Kurt
1/
DECENTRALIZE YOURSELF.
(Just not in the way Fediverse federation advocates are suggesting.)
Over all of the ecosystem promising an open social web, with decentralization and federation designed to protect user control, falls the shadow of reality.
(Long post, or read in one page on Medium via link)
https://medium.com/@flglchicago/decentralize-yourself-0d904cf74beb
Kurt
10/
Miraculously, somehow, a couple of days later, the restriction on the Meta side seemed to be resolved: whatever spam or malicious content flag Meta had placed on mas.to was removed, and although scrubbed content wasn’t restored, new posts could be made.
Tim Chambers
@FLGLchicago So this temporary nature of this would seem to imply it was what it seemed to be - a temporary spam flag based on the URL, versus some specfic vendetta to punish admins of the mas.to site, no?
Kurt
8/
So I took everything down from Threads and posted about the issue on Mastodon instead, exact same posts, exact same screenshots. Fediverse advocates for Threads federation (including a couple of Mastodon server admins) told me that I was imagining things, that they saw no evidence of any antisocial behavior like that...
Kurt
9/
...and had never seen any problems. I doubted my sanity for a few moments until some other users started sharing their identical experiences, mostly about sharing mas.to addresses or content but also some users on other smaller Mastodon domains.
Kurt
5/
Both the intent and the timing of the deletion seemed odd to me until I looked up mas.to’s list of suspended domains and saw that the admin had decided pre-emptively not to federate with Meta domains. There didn’t seem to be problems with other large Mastodon instances who hadn’t placed Threads in a suspended state, so...
Kurt
6/
...I made a series of posts on Threads outlining what happened, along with screenshots, making what seemed to me to be a reasonable observation or assumption that Meta was taking retaliatory action to use as leverage to coerce cooperation by essentially wiping mas.to from the Meta universe, true to the form by which the industry has known them to operate for years.
Kurt
7/
Those posts themselves were immediately marked as spam too, invisible on anyone else’s timeline, invisible on the new Federated Threads tag, dumped into the spam bucket when I tried to use one as a comment on someone else’s post to get some visibility. At the same time it was impossible to post anything from mas.to on Threads — sharing a post, or even just a link to the front door of the site — everything was blocked at the composition box.
Kurt
2/
There’s been so much enthusiasm this past week to federate the non-commercial Fediverse with products from large corporations like Meta — sounds great on the surface, all your stuff in one place! In the rush, however, to connect the world and bring everyone closer together (sound familiar?), the practical outcome serves to re-centralize social media activity under an illusion of convenience that is quickly overridden by vulnerability and threat.
Kurt
3/
Is this really necessary, and is it what we truly want? Here’s one real-world early-stage cautionary-tale example of why not, why we should be careful to decentralize in a different way.
Two days before the public announcement that Threads would be testing the ActivityPub protocol with a limited Mastodon presence, I received a notice from Facebook that a post I had made about Mastodon more than a year prior had been taken down for violating community standards as spam.
Kurt
4/
The post was completely innocuous —I was letting my Facebook friends know that I had left Twitter after 15 years, that I was happily surprised that there had been so many improvements in the Fediverse since I had last had an account, and I shared the coordinates of my fresh new mas.to address and invited them to look for me there. Brand new day.
Tim Chambers
@FLGLchicago We see no signs of this occurring thus far. Time will tell, but if I were Meta, "conditional Federation" is the LAST thing I would waste my time on for the 20,000 individual fediverse servers and growing. And I cannot imagine any server-specific thing admins would need to do to be "compliant with" or in any sense any way we even could "adopt" Meta's security systems. Nor would we. Again we will see, but this seems highly unlikely to my eyes.
Kurt
14/
Further, and this is projecting out a bit in time but not possibility, suppose Meta requires as a condition of federation with their products that instances of all types either provide Meta with their own validated signals, or inevitably requires instances to adopt Meta’s automated moderation and spam system.
Kurt
11/
Minor incident, right? Possibly a misunderstanding, blip in the spam filters, easily remedied. Maybe Threads just backed off from a petulant business policy. But, and here’s where it really becomes problematic: a week later, the Threads account from which I made the posts is itself now, and still, marked as spam and/or malicious content, and because of that the account is essentially unusable.
