I feel like all of these decentralized attempts to respond to the ills of social media/social products are going to never make it out of the curious POC phase simply because the vast vast majority of people do not care about hosting their own instances or owning their data to that degree.
When it comes down to it, there are 3 key issues with the "traditional" model of social products:
You are the product - This is true of 1) your data you put into the system explicitly, 2) the data you create by using the system, and 3) the predominant business model (ads) of these products. This could be fixed by:
1) some easy means to extract all data you create in an accessible format (only some users will care about this)
2 & 3) Allowing people the choice of whether they are the product or not. "Subscribe for no ads" seems like a good middle path alternative to selling ads without demanding a whole new architecture and operational model for federated solutions.
Separate of the above ideas is the fact that social products are really addiction mechanisms. Reddit/Hacker News/Tik Tok/Instagram are all heavily optimized to addict their users to their product. This is, in my opinion, where the real risk lies overall, because it is directly leading to aggressive micro feedback loops and information bubbles that are evolving faster than humans can process. I'm not sure what to do about this problem save for outright control/limitation attempts, which seems like a guaranteed mess.
The vast majority of users absolutely don't care about data residency or "privacy", however you interpret that word. You can convince individuals to care, but good luck convincing a large group.
What can happen (and has happened) is that the existing thing becomes just toxic enough to lessen the pull and invite alternatives. For the tech communities I've been a part of this has happened a number of times in the history of the web — off the top of my head:
• ExpertsExchange -> StackOverflow
• IE 6 -> Firefox
• Twitter -> Mastodon
In all cases the migration was driven by software engineers' disgust with the platform. The current Twitter -> Mastodon migration is mostly just folks in tech, but there are enough of us to forge decent-sized communities there.
And yet, today Firefox has single-digit-percent market share, and the dominant player is in some ways worse than Microsoft was. MS seemed to want to make the web stagnate, while Google wants to turn the web into (more of) a privacy-free, advertising machine, and has been steering web standards in that direction.
Google allows Firefox to exist (by funding Mozilla via Firefox's default search engine) only so they can point to the fact that there are alternatives, with alternative rendering engines.
It is unclear if Mastodon will actually replace Twitter. I'm skeptical. I think Twitter would have to actually fail (as in, go bankrupt, or at least have such severe technical uptime/stability issues) before the platform would be abandoned enough to definitively say Mastodon won.
ExpertsExchange -> SO mainly happened because all the decent answers on EE were hidden behind a paywall.
> And yet, today Firefox has single-digit-percent market share, and the dominant player is in some ways worse than Microsoft was. MS seemed to want to make the web stagnate, while Google wants to turn the web into (more of) a privacy-free, advertising machine, and has been steering web standards in that direction.
I think that reinforces GP point that consumers aren't driven by privacy. Firefox didn't take off due to privacy, it took off because IE6 was a miserable experience. The killer features that drew people to firefox were pop-up blocking, tabbed browsing and less crashes.
And even then, Firefox's decline makes the same point; Chrome didn't win because it was more private, it won because it was fast, stable (when Chrome first launched, its process isolation features were unmatched - suddenly, one tab crashing didn't mean crashing the whole browser!), and of course because Google abused its search monopoly and bundling deals to cram it down people's throats. But for all that they cheated, they could have just won fair and square because they gave people features that people actually cared about; privacy was never relevant, and security was secondary.
The IE 6 to Firefox migration started about 20 years ago. The world of browsers today is _very_ different.
You may not like Google's influence, but they don't hold back technical innovation in the way that Microsoft did with IE 6. Many web designers, me included, spent hours on each project hacking around its CSS bugs and JS drawbacks. I've not had to do that with Chrome, which makes it miles better.
>You may not like Google's influence, but they don't hold back technical innovation in the way that Microsoft did with IE 6.
No, the problem with Google's huge market share is that they can use their huge market share to push the web in ways that aren't standardized or in most people's best interest. They end up breaking the web.
IE6 wasn't "toxic". It simply was very bad and outdated. Firefox and Opera were better browsers, plus the browser doesn't really have major network effects.
The trans community is currently the wellspring which most reliably produces top developer talent.
