AlexAplin

This doesn't feel as much of a bellwether as I have seen celebrated than the cold fact that if you make it hard for teens to have relationships and sex with your chatbots [1], they will jump to the next boat. Character.AI has survived long enough to enjoy the squeeze in lawsuits, but the way users talk about these services always feels very much as fly-by-night as nudify apps: extract what you can, while you can, before it goes away or gets sanitized by restrictions.

Disney showing up here again after their Sora flirt is interesting, to say the least.

[1] https://blog.character.ai/u18-chat-announcement/, https://blog.character.ai/an-update-on-changes-to-our-under-...

>the mere knowledge that a solution exists "contaminates" efforts by both humans and AI to find alternate routes to the problem that reveal additional insights

This really expresses the heartburn you see across all fields, not exclusive to careerism. I certainly have friends in decomp and fan translation spaces that have been demotivated by the current rash of efforts happening there.

The rush to be "first" has always been over-celebrated, but it would be nice to believe there's a way to get beyond that thinking.

We do have some unique carve outs already for what we consider intellectual theft e.g. trade secrets. In this case the original artifacts might remain, but admitting the market that created them could go extinct is enough definition to be infringement at least. Fair use has been really resilient in these training cases so far but it is pretty damning to admit a negative market effect and that you're a direct substitute (see Warhol v Goldsmith recently).

Democracy, even at its most ideal, is often about overriding preferences. Netflix helps to float an entire industry of creators and a tangible pop culture. It's not entirely clear (or softened, maybe not clear to communities) what AI build out floats currently except AI companies and the tangible negatives of generative AI.

They aren't my indicators, but they read as sweat against the observable metrics being in decline. There is a perception that they want to pull on users outside of Bluesky that I don't really think exist, and I guess I share in that feeling when I believe the core still has so much basic parity missing with X et al that would demonstrate confidence in the current users you already have. Still interested to see what they ship as.

It's probably silly to make OP's judgement of any of these implementations in isolation, but you can sense a real fear about the growth (or lack thereof) inside Bluesky. Fiddling with the presented metrics (likes count getting moved ahead of reposts), promoting total user signups while being silent on collapsing DAUs. They are promising a lot at once with AI custom feeds from Attie and subreddit-style communities that smells like feature creep.

The attendance was clandestine and performed by actors that misrepresented themselves to gain an intelligence advantage. I'm not sure what else you would call it.

Feeds still have a lot of inertia, at least. Most blogs generate them out of the box, and Wordpress being as predominant as it still is means you can still find them even where you don't expect to. Even social media profiles can compress into feeds with a little help from solutions like Nitter (or in Bluesky's case, offered natively). What I think is really missed is the social component that Reader had, which you still can't really find even with the variety of great readers that came in its wake.

I know I will probably just migrate them to RSSHub when the time calls, but YouTube's channel and playlist feeds have been such a great way to follow a large number of creators. I'm still holding my breath for the day they unceremoniously stop working.

Any boast about AI accelerating the development lifecycle comes with an equal or greater stock that examples like this become more frequent when there are less opportunities for someone to interrupt them with a smidge of common sense.

I should have specified I meant the custom melodies. My understanding was that they're now all generic. https://sheets.works/the-bells-of-tokyo is a good reference for the transition.

JR East is already in the process of eliminating departure melodies as they transition to one-man station operations, so these will unfortunately be gone sooner than later. The Nambu and Joban lines got rid of them last year and it looks like the Yamanote is scheduled for them to be gone by 2030 [1].

I'm sure they can figure out a way to trigger custom melodies with RFID or similar eventually. Keikyu figured out how to recreate their departure boards [2]. JR might be less willing to come up with something immediately given the optics around automating someone out of a job.

[1] https://japantoday.com/category/features/travel/jr-east-axes...

[2] https://soranews24.com/2026/07/04/japanese-train-company-bri...

Besides the context in the other comments, they pushed the Japanese fresh food angle in a media blitz pretty hard last year (https://www.nytimes.com/2025/09/09/business/7-eleven-ceo-ste...). Egg sandwiches seem to be the most reliably available in the contiguous states, but you can also spot egg rolls and onigiri. They're also now bracing to close hundreds of stores and reopen a fraction of that number to match the new model: https://www.forbes.com/sites/pamdanziger/2026/04/17/7-eleven...

Dead internet theory seems mostly proven by blog posts gesticulating about it. Digital creation is easier, more collaborative, and just as fun as it has ever been if you stop thinking in terms of mass audience and following the herd.

