Protocols, not platforms: A technological approach to free speech (2019) | Hacker News Reader

Protocols, not platforms: A technological approach to free speech (2019)

knightcolumbia.org

95 pointsuberdru38 comments

I think this is largely a futile effort because the equilibrium the author thinks of when he talks about protocols does not exist.

All existing internet platforms already sit on top of protocols, the internet's running on them. Anyone can built a platform on top of ActivityPub, on whatever crypto network or what have you.

The old internet consisted of smaller islands not because of protocol technology, but because it had a small userbase overall and discovery was bad. The benefits of aggregation hadn't kicked in. If Mastodon gets to hundreds of millions of users I will bet that one or two instances will grow very large, develop their own rules, and what do you have then? A proprietary platform on top of what was once a 'first order' protocol. The protocol is still there but doesn't matter for end users, same way HTML doesn't to your average Facebook user or the blockchain doesn't matter to your average guy on coinbase.

People need to recognize that anything that increases connectivity between people tends towards agglomeration. Tech is inherently centralizing, even ironically enough the superficially decentralizing tools, if they lower the cost of communication. This isn't just the history of internet companies, it's the history of movement of peoples from fractured townships to large urban centers. It's not malicious platform lock-in, but simply gravity.

This is very true and I think it applies to everything, not just software. I was an anarchist at heart until I understood that if you just remove all the rules they all just come back with different names, and often in an inferior form.

So maybe we need the opposite, a completely authoritarian and centralized way of handling internet services. Imagine if there was a single mail service, provider and client, handling all possible uses cases, a single web browser, etc.

Its like the complete opposite end, instead of using a standard protocol to implement your use case, simply improve on _the_ service.

[Edit]: Oh, and by 'centralized' I don't mean BigCo is in charge of it and manage the servers for it. What I mean is Public Organization 'owns' the service, but this service could very well be self-hosted by users, decentralized, etc. It's just that it is 'forbidden' that any other organization works or develops a similar service.

IIRC that was how the UK phone network used to be. Also the USA, if I remember correctly that acoustic couplers predated plug-in modems because users were forbidden from connecting anything other than a phone to the phone network.

> If Mastodon gets to hundreds of millions of users I will bet that one or two instances will grow very large, develop their own rules, and what do you have then?

Are you saying that these large instances would refuse to interoperate with each other? There are already signs of that happening even among small instances, but people are able to move their accounts from instances that are blocked by (or block) too many of their friend's instances, onto instances that have moderation policies they are more comfortable with.

Better yet, because all account actions are covered by an open API, it would be possible to create an account on a service that lets you make posts on multiple sub-accounts on different instances, and aggregate messages from your friends across them, which should heal any fracturing of the network for people who really have friends who (are in communities which) hate each other.

> Are you saying that these large instances would refuse to interoperate with each other?

Likely. It's the same thing that happened to the big XMPP instances.

There are other ways that open systems turn closed. I'm not sure which one would happen to the Fediverse, but given how often it has happened, and the variety of proximate causes and exact end-states, I'm betting my money on the Fediverse turning closed somehow.

* The XMPP path where the biggest instances just decide they aren't going to federate with anyone else any more.

* The SMTP path where the big instances gradually accumulate anti-abuse mechanisms that make the barrier to entry too high for anyone who requires reliable delivery and isn't one of them.

* The IRC path where the network fractures into centrally-controlled coalitions that federate with each other but not with anyone outside the coalition.

* The NNTP path where the cost of running a server kept going up, the number of servers kept going down, and now if you want access to the entire network, you have to pay a subscription fee to one of three companies.

* The Git path where a single instance bundles value-adds that lead to most users just using the one central service.

The mastodon community of today is very different from what it would be if it had hundreds of millions of users. Once the plaform grows enough to critical mass, the costs of switching off that platform become high enough that it can close itself off without mass exodus.

The only way I see to end up with anything like what the author lays out is to mandate interoperability for platforms (possibly over a certain size). Unfortunately I have zero faith that our legistlators can craft and pass a bill that does that without bastardizing it to the point that is simply creates moats for the incumbent tech giants. I'd love to be surprised.

> mandate interoperability for platforms (possibly over a certain size)

The EU came really close to doing that with its new Digital Markets Act, but it seems that the interoperability requirement will (rather bizarrely) only apply to messaging apps and not social media platforms.

https://www.eff.org/deeplinks/2022/04/eu-digital-markets-act...

