Bah, I should have read this thread before integrating bubbles.town onto my blog.
Hm, I just looked over their Privacy page and as of now, there is no mention of Anthropic. Luckily it was archived on the Wayback Machine a few days ago.
So either they’ve figured out a non-LMM way to classify… or they’re now covering this up?
I was just about to ask you if you’ve heard, since I saw Bubbles come up in your last blogpost.
Thanks for spotting that change to the Privacy page. o.o I wonder what’s going on there. From quick ctrl+f on the official blog I’m not seeing any announcement about it or anything.
Not for running the site. Posts, votes, comments, ranking, search, and the category assignments all go through plain code. Categories are decided by a small statistical classifier (Naive Bayes) that ships with the binary and runs on the Bubbles server, with no external service involved.
For writing the code itself, I use Claude Code as a coding assistant during development. That is a tool used to produce the source, not a system the running site depends on. The deployed Bubbles never makes calls to any AI service at runtime.
So the category assignment system has changed..? And also, as others suspected, the code itself partially involved an LLM.
Thank you very much for the additional information. AI being added after word started spreading would explain the sudden drop in quality soon after I praised the site for its blog selection. It’s good to know the decision was reversed since then.
That said, I already deleted the domain from my browser history and start page and won’t be going back. Please don’t ask why.
huh. Interesting. Thanks for bringing this to us. I did not expect to see myself quoted here lol but I suppose that’s what I get for posting.
This does answer some questions, and it’s good to hear there’s an announcement planned, so, given the givens, not the worst possible outcome here I suppose. I’m glad that Ben apologized and seems invested in addressing some of the criticism.
With that said, this exchange leaves me with more questions.
The context that Ben provided here for the original decision was that it was quick. But why would it need to be quick? What’s the rush? And even then, why resort to Anthropic about it? Was Ben simply unaware that Anthropic is controversial and why…?
Even setting the LLMs aside though, I remain confused at the original reception to the premise. I’ve seen multiple indie web blogposts enthusing about Bubbles at this point, such as I Love Bubbles (shared to This Week In The Indie Web), and I’m like… what am I missing here? This kind of thing seems tailor made to exacerbate the kind of thought patterns that @candycanearter07 expressed – concerns that their posts won’t rank highly in this mass cross-topic ranking contest. I don’t know what to expect for sure, but I would generally expect this system to risk burying the blogposts with more niche appeal, including some of the topics that I’m actively interested in, such as perfume blogging.
i’m sympathetic to the premise of bubbles. “non-tech content gets drowned by the tech majority” is something i agree happens with a lot of existing independent web discovery mechanisms, and i’m rooting for any alternatives that are focused on non-tech interests! unfortunately, i’ve checked bubbles every couple of days since this thread was made and i’ve yet to find a single post that interested me…
ethics of the category assignment system aside, i don’t think it works very well! despite the goal of bubbles, the tech majority is still drowning out everything else. as soon as the category checkboxes were added, i deselected “tech”, yet six of the top ten posts i see right now are on tech related topics: a post on AI summaries classified as “writing”, a post on vibe coders as “culture”, a post about working in tech, with tech in the title, as “life”. most of the top posts are meta posts about blogging itself (or worse, about bubbles itself), which i think should be classified as tech! i’d rather read about anything else!
i think it would take a dedicated effort by the whole userbase to actively avoid upvoting tech-related posts for bubbles to work according to it’s initial premise. otherwise, the numbers just don’t add up. all the people i know with fediverse acccounts are interested in tech! sure, they each have additional diverse interests, but an interest in tech is something they all have in common, so tech topics are always going to get the most votes.
it’s possible that bubbles could evolve into something that interests me, but as it stands… eh. i don’t have a fediverse account, and i don’t think it’s worth my time to make one just to try to nudge the site culture in a direction i’m less bored by.
I completely agree. I think having posts be automatically aggregated and categorized is enticing, and it means that it’s very hands-off, but this is the trade-off.
This could be solved with allowing more direct community intervention, such as having downvotes rather than merely not upvoting passively (I think the Internet’s removal of downvotes has done more harm than good, maybe an unpopular opinion). But also if there was a team of moderators (or experienced users) that were able to manually change categories. At this point it’s probably an entirely different project, though.