When trans folks saw the toxicity and hate Elon was enabling with Twitter they fucked off to Mastodon, and the barycenter of tech discussion shifted in a less Twitterly, more Mastodonish direction.
It has enough traction that it’s a vibrant set of communities. I use it to share bird photos, talk about adhd/autism, and follow some climate science. Its smaller than twitter, but its big enough to work now.
not the GP but I've been happy on fosstodon.org. It has been reliable and I've found a lot of good people to follow on it, including the founder of MNT.
Exactly. Everybody seems to look for the place "where everyone is".
Truth is that there is none. People are always spread around many different places (and it’s a good thing).
Is everybody on Mastodon ? Absolutely not (but you would be surprised how many people never were on Twitter either).
Is there enough people on Mastodon to keep you informed, entertained and making you discover new stuff? Absolutely.
Is your niche interest represented on Mastodon? I don’t know but it might be, although with different "codes/traditions" which might make it harder to find at first glance.
Will Mastodon be the next big thing? Well, nobody on Mastodon seems to care about that question.
Is Mastodon a success? Of course it is for the thousands people using it daily.
heh, yeah, I was going to say -- I don't use twitter and have used mastodon for years and do not expect it to become the next big thing.
is that important?
I put it like this. If you expect 1-to-1 mapping of your twitter social network and posts to mastodon to work before you switch, you will fail. Expecting that or something close to it is not unreasonable to want (okay, it's historically very unreasonable), but it is not what we have.
When I joined, I had one friend on the network. Then two, then I followed random people until I built a new social network organically over a year, and unfollowed people and now everything works nicely for me. I have a new social network, and all it really took was being willing to click follow and be patient.
It was slow though, or else would have been work, so I do get that this isn't a recommendation. It's honest though, and that will keep it from being the next big thing until everyone on Twitter is gone and the networks built from scratch are the place to be (and now more gentle to do that because you're a latecomer).
Next big thing though? Nah. I think it's a sustainable thing though, like Linux pre-smartphone, and one that doesn't need to be the next big thing to be big enough to hit perpetual improvement and growth until it does become closer to the big thing a few down the line..
We've got a lot of proprietary sites to crash and burn their VC funding by which point federated services are going to look like an oasis of stability where you can go back and find your friends when the trendy new network fails.
I don't know if mastodon will be a household name, but I bet it (or another federated service) outlasts bluesky, truth social, gab, reddit, twitter, instagram, youtube, tumblr, facebook, makes me wonder more at what point it's got a signal, email, dropbox, or online collaboration system somehow. I see no reason to think the network is boring enough that it will just peter out without active sabotage so far. Maybe everything will be more boring in this context in the future, that would be okay with me.
While I am also not very bullish on any fedi-xyz that requires special extra effort from users, I think there's another problem with the current popular schemes that makes your (2&3) solution problematic: money incentives, over time.
Even if a company today were to offer "subscribe for no ads", I don't have especially great confidence that it will stay that way, or that the quality of the product will be maintained over time. Fundamentally, if you have a situation where making money is the goal (whether via ads or subscriptions, etc), then that is the goal that will be pursued. But a healthy community is not about money.
One way to "keep the money-grubbing company honest", as it were, is to have genuine competition. Maybe there are other good ways, but that's at least one. There needs to be an existential threat to that company, at all times, that its users (and thus revenue) could leave and go somewhere else. A federated approach at least somewhat pushes things in that direction. I'm curious to see how much traction it gains.
Maybe there could be a non-profit approach. Or maybe "online community as a public utility" approach; I really don't know. But the money-driven incentive tends to be a slow poison, I think.
Hosting your own instance isn't a requirement for joining any of the alternatives I've seen so far (e.g. Bluesky, Mastodon, Lemmy etc.) but I'm open to being corrected if you've seen this stipulated. As for the rest of this comment, it's really predicated on certain ideas about what websites have to be based on what they have been over the past 15 years. In fact it's highly possible to participate in community-run efforts that aren't driven by profit incentives, which also don't aspire to drive addiction.
There is mass attention on the pitfalls of centralization right now - Twitter, and now Reddit, two public squares annihilated by corporate greed.
