Hacker News Reader

An excellent counterargument. Yes, one might concede that protocols have not and probably will not win out over platforms in a free market. But if we're advocating that the big tech utilities be required by law to open source their protocols then your argument doesn't apply, as of right here:

> Overtime, the most popular client will get a quasi-monopoly and will become incompatible with the rest of clients, no one will notice except for us nerds on HN.

As soon as those popular clients get incompatible, they cross the line into a new protocol - and we force them by law to open-source license that too. Sure, the major platforms will still own 90% of the traffic since the public is mindless (for now, until its educated over generations), but third party options will again be possible.

So all this advocating for protocols is still very relevant, it's just probably not going to happen by itself or from some magic invention.

I also think regulation is important. Like energy markets, some software, particularly communication systems, favor big players due to network effects.

So regulators should prevent anticompetitive behaviors such as what happened with Google Talk. When Google Talk was a new product, it was a fantastic XMPP implementation. They even contributed the Jingle extension to implement VoIP.

Once they became a dominant force, XMPP federation was removed effectively making it impossible to communicate with Gmail accounts from other providers. Later on, with the transition to Google Hangouts, XMPP was replaced by a proprietary protocol making it impossible to use even a different client other than Google's.

An open-source protocol won't prevent a platform from accepting content only from approved sources and serving it only to approved clients.

Facebook could start using on open protocol today, then block anyone not signed in through facebook and the result is nothing at all would change.

Maybe you're thinking platforms will somehow also be required to accept and share content from anyone? If so, all platforms become like 8chan (or worse). If not, there's someone, somewhere deciding what can and can't be shared on each platform (like there is now).

I think the fundamental confusion here is the idea that some kind of protocols vs. platform dynamic even exists.

Social media platforms are trying to achieve a captivating user experience built off our social nature to monetize our attention. Protocols are used, but they are deeply buried implementation details.

I don't know about that. Authentication mechanisms seem within scope of any protocol, and if an implementation refuses service based on some extra requirement (such as being signed in to Facebook) then that implementation is simply not up to spec.

> Facebook could start using an open protocol today

Richard Stallman would prefer you to use the term "free protocol", which I think is what we're all trying to say here ;)

Authentication mechanisms are in scope (obviously). The set of authorized accessors is not, right? I guess people are not accounting for the distinction, but it’s rather critical here.

I agree what you described - one client getting really popular is a huge risk, but the risk is not the popularity of one client, it’s the popularity of the client without any user knowledge of the underlying protocol. I recently wrote about XMPP and this problem https://salibra.com/p/the-death-of-jabber-8ea395e82f5b

The risk you’ve described isn’t as insurmountable: So far, bitcoin, ethereum, & BitTorrent serve as strong counter examples to the problem you state. They’ve remained clearly defined protocols that users seek out instead of a particular client and have proven hard for popular clients to fork.

When France had a lot of socialized businesses, they used them to enforce Proper French. There are variations on this theme that could potentially work here.

Years ago there was a push to get government to make all public documents exist in a standard format so that nobody was locked into a particular vendor's (potentially for-pay) reader.

If some independent entity as large as the Public Sector supported an open standard, then it would provide a lot of friction toward efforts to balkanize the standard. Because if you weren't at least backward compatible with the standard, then people would notice.

It is not “practically impossible” for individuals to set up their own email servers with good deliverability. I run my own and emails get delivered to my intended receivers all the time. If you want email to become more distributed someone ought to come along to make a Helm that isn't so expensive.

You just have to make the protocol sticky enough and distributed enough so one corp doesn't wield Total Control.

> I run my own and emails get delivered to my intended receivers all the time.

That's a pretty low bar, unless you mean "all the time" as "every single time". :)

Before I gave up and moved to AWS Workmail this year, I'd run a personal mail server for about 20 years. Major issues were not too uncommon, and minor ones were constant:

"Why can everyone in the world get mail from my server except outlook.com users?" I can't remember if I ever resolved this.

"Why do more and more corporate servers ignore SMTP retry rules?" It was nice when only spammers did that and graylisting was effective.

"No, it's not my server that's sending those spam emails that spoof my domain; I checked. Yes, really." This happened more than a decade ago, and I assume it either stopped or was solved by the spread of TLS or something.

I have had a Gmail account since the beginning and have, not infrequently, had my sent email arrive in spam. These arguments against the ease with which we may decentralize basic ubiquitious protocols lean on, I don't know, a basic bleak outlook?

People thought I was crazy for unplugging from every Facebook service. No one would follow me over to my messenger of choice they said. I proved them wrong. And you can too. But it takes a little patience and a little persistence and little sticking to your ideals. Our basic internet freedoms are worth this.

You are a fluke of a customer.

My bodega owner is a customer. So are his kids that help him. And an old mexican lady who works in a hipster coffee shop because hipsters that own it cannot be bothered to get up at 6am to open at 7am.

They use email. None of them are interested in "setting up their own email server". They are interested in being able to send email and get email.

Please read my comment fully. Helm is one such solution that gets out of your way but is too expensive. The price on these do it for you boxes needs to come down. It doesn't take much capability beyond email for one of these boxes to convince an everyday user.