Mastodon is not built for a single community comprising "hundreds of millions of users". Individual users and communities will interact on a far smaller scale, much as they might with email.

While gmail never reached critical mass and completely closed off, it's anti spam efforts (among those of others) have made running a small mail server fairly impractical.

Ending up like email would probably be the best case result for mastodon. It is interesting to think about what stuctural aspects of email prevented hotmail or gmail from reaching critical mass and being able to fully close off interoperability.

Yes, and I don't think the author fully grasps the role of capital: you can create all the egalitarian protocols you want, but if there are massive wealth inequalities, capital will ultimately capture whatever communication channels are built on top of them.

But not all, unlike communism. That is where you can still have phoenix. Whilst rich man eating Twitter, people can revolt.

Given option out is key, not what is in.

> It's not malicious platform lock-in, but simply gravity.

One effect doesn't rule out the other.

Platforms are definitely working towards locking customers in. The whole focus on engagement for example, prioritising time spent on a platform (and by extension the ad revenue) over the qualify of that time..

Users don't even venture out of their bubbles anymore because they don't know how. This isn't all natural discovery and network effects.

Users don't even venture out of their bubbles anymore because they don't know how.

Added to https://github.com/globalcitizen/taoup

I don't see how this approach stops a centralized service from taking over via the standard approach of Embrace, Extend, Extinguish [1]:

- Embrace: BigTech implements the protocol. For example, say text messaging.

- Extend: BigTech implements some custom additions to the protocol, that only work for their service. For example, add support for sending photos, but only ones stored on BigTech's servers.

- Extinguish: Once people get hooked on your custom features, make it hard for competitors to replicate. For example, stop competitors from being able to load photos from BigTech's servers.

1: https://en.wikipedia.org/wiki/Embrace,_extend,_and_extinguis...

The most successful example of EEE is not Microsoft but rather Linux. It embraced the Unix* style, extended it with its own stuff and now all non Linux unices are basically an afterthought

It doesn't really undermine your point, but, well ... iOS/macOS/watchOS/tvOS/etcOS is hardly an afterthought, and they are all non-Linux Unix systems.

We've already seen this happen with Google and browsers. In this case, they extend so quickly and also manage to spread much propaganda about how their way is better, that they're basically using change as their weapon.

Protocols aren't the solution when people are so easily manipulated. Perhaps stagnation is the solution. Something that stays static for a longer time is also going to have more implementations.

It depends on who cares, and how much. The world isn't one-size-fits-all, and if 90+% of internet users are happy with a platform that arbitrarily bans users or censors content as long as it supports the marquee tag, that's their business. Conversely, I could see using a system that only gets me a very limited amount of content as long as I consider it to be high quality, and if my friends' ex-wives or people I haven't seen since high school and wasn't close to then never use the system, well, I think I can handle that.

[2] is the step that can be addressed.

Extend? How do you address it?

"Protocol extensions must be able to be implemented by all users of the protocol and may not require a particular providers service but allow any alternative so long as it is protocol compatible"

Basically force those extensions to be equally cross-compatible.

Two key issues with a rule like that:

1. What forces me to comply with this rule?

2. What does "able to be implemented" mean? In theory, you can re-implement my entire codebase. In practice, you'll find it very difficult to maintain feature parity with me if I repeatedly extend the protocol faster than you can implement it. Especially if I don't publish my code or document the extension or if I simply have more resources available to me (either in terms of server-bandwidth or programmers).

Though protocols are not law. You can have non-conforming implementations. If these are popular enough (potentially due to $proprietary_thing) then people will just use them. Other services will need to decide if they want to continue to interoperate.

(2019). Previously discussed at:

https://news.ycombinator.com/item?id=25942632

https://news.ycombinator.com/item?id=20841059

Thanks! Macroexpanded:

Protocols, Not Platforms: A Technological Approach to Free Speech (2019) - https://news.ycombinator.com/item?id=25942632 - Jan 2021 (269 comments)

Protocols, Not Platforms: A Technological Approach to Free Speech - https://news.ycombinator.com/item?id=20841059 - Aug 2019 (67 comments)

> Rather than relying on a few giant platforms to police speech online, there could be widespread competition, in which anyone could design their own interfaces, filters, and additional services, allowing whichever ones work best to succeed, without having to resort to outright censorship for certain voices. It would allow end users to determine their own tolerances for different types of speech but make it much easier for most people to avoid the most problematic speech, without silencing anyone entirely or having the platforms themselves make the decisions about who is allowed to speak.