I'll grant you: Flash is a hole that never fully healed back. Search engines might not be especially great for discovery now either. They weren't especially great for Geocities shrines either, though.

The advertisers that evaporated and left behind a lot of no label dropshipping scams seem to think so. Did a lot of them eventually come back because there is some audience to squeeze numbers from? Sure, but I also wouldn't negate that many didn't and aren't coming back because it is Elon's playground now.

We have probably crested over some peak, but you would not look at the broad numbers and say 3% of a peak is organic to that trend. That is a dying/dead website, at least from the position of someone running socials for EFF.

https://flowingdata.com/2025/10/03/passed-peak-social-media-...

Notably, this primer on Sora safeguards was published only yesterday: https://openai.com/index/creating-with-sora-safely/

Not a great look that either the teams responsible for Sora didn't know this was coming or the decision was so brash that things changed overnight.

I'm pretty forgiving about accessibility (I'm able to say this at all because I don't have to rely rigidly on accessibility tools) but nav menus feel like a baseline we shouldn't muck with. Tabbing doesn't seem to respond very well in the live example, and at least in the limited demo you can't expand the listing without using a mouse (I thought it would respond to a space with the :checked pseudo, but seems not).

The current utilization of generative video is almost universally horrible, which this article does suggest, so I'm not too surprised there are players trying to differentiate themselves. Slop for thee, not for me.

Comparing views cross-platform is not a very useful study and YouTube routinely adjusts what a view means. Shorts changed earlier this year to count all playbacks and loops without a minimum watch time requirement. https://support.google.com/youtube/thread/333869549/a-change...

For what it is worth, they have communicated that they do eventually plan to retire the domain themselves [1]:

>Re-enrolling your security key will associate them with x[.]com, allowing us to retire the Twitter domain.

[1] https://xcancel.com/Safety/status/1982278858457174522

Cloudhiker is pretty healthy as a StumbleUpon revival. I've found lots of great personal blogs and sites across a lot of categories through it. https://cloudhiker.net/

Google had a tailored fair use argument because they never made more than snippets public and searchable. It was also prior to Hachette that controlled lending with one-to-one digital copies for every physical copy was a status quo that publishers largely accepted, which IA deliberately tried to upset with the National "Emergency" Library.

I think it's worth fighting back on copyright as a broken institution, and it should be part of the IA's mission, but you have to be responsible on your approach if you're also going to posture as an archival library with stability of information and access. I understand Kahle might lament losing some of the hacker ethos, but the IA is too important to run up against extremes like this without an existential threat.

Stick figures run through a lot of amateur digital animation, for probably obvious reasons. Pivot reigned on a lot of early YouTube and the stubby stick figure style ran through a lot of Flipnote Hatena. I'm not sure if it's simply that standards for amateur digital content have evolved, or if we have lost the character of small platforms like SFDT and Flipnote, but I do find stick figures absent today on the large platforms we've all herded towards. A lot of what I see is definitely buoyed by Flipnote diehards.

This exact arbitrage, performed by Doordash, was exploited.

https://www.readmargins.com/p/doordash-and-pizza-arbitrage

Some of the major hosted feed readers (Inoreader, Feedly, Feedbin) include subscriber count in their user agent. I usually run a filter on requests to the feed links in my access logs to get an idea of how they're changing. Anecdotally, my subcriber count reaches into triple digits with only those counts, but I've never gotten email from readers and generally only get feedback promoting new posts on socials. The counts are about as nebulous as follower counts, which is to say most people probably subscribe/follow and forget.

Quibi launched in April 2020. TikTok by this point would have 2 billion downloads [1]. It's difficult to assess they were trailblazers here. I might even say a component of their failure is free mobile video was widely accessible by this point.

[1]: https://www.theverge.com/2020/4/29/21241788/tiktok-app-downl...

Watching the firehose events they're probably clearing Mastodon for now. We'll see how that looks after it stabilizes again, the surges tend to have pretty steep dropoffs so far.

https://bsky.jazco.dev/stats https://bskycharts.edavis.dev/edavis.dev/bskycharts.edavis.d...

Neil Cicierega's Ariel Needs Legs somewhat infamously had the audio corrupted on its YouTube upload over time: https://www.youtube.com/watch?v=_nH6ya5g2-s

It actually seems better from when I last looked at it, but you can you still hear skipping and audio jitter around spots like at 0:18. I've seen similar behavior on Twitter video uploads over the last few years.