Mike Macgirvin, on cross-platform federation and protocol standardization

Some people here may not always agree with what Mike has to say, but he raises a number of very valid points here.

Mike Macgirvin wrote the following post 4 hours ago

I hope we don't have to rehash the last decade of the state of the federated social web and standardisation efforts and specifically how that relates to hubzilla. I know quite a lot about federating the social web. Been there, done that. Yes I know exactly how to federate with GNU-social. I know exactly what it would involve and how long it would take and what bits would work and what bits will never work because the projects are fundamentally incompatible. In any event, no - I'm not doing it again. I refuse. Full stop.

If you want to federate project 'x' with project 'y':

Step 1.

Obtain an account and hopefully code access to both projects. Study the projects and learn their capabilities and integration points. Make a list of things that are likely to be a problem.

Cross network privacy issues are often show-stoppers unless you can convince one of the projects to change their privacy implementation; and this is rarely possible. So you must come up with some convoluted workarounds so that privacy is not leaked through the less private project and so there is also no significant loss of functionality on the more private project.

Step 2.

Roll up your sleeves and get to work.

----

That's it.

All other approaches have traditionally failed.

Repeat this process for every project you wish to federate with.

Sure, you can rely on a committee to define how best to do these things, but you still need to perform these steps in the proper order. You cannot change this.

Note: after ten years of trying to create a federated social web protocol through various consortiums, they're still stuck trying to invent a "universal RSS feed". Nobody in these committees are seriously looking at privacy and spam (the killer issues) and how to provide a universal privacy policy control mechanism. Many do not even understand what privacy is. (Hint: it is not anonymity). And if you mention this topic, it starts a massive war amongst implementers and everybody gets mad and goes away for another six months. Then they start over again defining the RSS feed. We've redefined the feed format a number of times now.

It can be achieved. Stop worrying about the danged protocols. See my previous post and follow the steps. If 10-20 people out of the 7 billion people on the planet did this we'd have a federated social web. Once these 10-20 people get together and discuss how to reduce the pain and how to resolve the privacy mish-mash between projects we'll figure out a protocol.

The problem is they're doing it backward - trying to define a protocol when they haven't even identified what it's supposed to do (except to describe the content of social messages). That's the easy part - always has been. Don't even worry about the format/content of the messages. We can solve that. Sending messages back and forth is also a no-brainer. We've been doing that 40+ years.

Privacy and spam policy differences are the hard issues - every thing else is just a smoke screen. But you won't know that these are the hard issues unless you actually try to federate two services that implement them differently. That's why I specifically mentioned

All other approaches have traditionally failed.

You also need to be aware that all these federation efforts invite the large players to the table. The large players are sitting at a table discussing how to federate different services with one hand whilst the other hand is trying to build a walled garden and lock their communities into their silos. So is it at all surprising that progress is slow? Don't take my word for it. Read the author names on the relevant spec documents. See who employs them. Here's ActivityStreams:

Activity Streams Working Group
J. Snell
IBM
M. Atkins
SAY Media
W. Norris
Google
C. Messina
Citizen Agency, Google
M. Wilkinson
MySpace, Facebook, VMware
R. Dolin
Microsoft

Notice who you don't see: Diaspora, Redmatrix, Friendica, GNU-social, and Pump.io