People are learning that the rewards of social media are possible without the ennui machine sucking their souls. They do care, but many don't understand enough to navigate this combination of tech and social structures. Now is the time to educate them and take advantage of that.
One of the things people may have to learn is that it might not be possible (for several reasons) for privacy respecting user centric online systems to equal the ease of use and friction free on boarding of centralized silos.
There are both technical and economic reasons for this.
If people want freedom, privacy, and control of their own data they may have to tolerate a somewhat higher friction user experience. This may be unavoidable.
The cost of online freedom may be a few extra steps.
> Allowing people the choice of whether they are the product or not. "Subscribe for no ads" seems like a good middle path alternative to selling ads without demanding a whole new architecture and operational model for federated solutions.
The main problem with social media as it exists today is not advertising (as much as I despise advertising), but the toxic, addictive, engagement-at-all-costs nature of how the information is presented.
I agree with you that most people don't care about their data ownership/privacy to the point that they'll host their own instances. And if they don't do that, then they are beholden to whatever "algorithms" the centralized players want to use to ensure high levels of engagement that keep people on the site/app, and coming back to the site/app.
But it's the advertising that drives all of the stuff you do hate.
When a business is ad-driven, human attention is the raw crude that it harvests to turn into a product (ad impressions) that it then generates revenue from. The more attention it can siphon up, the more ad impressions it can deliver and the more money the business makes.
Does your coffee maker chime every twenty minutes to helpfully suggest you might want another coffee? No, of course not. Because the people who made that coffee maker already got as much money from you as they're gonna get when you bought it. As long as you're happy enough with it to not switch to the competition, they don't care how much you use it. They get no additional revenue from your additional use.
With ad-driven businesses, every user is a potentially inexhaustible source of future attention, which is what drives all of these toxic user experience choices.
I've been using Pixelfed since I discovered it after Elon bought Twitter. It's helped me get back into photography and I post daily. Of course the userbase is tiny compared to Insta, but I don't care.
I'd rather grown a small but sincerely interested community around my work, than a large, mostly indifferent one.
When I was still active on DeviantArt regularly, but before I got Gallery Director status (2004-2008 ish) I sold little prints, but consistently. After GD, I overall sold about double, but the interaction was like an EKG, and full of noise and stupid complaints and random stats-whoring from commenters. It felt like too much of a rollercoaster managing whatever turdstorm may or may not arise.
I'm still trying to bootstrap a regular pace with PF, as I'm still a very active IG user, but I may just test out if the Import function can handle JSON file with over 10K posts...
Yeah, I joke with my daughters that I have EIGHT followers, and my financial windfall is just around the corner.
I used to do quite a bit of photog and even majored in visarts, but lost interest until recently. Nice to have something simple like daily posting reinvigorate an old interest. It's like writers who have to write every day. After a while you get decent at it.
I started posting my art online 20 years ago with DeviantArt and my original blogging efforts, but it was predicated on the fact that I had an art practice for my own edification, so the online part just became an adjunct aspect. I was "gonna art" regardless, so why not syndicate what that is, regardless of what it generates.
Keeping that attitude has been healthy in terms of what I produce, but I do find that I would prefer to be using a service that allows for some actual community elements, which DevArt has allowed to go fallow, and other sites never quite had (e.g. Society6) but at least have good artist services (e.g. prints)
Just curious then...why use it? Do you take photos for yourself, or do you like to do it for a small audience? And if the audience on it is small, and your potential audience on IG is small, why not just use IG where you have a chance of getting a bigger audience?
I have a couple of goals where Pixelfed serves me well. I want to improve my photography, as well as get outside (away from all my screens) more. So I've encouraged my OCD/hyper-fixation to post daily. To improve something, I think you need to have criticism of some type. I measure that in terms of likes or reposts on pixelfed. It's not perfect, but it is some validation. Sometimes people post comments, but that's unusual.
I prefer Pixelfed since I don't want to feed the Meta beast.
