Armin Ronacher claims "Scaling Mastodon in Impossible," a claim which may well be true, given how one defines "scaling" (or "impossible").

https://lucumr.pocoo.org/2022/11/14/scaling-mastodon/

But his primary claim, in big letters, is "Decentralization is a Questionable Goal," which seems a pretty bold claim to make during the period of time in which American oligarchs are demonstrating why centralization, using Facebook and Twitter as example, is a goal for which the question has been answered in the negative. Let's try and follow the logic.

> Decentralization promotes an utopian view of the world that I belief fails to address actual real problems in practice.

I'm not sure how decentralization does that. If I believed in a utopia, I would assume that the issues at Facebook and Twitter would work themselves, or never occur in the first place. We're far from a utopia, which is something the writer goes on to acknowledge two sentences later after a contradictory footnote.

> All of these things [including mastodon] have one thing in common: distrust.

So is it utopian, or dystopian? Distrust isn't something I associate with utopia! Also, I skipped over a fascinating footnote, which includes:

> Traditional banks are also decentralized, but they follow shared rules. I can send from my Austrian bank to a bank in Estonia and it will work. The tech behind the scenes is not even all that terrible. It does not really look like a decentralized thing because there is a lot of regulation and you can't just start a bank, but it would be hard to argue that it's not decentralized.

So far in this essay questioning decentralization, we've seen a reversal of utopian and dystopian, plus a great example of decentralization working very well. Without even mentioning telephones or email yet!

Next we get to the meat of his argument, I hope, as the author explains that decentralized package management has been deprecated in the Python programming language in favor of a centralized package management system called PyPI, and makes an isolated reference to NPM to remind us that Node.js has done the same thing. This isn't described as an unmitigated good, and the author is at pains to highlight past complaints about PyPI, and doesn't even mention NPM's infamous left-pad incident, and in fact both PyPI and NPM are often brought up as example of how centralization has hurt both language ecosystems.

Yes, trade-offs aside, there are definitively advantages and efficiencies to centralization. So far no oligarchs have set their sights on controlling Python or Node.js packagement management, so perhaps the biggest issues haven't surfaced yet, but Twitter isn't in that enviable position.

The author seems to understand that they are on very shaky ground at this point, as their several feints against decentralization seem to keep circling back around to praising decentralization, so we shift tactics:

> Mastodon encourages not just decentralization, but federation. You can pick your own mastodon server but you can also communicate with people on other instances. I will make the point that this is the root of the issue here.

Mastodon is brittle. ActivityPub (the protocol which underlies mastodon) is messy.

> The thing does not scale to the number of users it currently has and there is probably no trivial way to fix it up.

If I'd invested a dollar into $TWTR every time someone said something like that during the four years or so before Twitter was moved from Ruby to Java, I'd have bought many, many shares and then cashed them in for $54.20 per share a few weeks ago! And yes, mastodon is written in Ruby also, so it seems likely that a rewrite in something else--a non-trivial task--will be necessary. In fact, mastodon is not the only ActivityPub implementation, and other compatible software already exists written in things that are not Ruby. I'll keep refering to mastodon in this article, but please substitute any other ActivityPub software in its place.

But, the author points out, the real issue is that we can't even universally agree on what things should look like! Some people want huge servers, other people wish everybody would run their own servers-for-one, some servers are designed for trolling and chaos, while others are so tightly managed that they block servers that don't sufficiently bend the knee to their judgement about third parties.

It's true! This "wild wild west" approach that is inherent to mastodon is very different, and is possibly one of the weakest points when compared to a centralized service. I think most people would say it's dramatically outweighed by the stronger points, but if you're the sort of person who assigns zero points to the fact that Facebook and Twitter are now controlly eclusively by two American oligarchs, I guess you're going to focus on moderation challenges more.

I'm on that qoto.org instance he mentions is still blocked by some number of servers. I'll link to the server owner's explainer for why: https://qoto.org/@freemo/109319817943835261

The author now suggests we are unlikely to agree, but there is another issue: Unpaid Labour and Opsec. While Twitter and Facebook have lawyers and can buy and sell politicians, mastodon server owners don't. They give the example of a large company with 140k followers on Twitter joining a small instance, and... wait, why would that ever happen?

It seems like some of these problems are solved by others. Why would a large publicly-traded company ever join an independent mastodon instance in the first place? They don't user elilillyceo@hotmail.com for email, do they? They've run into trouble recently by trusting Twitter.com to protect their brand identity, which turned out to be a mistake on that centralized service. The popular story is that someone paid $8 to get a blue check put on their own Twitter account, impersonated Eli Lilly to highlight the egregiously-high price they charge for insulin, and Eli Lilly lost $16 billion in the stock market as a result. I think the entire pharmaceutical industry took a huge hit around the same time, so I'm not convinced, but it seems to me that decentralization solves both the imaginary scenario the author outlines and the real-life scenario with the centralized Twitter at the same time. The Eli Lilly company controls the lilly.com domain, which they already use for email. Emails that come from elilillyceo@yahoo.com are obviously fake, while emails that come from d.ricks@lilly.com should probably be believed. Similarly, it turns out that tweets from @EliLillyandCo@twitter.com were not to be believed, despite the blue check, but mastodon posts from @pr@lilly.com are clearly official, since they come from the official Eli Lilly domain.

I believe all companies that run their own email servers today should also be running their own mastodon servers.

Any companies paying another company to provide their email service (but using their own domain name) today should also be paying another company to provide their mastodon service (but using their own domain name).

That solves several problems at once, and seems obvious.

The author goes on to suggest they have had very-private conversations on Twitter, which... don't do that! If the conversations are very private, use Signal. Neither Twitter nor any mastodon server is private enough for private messages.

The author is not wrong that mastodon faces challenges. They are only wrong in ignoring the even bigger challenges Twitter faced in the past and present and future. Will things be rough going forward for mastodon? Undoubtedly! Will there be a future mass-migration away from Ruby toward something more efficient, and maybe an incompatible version of ActivityPub with beaking changes, and other assorted technicaly challenges? Probably! But scaling mastodon is no more impossible than scaling Twitter was, it's just difficult. Decentralization is not a questionable goal, it's the only option when oligarchs run roughshod on the world stage. The future is going to be interesting, and the present is wild.

I, for one, am here for it.