Also, sorry! I didn’t mean to infringe, it was just a really good point and I wanted to give credit for it.
Thank you for saying it so I don’t have to. It makes sense for this system to converge toward what visitors have in common… and as it happens, the biggest thing that Bubbles visitors have in common is looking at Bubbles.
(There may be a novelty factor that could fade with time, of course, but I’d expect the meta-blogging posts to be a mainstay.)
I’m also wondering about how much of this outcome could be chalked up to the initial selection process. The About page says that Bubbles is scoped to “independent” blogs, and the FAQ mentions that feeds were drawn from ooh.directory, indieblog.page, Bear Blog, blogroll.org, IndieWeb Webring, and personalsit.es. So I’d infer that the initial roundup is unlikely to include any blogs hosted on Tumblr, Medium, Substack, or Dreamwidth, and it may not have much if anything from Blogspot or free-tier Wordpress. For that matter, I’d be surprised if it included anything like Now Smell This or Coyote Yipps. The project seems to be more narrowly focused on blogging with a high technical threshold… which is necessarily going to skew toward people with interests in tech.
If you are genuinely someone who “enjoys reading blog posts about things other than startups and JavaScript frameworks” (per the FAQ), then you don’t have to shoot yourself in the foot like that.
I figure what you’re describing sounds more like LJ/DW/PF-style moderated communities. Or subreddits, if that’s your closest point of reference.
I’m sure you didn’t mean it that way, but your assumptions come across as rather condescending. There are indie blogging platforms (like Bear and, yes, Dreamwidth) that don’t require anything beyond knowing how to register and type posts. Many of us on these platforms don’t read or write tech blogs, either.
If bubbles is meant to gather indie web links, why should it include places like the ones you mention? Out of those, Dreamwidth is the only one I’d consider indie. If bubbles should include Tumblr and Medium, why stop there? Why not include Facebook as well?
Finally, I don’t believe that directory creators absolutely need to please everyone or include any links that don’t fit their vision for it. There’s room for many directories, including existing ones that are exactly what you’re describing.
P.S.: A very simple solution to people worrying about votes is to simply not have that mindset.
In that case I appreciate your faith in me and my intentions otherwise.
Dreamwidth is the only one I’d consider indie as well. What I had in mind, and what I’ll try to spell out more explicitly, is that an emphasis on independent blogs and an emphasis on non-tech blogs appear to be in tension with each other. They’re not mutually exclusive, but a stricter emphasis on one can come at the expense of the other, it seems. That’s how we get for instance @ribose expressing an interest in the premise of Bubbles and yet being disappointed that it’s ended up with so many tech posts. My suspicion is that if Bubbles were to broaden its criteria for “independent” blogs, then it could aggregate more blogs and more posts on more varied topics.
With that said, this would not necessarily solve the audience voting problem that’s liable to converge back on tech posting and blogging-about-blogging.
I do not have ideas on how to fix this because I still balk at the foundational concept of the site. It’s interesting to think about though, and I wonder how we could view Bubbles as an indicator of unmet desires or even conflicting priorities in indie web spaces.
That’s the thing: I disagree that there’s any tension between independent blogging and non-tech blogging and I don’t think the criteria should be expanded to include corporate websites known for flooding the web with low-quality stuff. Again, I would say that a directory’s criteria are up solely to the directory’s creator.
I’m going to bow out of this discussion. It’s clear that you have a specific vision for what a directory should be like, which is fine, but I don’t feel like reading posts reiterating the same things over and over will accomplish anything on either side.
Sorry for frustrating you then, effigies. I’m not sure I’ve been explaining myself sufficiently, since including corporate websites known for flooding the web with low-quality stuff isn’t something I want either.
Anyway, I’m wondering if another part of the appeal for Bubbles is the layout that displays a mixed feed of the posts themselves, as opposed to a directory that features only the site-level blog titles. Feels more… direct, I guess? So that’s one way that Bubbles distinguishes itself from the other directories it was initially drawing on. I wonder if other folks might want to iterate on that template with other projects in the future.