The captcha to sign up shows AI-generated images and asks things like "select images of a dog" but there are zero dogs, just dog approximations. I really hope these identifications won't be used as training data.
hCaptcha has been doing this for a few months now to my knowledge - I'm guessing this is being used to train a CLIP-like system (or alternatively: it might just be that poorly generated ai images are the best candidate for a captcha in the age of modern subject recognition, it's hard to know)
I mentioned noticing it several months ago, but it sounded like others hadn't seen it yet. It's always hcaptcha (or at least the atrocious ones are, maybe the others are just super good).
I have the same thought at first, it's really amusing to think that they expect me to select silly wrong things to pass a test and mistrain their system. I like to hope.
Alas, statistically they can totally do this by showing the same images to multiple people so that they can estimate what most people like and then define a threshold outside which they predict that it's not a human. One dimension is related to their AI tuning variables or whatever, the other is related to their analog signal from human measurements of the same images. They surely do this already since humans also have noise in their responses to a survey with anything that's not obvious.
They can still apply on-off thresholding to the analog signal that is a survey over many participants. It still gives these ridiculous gas-lighting training systems though.
One thing that's probably true -- they're finding out what the lowest quality image is that people will identify as something, which is unique. Usually people vote for what they like best, but if the stupid thing forces you to click on a certain number they can adjust that threshold.
Maybe they intend to touch up subsections of images such that they're closer to these minimum quality thresholds, who knows... just something different.
Tangential caution to anyone wanting to try Pixelfed. Host your own instance and post on it. The official pixelfed.social is no better than Instagram in banning accounts for no reason except that the main developer of Pixelfed (who is also the admin of that instance) thinks of something as spam without reading and without issuing warnings or any indications. Unlike Instagram, which at least has an appeals process (in theory), he doesn’t respond to emails and doesn’t respond to queries sent through other platforms where he’s active either.
I learned it the hard way that whether you choose a centralized platform or decentralized platform, you’re at the mercy of algorithms or someone’s emotions and trigger happy fingers.
Your data is never your data when you’re on someone else’s hosted platform.
> Host your own [Pixelfed] instance and post on it.
I'm leaning towards that but hoping that someone comes out with a more resource-friendly implementation (hopefully based on Rust or Elixir with SQLite database.)
The next thing I'd like to see is a way to make it ongoing - specifically I'd like to have a automatically updated backup of my wife's Instagram account. She posts a lot of things that are relevant to our family, and it'd be nice to know we'll still have it even if something were to happen to her account.
It also sounds like Stories are not currently imported. That makes sense for a one-time import since they are meant to be ephemeral, but for an ongoing importer/backup, it would be nice to capture those as well.
> specifically I'd like to have a automatically updated backup of my wife's Instagram account.
Instaloader is pretty good for backing up Instagram accounts but it's easy to trigger Instagram's anti-bot protections using it. I use a secondary account to avoid locking out my main and even then it took ~4 days to download my 9k posts because of the lockouts and rate-limiting. But once you've got the main bulk down, you can run it once a day to fill in anything new and that's pretty reliable.
The thing that annoys me is that there is no propper way (at least to my knowledge) to run multiple linked fedi accounts. I have a selfhosted misskey instance, but then should I also run pixelfed? Or should I just use the singular instance?
Everyone just has multiple accounts, but I'd say it still works better than having separate twitter, instagram, reddit, and youtube accounts. You may need a peertube account to post a video, but just about anyone can comment if they like. Can't comment on a youtube video from twitter.
I have the same problem. I would love to be able to toot from mastodon and post videos from PeerTube to the same account. Right now it is very segmented, even though most fediverse apps can process incoming posts of various formats (so you can follow PeerTube users from Mastodon no problem) but you can only post a limited set of content. They have mixed identity and applications.
I've been toying around with the idea of an ActivityPub proxy/aggregator/delegation which would "front" all your services and act as a unified identity for the proxied instances.
It would inherently require trust, so the only way it would scale beyond self-hosters controlling their own domain would be co-hosted alongside some SSO-unified fedi services (or if S2S delegation becomes an adopted thing).
(If you run with this idea, consider AGPL for all parts :))
It would also require that (almost)all the existing fediverse services be redone from scratch because each of them owns the underlying ActivityPub actor (that represents a user) and are unable to work with one coming from a third party.
