny how useful they are!!

[hypolite@friendica.mrpetovan.com]

@jplebreton I’m here waiting for AI agents to pick up the slack in app usage and review to keep the successful AI-coded app myth alive.

[chateauerin@mastodon.social]

@jplebreton Noise to Signal Machines

[inglogamesdev@mastodon.gamedev.place]

@jplebreton I dislike AI as much as the next person on Mastodon, but this graph shows absolutely nothing. The increased velocity in releases and the drop in reviews both began well before the "Agentic AI" region. Plus, even if you claim they're still related, correlation is not causation.

There are a million reasons to avoid AI but this isn't one of them.

[jplebreton@mastodon.social]

@IngloGamesDev it's hard to prove that something isn't happening, and this is certainly well shy of what would be needed to "prove"

[tomwor@mastodon.social]

@jplebreton The only winners are the platforms collecting 30% cuts of every sale: Individual app revenue might plummet giving every developer a narrower slice of cake, while the cake size stays roughly the same. The platforms enjoying worker self-exploitation, with few winners they put on pedestals to keep the wheels greased.

[jplebreton@mastodon.social]

@samim @jack why do you fucking people always feel the need to chime in like this? don't bother answering, i'll never see it.

[lritter@mastodon.gamedev.place]

@tnorinder @jplebreton theory: every thoughtstopper has the potential to become the idol of a cult

[ariarhythmic@ohai.social]

@jplebreton Good data, bullshit graph, I wish people realized that if the y axis doesn't start at 0 they're missing the point of a graph.

[julijane@ruhr.social]

@ariarhythmic Seems this one is on you. The graph shows relations in percentage and puts 2024 average at 100% which is the "zero" here. It goes both above and below that. It absolutely makes sense.

@jplebreton

[ariarhythmic@ohai.social]

@julijane @jplebreton Oh, nevermind, thanks for clarifying!

[wooshell@chaosfurs.social]

@jplebreton .. and that's on iOS who at least do *some* vetting of submitted apps. It's probably three times as bad on Android where nowadays you can get almost every bullshit into the app store.

[larsmb@mastodon.online]

@jplebreton I think the GenAI hype is a cancerous pit.

However.

The criticism of this graph rests on the same logic that produced GenAI: that only hyperadoption and market share growth matter.

A claim made by GenAI is for more custom software to be adopted, which naturally has lower user rates on average.

Whether or not that's happening can't be decided by the currently available information, I think?

[morj_lecteur@h4.io]

@jplebreton Unfortunately contrary to what I thought, this does imly that LLMs boost programming productivity. That that productivity is spend on garbage (borderline spam) is a separate concern

[cupz@mas.to]

@jplebreton Most software ideas are not useful. Working very hard on an idea for a longer period of time is like vetting that idea. Now all dumb ideas can get some form of implementation.

[petealexharris@mastodon.scot]

@morj_lecteur @jplebreton

It's not impossible that it's helpful for some part of the process. Even conceding that, all the graph seems to show is a x1.8 not a x10. Which is not nothing, but not revolutionary.

The relative fall in apps with significant usage suggests usage is being spread out over a wider area but no significant new scope of usefulness has been discovered. Even if we get to x10 development, there aren't x10 use cases to apply it to and x10 income to attract.

[hypolite@friendica.mrpetovan.com]

@petealexharris @morj_lecteur This graph shows the relative change, so the actual app volume chart would be a pretty fast exponential curve. Aside from that, I agree with your analysis.

[petealexharris@mastodon.scot]

@hypolite @morj_lecteur

Yeah the volume of apps would be going up faster than linear, since it's already a graph of rate. I feel like rate is the relevant measure since, per project, the absolute number of other apps that exist doesn't matter for how much easier it is to write one now.

The rate at which apps are being produced per month only seems to have gone up by x1.8. It might go higher, it might stabilise, it might fall again. I don't think it'll reach x10 any time soon.

[oatmeal@kolektiva.social]

@jplebreton I didn’t read yet but wonder if there’s a carbon footprint cost estimate to this flood of garbage (also I’d like to see how many exist only for demanding a monthly subscription).

[dotjayne@tech.lgbt]

@morj_lecteur @jplebreton imo, it only implies that LLMs boost programming output of garbage apps. The corresponding decrease in apps with significant usage could indicate users bouncing around from app to app, not finding anything better to settle on, but still displaced from long-time favorite apps as the whole software ecosystem loses stability to LLM cruft

[bri7@social.treehouse.systems]

@jplebreton why buy the milk when the cow is free (for now)