As someone who does prefer more indicative post titles, I wonder if recontextualization may be playing a role here as well. Is the complaint about generic posts showing up from a blog that the reader deliberately subscribed to, or is it that those posts are showing up in an aggregator? The original complaint begins “I’ve been scrolling the Bubbles new feed,” so I think that’s our clue.
Not every blogger is necessarily expecting all their posts to get automatically broadcast to an aggregator, let alone writing with that broadened audience in mind. Meanwhile, if someone is browsing a site that encourages you to vote on what you see, a generic journaling post might look out of place. So in light of that, I’m inclined to point out that Bubbles is recontextualizing people’s posts as part of a competition that not all of them necessarily asked for.
See also this reply from @AetherAnne on how these kinds of aggregators can discourage providing an RSS feed. I would think that many indie web advocates would consider this a detrimental effect on the community.
Hot take: I think things like bubbles.town are bad for the indieweb, smallweb, boringweb, blogosphere or whatever you want to call it.
That ranking mechanism is just a drop in replacement for any other sorting/scoring algorithm.
I am seeing many posts very obviously doing “optimizations” like publishing replies to the top posts, using similar titles or talking about the same topics…
This is introducing the toxic incentive to try to rise to the top of the list and I hate it.
Personally I appreciate reply posts, but the overall point stands about how Bubbles has reframed blogging explicitly as a competition.
Note I think these concerns are worth bringing up because I’m continuing to see Bubbles being recommended and endorsed again and again and again and again, to the point that I’m contemplating whether it’s worth organizing my thoughts into a separate post.
Speaking of aggregators discouraging RSS feeds … I just stumbled across this one via Mastodon. For some reason, the individual behind this aggregator thinks it’s completely fine to use generative AI to create logos and summaries for the feeds featured there. I would be beyond livid if someone did that to my blog/feed.
This stuff should be opt-in only. I can understand the need to curate an RSS aggregator during the initial process of getting it off the ground, but would it be that hard to reach out to bloggers first? Some people don’t care if their feeds are subjected to a popularity contest; others do. Some people don’t care if generative AI is used to create logos for their feeds; others do. It’s poor RSS etiquette to assume another person’s feed is free to (ab)use however you like.
I haven’t caught up on this thread but I can at least say that I haven’t been using bubbles very much to find new blogs to read.
I do find it (interesting?? ironic??? not sure what word I want to use) that for an aggregator that is supposed to be a “Hacker News but for non techy blogs” has the Tech category dominate the hot topics. As I write this 16 out of the 30 hot posts are in that Tech section…
like i said in another forum, the issue i think is more that the kind of person who would be there would also be interested in technology, so since the majority of people have that interest it just ends up being the majority.
on the other hand, i think that randomized aggregators, like sortition social, are pretty nice too probably
ill admit i agree and it does trigger anxiety a bit (lot) but i still do like bubbles for like the variety of blogs, i jsut cant subscribe to the unrated feed because theres too many posts and i also cant vote because i hate mastadon so
As I’m so new to ‘this’ (as I gesture broadly at everything), something I alluded to in that very post, I would love to gain a far better understanding and perspective on things.
Yet to add the personal context that is missing from my asynchronous post, Bubbles - be it good, bad, indifferent, or something else - is what helped me surface so many engaging conversations this past weekend and I simply wanted to share that with others. If I “mixed the baking soda & vinegar” so-to-speak, I wholeheartedly apologize.
Edit: Bolded the “link” in question within the quote. My trust level required me to remove them.
I know I’m being naïve here but I personally don’t see having upvotes and the arbitrary formula* as competitive. What I find on page 7 or 8 vs. page 1 is not lower quality, the sorting to me is just a way of sorting.
I do agree these kind of things should be opt-in, and that the creator of Bubbles was reckless to rapidly accumulate 5,000 blogs on his own (especially with the help of genAI). And the fact people disable RSS because of aggregates or genAI or any other reason is bad, as it is shutting down an accessible way to interface with the web.
But I do find Bubbles useful for what it is, I think it does good in encouraging more discussion and communication and connection between people and their individual sites, and I’m unsure if it’s doing more harm than good. I also know of other aggregates like https://scour.ing/ and https://nuchronic.uk/ and I use Bubbles as just one of several ways to interface to find the work of others that I haven’t yet added to my RSS reader.