I don't think that would suffice. I want all my data to be in one place, hosting one service, while still being able to use frontends for all of the different applications. It's a big ask for sure.
I'm a big fan of liberating data (especially my own!) from walled gardens, but I have to assume that whatever they're doing is against Instagram's terms and Instagram will do everything to stop/break this.
It would appear that you assume incorrectly. Instagram will let you dump all your data into a JSON. All Pixelfed is doing is reading that JSON. Pixelfed has no contract with Meta that would block them from reading in that JSON.
As a victim of Meta's Cease and Desists for open-wa, the contract terms is in the Terms of Service that anyone associated with Pixelfed, or anyone associated with the development of Pixelfed has ever accepted regardless if it's a personal or testing account.
In these ToS agreements, Meta believes they can prevent anyone making any tools that they deem break any of their ridiculous terms of service.
I literally had to stop working on my OSS project/tool due to this bullying. Unfortunately I, or other Devs, do not have enough resources to fight this immoral megacorp so therefore they win our compliance.
The neat thing is that it unzips the file locally and only uploads the posts you select, which Dansup says helped get around very large instagram archives and also means you don't have to worry about posts you don't want imported being sent to the pixelfed server.
IG has to allow an export function, if nothing else to help with compliance (e.g. CCPA) and in a digestible format. All PF has done is built a parser for said JSON formatting.
This is marketing speak. Do not call it walled gardens. Called it curated, fixed, whatever.
Gardens have delicious fruits, vegetables, and beautiful flowers. It creates a positive emotional spin on what is otherwise something bad for consumers.
I've only ever heard "walled garden" used in a negative sense, so I think most people get that it's not meant to be a good thing. Emphasis is placed on the "wall" rather than the "garden".
TIL Pixelfed exists. It's a pretty small network though, they say they have 163K users. They need some stupid Instagram policy change to get more traction.
No. Pixelfed is part of the Fediverse. You can follow anyone and be followed from Mastodon, Pleroma. This means that we have already almost 10M users to interact with. [0]
Stop thinking in terms of silos and start embracing the re-emergence of the World Wide Web.
One of the cool benefits of federation. I've already seen a bunch of Lemmy/kbin instances instantly have a bustling, unique community because people could post to and follow communities from Mastodon.
Agreed. As a developer it's also a lot of fun because i can toy with creating my own platform and depending on the compatibility i choose to opt into -- ie Mastodon/Lemmy/Kbin -- i get interop and i don't create a silo. I absolutely love that.
I suspect there's tons of issues with ActivityPub, ranging from inefficiency of synchronization to discovery of instances, but those are things we can evolve for ActivityPub.
But i absolutely adore that creating a new platform with unique features can still be compatible with the larger ecosystem. It's like creating a new Reddit client (Apollo/etc) but it can have new features like shared economy or integration to PixelFeed/Mastodon/etc.
Despite not loving some things i'm about inefficiencies ActivityPub, i adore what it's giving us. I hope we run with it.
> I don't want the very nasty users of that 10M to interact with me or my friends.
> This is just opening everyone to harassment we didn't ask for.
I fail to see how this is much different from existing social media, where there are also any of millions of users who could harass you or your friends.
With all due respect, I'm going to nitpick a portion of your comment there... While the 163k users might accurately describe pixelfed...that is, pixelfed users, that doesn't even begin to describe "the network"... which is composed of millions of fediverse users. Your comment was like saying that email is not that big of a network because there are only X numbers of thunderbird users. Again, all due respect. But pixelfed is only one of many ways to be present on, and interact with millions of people on the fediverse. ;-)
And if that goes too deep too soon, then the home page of the same site can help provide a brief primer. Of course with the fediverse gaining recent attention (due to mastodon and pixelfed), there are plenty of other guides out there as well. Good luck!
> Did anyone try it out already? I'd love to hear more about how it worked it practice.
Just tried it but it's not a direct import - it's "dump your data from Instagram as JSON[1] and we'll import that JSON". i.e. a one-off backup restore.
(If ActivityPub allowed backdated posts, I'd do the sync myself from my instaloader downloads but alas, most servers do not.)