Most platforms have APIs that allow people to "design their own interfaces, filters, and additional services". Many also allow people to build their own platforms on top of them - subreddits, discord servers, facebook groups, etc. that they can moderate as they please. It would also be possible to build browser extensions that automatically block certain speech. But this is all besides that point, because this debate is not about letting people "determine their own tolerances for different types of speech", it's about preventing other people, who are willing to tolerate certain types of speech from doing so. This is the reason subreddits, discord servers and facebook groups get banned. The people who joined them have decided that they are willing to tolerate certain types of speech, but others are not.

Disagreements with this article via a claim that says technology inherently centralizes, to me, seems flawed.

The concept of democracy was evolved as a system that could distribute social power in order to catalyze the wider involvement of its members in their own social and economic concerns.

All the way back to Alexis De Tocqueville in the 1800s (in his book Democracy in America),

Tocqueville attributes the ultimate success of liberty in America, however, to a feature of American political order he refers to as “administrative decentralization,” which–going beyond federalism as such–fosters the participation of citizens in “real . . . political life” beginning at the local (sub-state) level. [1]

To me, a series of decentralized, shared ownership systems that come together to provide and replace the functionality that has today been built into centralized networks is the only way to achieve any kind of "liberty" (however one may define that word).

The missing link for these systems has historically been cost - who runs the servers that runs these systems, and who pays to develop them? It is exactly for this reason that they were centralized in the first place.

Today, we've got one way to solve for these problems (tokens). I'm sure there will be more. But they will all be catalysts towards this direction.

Edit: To comments talking about how protocols won't just be replaced by big companies.

All companies ultimately care about one thing -- distribution. They will continue to be the direct link between customers / users and the service they provide. The difference is that they will need to build upon the protocol that has the most distribution.

If we imagine a new open protocol for "online document comments" that existed, Google Docs AND Office online could use it and have interoperable commenting from each other's identity systems vs the state of the landscape today. what would their motivation be to do it? Same as using open source libraries like ffmpeg or openSSL -- except this time, they don't pay to host the APIs that power this protocol.

[1] https://www.catholicculture.org/culture/library/view.cfm?rec...

> The concept of democracy was evolved as a system that could distribute social power in order to catalyze the wider involvement of its members in their own social and economic concerns.

Even in democracy there is a tendency to centralize. For example, there are big differences between ancient Athens, modern Switzerland, the USA, and the EU.

They all are implementations of democracy, but as democracies have grown larger, it becomes more centralized.

> Democracy in America

Relatively speaking. Written by a French aristocrat.

Just out of interest: social history wise, for those more familiar with European modern history, what kind of wine would they have been drinking back then?

Hiding abuse from the end user only stops direct forms of abuse, but it won't stop all abuse. Revenge porn is a classic example because that is usually directed at everyone in a target's life without needing to reach the target themselves.

Imagine a scenario in which the target uses a client that aggressively hides content. However since this system is protocol based, other users would be using clients that aren't as aggressive. What happens if an abuser responds to every post their target makes with a revenge porn image? Maybe the victim of the abuse never sees that image because of their clients settings, but numerous people following them do. Isn't that still abuse even if the target doesn't see it? How do you stop this type of indirect abusive behavior without some centralized authority?

At a certain point we need to remember that the legal system exists.

Clearly a sizable audience wants protections above and beyond what the law requires or else there wouldn't be such strong support for these systems curtailing abuse. Saying the legal system will handle it is not a neutral position. It is a clear decision in favor of the anti-censorship side over the anti-abuse side.

Revenge porn is illegal, at least in the UK.

That's my point. Don't ask the world to do the courts' job.

Protocols, sure, but then you still need to have HW running them, which enables platform building. Also, that HW (and bandwidth) has significant, non-zero costs, which again tilts the field in favor of those with the capital to pay for it, and they'll want to monetize their investment, which again means platform building.

And coming from the other end…

“Protocols, not APIs”

Reply on news.ycombinator.com