> Helm is one such solution that gets out of your way but is too expensive.

My bodega owner lost interest the moment you said "solution that gets out of your way".

He simply does not care to hear about it. He wants to click on an envelop icon and see his email. He wants to type his message, hit send and know that his message is delivered.

Until the OSS, Protocols not Platforms, "Use firefox and not chrome because its technology is better" advocates internalize that not a single argument other than "does it just work?" matters to users, SaaS, Platforms and Chrome will continue to clean the clock of OSS, Protocols and Firefox.

I would wager than fewer than 1/100,000 adults on the planet can succeed at setting up and maintaining their own email server with good deliverability. That's "practically impossible".

Spolsky wrote a great piece about embrace and extend titled "Fire & Motion"[^1].

The most important thing for defending against these incumbents is to aggressively prevent and repeal regulation that would stifle competition. Nassim Taleb's view on incumbents is that as long as the mighty can fall (i.e. no bailouts, no protection), it gives disruptors a chance to rise to the top (so they can later fall).

- [^1]: https://www.joelonsoftware.com/2002/01/06/fire-and-motion/

I do have my own email server, but it is only for receiving and not for sending; the ISP's SMTP server is used for sending. Probably anyone could do this, if they have a ISP with email (which they ought to have, for this reason).

(Also, it seem that in Ubuntu, when installing Exim there is an option to do specifically this.)

You only don't believe it because you lack the imagination to build it.

Which is the case for everyone but the person who does.

Unfortunately, that person may have killed himself because of JStor.

Ultimately, you apply modalities not concepts which explains why you can't see it.

Email doesn't need to be email it only needs to provide equal or greater utility.

> Now you might say, "but how about email"? Well, first of all, email is an oligopoly and it's practically impossible for individuals to set up their own servers and expect their emails to get delivered all of the time.

I do it and it works well. My advice is to try it before you dismiss it as impossible.

Regarding your overall point, this is partly a social/cultural problem. There's no inherent reason why people would not search for the protocol and not the monopolistic application nor why multiple "best" application could not co-exist. After all, not everyone has the same use case so different applications can specialize to catering for different use cases.

The idea of protocols needs to become more mainstream and ingrained into the minds of common people and this is why essays like these are important.

> Overtime, the most popular client will get a quasi-monopoly and will become incompatible with the rest of clients

Yes, but it doesn't have to be this way. It's possible to design systems that align all those predatory capitalist incentives of participants with keeping control decentralized. Bittorrent and bitcoin are prime examples of this. And it definitely can be done for more applications. There is one problem though, creating such systems is not lucrative, it can't make creators rich even if it changes the world.

It cannot make you rich quick.

It can still provide a comfortable living to those who offer a superior implementation with a paid service, like Proton Mail, or whatever IP telephony provider you respect.

Article addresses the “popular client” problem and Gmail in particular. Yes, oligopoly but still possible to shift providers if any one client gets too controlling. Leading to incentive for popular client to play nicely.

I don’t see it as an end-all, but do think it’s ridiculous that, say, iMessage only lets you iMessage other iPhones and falls to sms/mms for others. What if Gmail only let you send rich text to other Gmail users, falling back to plain text outside their bubble? There is zero that iMessage does that couldn’t be done in xmpp, or at least an open protocol. Similar with social media. It’s all the same thing: friends, posts, wall. Use ActivityPub or something similar and open it up. Won’t be perfect, but will be better than one platform holding all the cards.

Political problems require political solutions. Us nerds in tech too often assume some magic bullet protocol or other data structure or algorithm will fix how humans relate. But at the end of the day, power is about power, and a protocol or a platform is just on form of power.

>Well, first of all, email is an oligopoly and it's practically impossible for individuals to set up their own servers and expect their emails to get delivered all of the time.

What the hell are you doing to your mail servers?

edit - also, can the client be in the protocol?

A possible remedy to this might be the emergence of “fat protocols” http://www.usv.com/blog/fat-protocols

One question: During the time which Facebook Messenger and Google Hangout are both using XMPP, is it possible for a Google Hangout user to send message to a Facebook Messenger user via Google Hangout client?

Hmm, I don't remember, but I doubt it. The point of FB Messenger using XMPP was that you could add your FB account to your XMPP client and chat with your friends from there

I forget where I read it now but if memory serves that was part of the reason why Google Talk xmpp federation was turned off because Facebook was importing Google Talk users but then not also sharing the xmpp federation so in order to stop bleeding users, Google Talk turned off their xmpp federation.

This is an anecdote until I can find a source.

Facebook never supported XMPP federation - only chat with Facebook contacts.

So the outcome of this would be that Nazis wouldn't be banned from the protocol but the client(s) that 99% of people use would not display anything the Nazis post (Mastodon is getting there already). Sounds fine to me but it would probably just shift the debate from deplatforming to client-stealth-banning.

No, that is not what I'm saying. I'm saying the most popular client will make its protocol proprietary and incompatible with any other clients, nazi or not. And no one will notice. Just like no one noticed when google hangouts or facebook messenger dropped xmpp.