[1] "It may take up to 14 days to collect this information and send it to you."
It looks like the instagram->pixelfed import actually preserves the original post's date. Dansup imported a post from 2016 and it shows up as "7y ago" in my mastodon UI.
Oh yeah, if you have direct access to the database / backend admin (as I'm sure this importer does), you can frob things to get the correct dates and avoid pushing them out as new posts. Just not possible via the standard APIs at the moment (I had to patch my Akkoma to allow backfilling of posts from certain accounts.)
Got my download - 2.4GB. Uploaded it - Safari chugs badly here.
If you leave that tab at all, it breaks and restarts which cancels the import process.
Once it's uploaded the file, Pixelfed wants you to tap each image you want to import in a 3-wide scrolling grid and there's no "select all" I can see. Plus it is now DOG SLOW.
I have 9052 Instagram posts. Bugger that, frankly, because I'm not spending upwards of 3 hours clicking on a web page to import photos.
In summary - nice idea badly implemented.
[edit: Tried it in Chrome. Upload is much smoother and more reliable but still no "select all" and it still chugs like a 1970s PC - a good 30s between clicking and the image being selected.]
Apologies, should have been clearer that it's the MastoAPI which has no concept of backdated posts - or even of creating dated posts at all. [1] has no `created_at` or any other (useful) timestamp. Hence having to local-patch my Akkoma to allow passing `created_at` for my backfill needs.
Trying it now...it will only let you import 100 photos at a time, and you need to single select each one (My Instagram has hundreds and hundreds of photos going back a decade).
So far nothing has actually imported. Not sure how long I should wait.
> it will only let you import 100 photos at a time
Oh bloody hell. I've got no hope with my 9k photos then. I appreciate the effort from Pixelfed but they need to clarify their messaging from "Import From Instagram" to "Partial Restore From Instagram With A Terrible UX".
My worry is that it is expensive to host and serve images with decent speed. I wouldn't have pixelfed as my main image sharing platform, purely because you'll need quite some money to keep it running, and without monetization - considering a lot of people just wanna post and not pay - not sure how it will survive in the long term
I tried it. You have to click on each photo you want to import. Maximum 100 photos, but you can "re-upload" the zip file and click on another batch of 100 photos. Then they are processing for a long time (which is understandable). The next morning — nothing happened, import disappeared, photos didn't appear.
Well, I tried building pixelfed and ran into problems. I tried to follow reported, docker related issues on their github but without much luck. There was a momentum and even PR to create an image but it somewhat got stuck.
tl,dr; not skilled enough in this particular toolstack to get it working.
> given that nobody has posted photos, captions and hashtags to their feed in 2 years
Well, let me throw my anecdata hat into this ring and point out that it is not "nobody" because I have a constantly refreshing feed of people posting just those things every day.
He may be exagerating, but it is also my experience. The only "posts" i see are from entities that have something to sell. The "regular" users just do stories now (probably because of the disappearing nature of them). 2 anecdata, i guess.
Depends on the generation, but it's "not cool" to post photos like we used to do it every week or so in my generation (younger millennials / older gen-z). I've seen some people doing once every month, quarter, or life altering events to give a big "fan out" news to their friends, but that's it.
When it comes down to it, there are 3 key issues with the "traditional" model of social products:
You are the product - This is true of 1) your data you put into the system explicitly, 2) the data you create by using the system, and 3) the predominant business model (ads) of these products. This could be fixed by:
1) some easy means to extract all data you create in an accessible format (only some users will care about this)
2 & 3) Allowing people the choice of whether they are the product or not. "Subscribe for no ads" seems like a good middle path alternative to selling ads without demanding a whole new architecture and operational model for federated solutions.
Separate of the above ideas is the fact that social products are really addiction mechanisms. Reddit/Hacker News/Tik Tok/Instagram are all heavily optimized to addict their users to their product. This is, in my opinion, where the real risk lies overall, because it is directly leading to aggressive micro feedback loops and information bubbles that are evolving faster than humans can process. I'm not sure what to do about this problem save for outright control/limitation attempts, which seems like a guaranteed mess.