As a regular critic of the #Bluesky "decentralized" baloney, been trying not to pile on as yesterday's near total outage makes it clear that it is not, however there is one aspect of the reporting that is confounding.

Supposedly, the outage was due to a DDOS attack on the company's PDSs, and the few users running their own PDSs were not affected. (https://techcrunch.com/2025/04/24/wait-how-did-a-decentralized-service-like-bluesky-go-down/). Does this make any sense?

1/

Wait, how did a decentralized service like Bluesky go down? | TechCrunch

It turns out that decentralized social networks can go down, too. On Thursday evening, the decentralized social network Bluesky experienced a significant

TechCrunch

Apparently #Bluesky has one centralized service, the "relay", which did not go down, but it has distributed its users across multiple PDSs (and users can roll their own PDS). The idea is that any individual PDS can go down and so long as the relay keeps working all the other users are OK.

But, ALL of the company's PDSs went down. 🤔

We are being told that a DDOS attack targeted ALL of #Bluesky's PDSs and they all went down simultaneously. Suppose this is possible, but it does seem odd.

Well, here is a little more information on the outage:

Beebom | Bluesky Suffers Hour-Long Outage Despite Decentralized Design

https://beebom.com/decentralized-bluesky-suffers-outage-now-back-up/

6:55 PM "We are investigating a major outage with Bluesky hosted PDS instances.”

7:38 PM "We have identified a likely root cause and are rolling out a fix to the Bluesky PDS fleet”

Suppose this could be consistent with a simultaneously DDOS attack on all their PDSs, but it seems more likely that it has something to do with the "PDS fleet."

Bluesky Suffers Hour-Long Outage Despite Decentralized Design

Bluesky went through a serious outage on Thursday leaving users unable to access the platform. Read this story to learn more.

Beebom

As @thenexusofprivacy points out below, multiple things can be true. Perhaps the PDSs in "the PDS fleet" have some interdependency or share a computing resource that was overloaded by a DDOS attack on one or more of them.

It could be they are all completely independent and were all attacked simultaneously and the fleet all went down at the same time, but it seems like there is more to this story. It would be good to get more details from the company.

Two more tech media articles that provide a little more information and also report the outage related to something affecting all #Bluesky PDSs, but no real specifics.

https://winbuzzer.com/2025/04/26/blueskys-decentralized-network-faces-major-outage-caused-by-centralized-servers-xcxwbn/

https://www.perplexity.ai/page/bluesky-suffers-outage-despite-b5tuvcMDT.6SI7QyiCS5EA

For anyone interested in the specific communications from #Bluesky CTO Paul Frazee on this (https://bsky.app/profile/pfrazee.com/post/3lnnfwf3tns22).

7:15am: All of our PDSes got DDOSed. Any PDS that wasn't stayed online. If the same happened to AP, they would've had the same outage

1/

Paul Frazee (@pfrazee.com)

1. Bluesky’s about interop and scale, mastos about topological distribution, blockchains are about censorship resistance. Decentralization means different things 2. All of our PDSes got DDOSed. Any PDS that wasn’t stayed online. If the same happened to AP, they would’ve had the same outage

Bluesky Social

#Bluesky CTO continued:

Frazee: "ActivityPub doesn't scale correctly for the social media usecase anyway so this is a pretty pointless debate."

To which ActivityPub developer Christine Lemmer-Webber (here @cwebber) replies "hm"

2/

#Bluesky CTO continued:

A user asks about the source of the DDoS attack, and Frazee responds that they think it was an "accident by a 3p".

So, presumably not malicious.

"3p" apparently means third party.

So, it seems like #Bluesky is saying a third party, whatever that is, accidently did something that DDoSed all Bluesky PDSs and caused them all to fail simultaneously.

This is still confusing, and it would be good to get more details.

3/3

@mastodonmigration Bluesky has always been decentralwashing their centralized product. Now their trying to decentralwash their central outage.

Their right in one thing: The discussion which protocol better scales for decentralized social media is pointless, as only one of the two protocols is used that way.

@mastodonmigration Yeah, I'm very curious too to hear what exactly happened, I've heard some conflicting versions - for now Paul says that "I can't talk about in detail since we're still resolving the problem"

@mastodonmigration @cwebber It goes well beyond that.

Moderation lists that are designed to cut down on targeted minorities gets suspended, while people who engage in targeted of said minorities pretty much get away scott free on there.