Kurt
12/
Thankfully it is a new and small profile I had started from scratch without having an Instagram base for that username, and still I had been slowly building followers and getting engagement until I had made those posts about mas.to. After that, and through today, nothing: no new followers, no comments, no likes, complete silence. When I went in to check from another profile, every single outbound post and comment was invisible.
Kurt
13/
Now, I can always go back and start over on Threads. The account was an alternative presence and I really wasn’t that far into it—but what if it was large and longstanding? And suppose Threads had been fully federated, and had not only filtered my account on their own Meta services, including “accounts by the same user,” but had sent signals to other services and domains in the Fediverse that they considered my account to be a spam account, and any account linked to it to be a spam account.
Kurt
15/
I say inevitably because machined intelligence layers are the only way Meta can moderate at scale, and beyond simple signals they are likely, if they haven’t already, to develop a reputation scoring framework like most of the other platforms are doing. Say hello to Black Mirror Season 3 Episode 1, “Nosedive.” (Watch on Netflix or look up the trailer on YouTube.)
Tim Chambers
@FLGLchicago LIke you saw, they already track server URLs for spam. Which in your instance may have been a false positive they corrected. They would be responsible actors if they looked at or creating a server block lists akin to how responsible moderators of each server on the Fedi do. https://writer.oliphant.social/oliphant/the-oliphant-social-blocklist
Tim Chambers
@FLGLchicago Again: I don't see any "validated signals" that ActivityPub would allow us to give if we even wanted to, which we wouldn't. And there is no
"automated moderation" that Fediverse servers even could implement. The protocol just doesn't allow for such. The protocol itself was written to NOT trust other servers and to only give minimal data.
Kurt
@tchambers If that’s the case, I don’t see how Meta would be able to fully federate given some of the business requirements I mentioned in my post. Then a loose association to the Fediverse, for Meta, becomes a regulatory out as long as they can push undesirable accounts on someone else.
Tim Chambers
@FLGLchicago What seems a cconflict there? They could still take in Fediverse content, like any other server does. And they are free to moderate said Federated in content ON THEIR side. But not on ours. And not impacting us.
Tim Chambers
@FLGLchicago Pretty sure everything you are thinking of only applies on the Threads side, not ours. They don't NEED it to apply to our side of the fence. And any non-standard activtypub shenanigans would immediately be rejected by the rest of the AcitvityPub dev and admins space.
Kurt
@tchambers Right, so they will turn it on its head: Fediverse has no systematic approach to this, so we have no choice but to require ours. If ActivityPub needs to adjust protocol to accommodate, we’ll wait, forever, as long as we have air cover for regulatory challenges.
Tim Chambers
@FLGLchicago I'd suspect both/a and techniques. Some known block lists of servers (like all responsible masto admins do) and then they moderate all federated in posts just like they moderate all native all native threads posts. Using the same tools they do for all native threads posts.
But none of that touches or impacts our Fediverse side of the fence.
Kurt
@tchambers I don’t see how they take in Fediverse content and place it into an environment advertisers would be comfortable in. Brand safety requirements alone are onerous. If Threads cannot evaluate individual accounts or posts in real time, they will have to block domains wholesale.
Tim Chambers
@FLGLchicago They are already applauding Flipboard's entry into the Fediverse and saying they look forward to interoperating. Time will tell though. All just words until it happens.
Kurt
@tchambers Guess we’re all going to find out if there is a fence in their minds.
Kurt
@tchambers So, even if Meta chooses not to try to impose moderation standards on Mastodon servers, or it’s not possible, and even if Mastodon admins wouldn’t comply anyway, do you really think for a moment that Meta will federate with Tumblr, or Flipboard, or any other former or semi or current commercial platform, without imposing requirements on them? Center of influence shifts quickly and entirely from the Fediverse as it is currently known, to a new universe with Meta at the absolute center.
GNU social JP is a social network, courtesy of GNU social JP管理人. It runs on GNU social, version 2.0.2-dev, available under the GNU Affero General Public License.
All GNU social JP content and data are available under the Creative Commons Attribution 3.0 license.