Hacker News new | past | comments | ask | show | jobs | submit login
Protocols, Not Platforms: A Technological Approach to Free Speech (knightcolumbia.org)
235 points by pwg 17 days ago | hide | past | web | favorite | 67 comments



The "Protocols, Not Platforms" debate is naive and misses the point. Here is why:

- Let's say someone builds a social network protocol (including messaging, photo sharing etc).

- Let's be generous and say this protocol becomes so successful it turns into a de facto standards.

- The market will rush to build the best client for the protocol. Facebook, Google, maybe an outsider will come up with the best client that has all the features and added goodies (magic emojis, special video effects, something new etc).

- People wanting to get on the networks will never google the protocol just like people who want to get an email client do not google "imap". They just google the most popular client such as gmail.

- 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.

- This already happened with xmpp (facebook messenger, google hangouts and others used to be on xmpp).

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. So I wouldn't call email such a great success story. Another thing is that email is work oriented and so it makes sense to keep compatibility between businesses. Email also arrived very early and had time to spread slowly at the beginning of the internet among various small to big players, which explains why no company could take over it.

For all these reasons, I do not believe in a magical protocol that would save us from big tech. The protocol in itself would not be enough. A non-profit would need to be in charge of the most popular client such as Mozilla with Firefox, but even then this did not turned out as expected although Chromium is open source so yay.


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.)


> 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.


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.


> 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.


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.


>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?


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.


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


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.


What I'd like to see is a model where you a federated network of moderated groups. Something like Discord except more micro-blog than chat and where you can see popular posts of sister/brother servers, very popular posts on cousin servers and so-forth. Basically, a filter-tree keeping things reasonable and creating groups of folks with some relation to each other.


Decentralised group chat communication seems like it would work very well as the future model of social networking. Content online could spread organically instead of the surveillance panopticon telling us exactly what we want to hear.


Just to explain my preference for something allowing post/micro-blogging/blogging, I use discord for organizing real life activity but it really doesn't seem set-up for discussion of "issues" (political, programming, etc), here Facebook, Reddit or HN does best imo.

That said, topics are a bit like tags and a bit like sub-channels and if you could combine these qualities, you might have a nice merger of the approaches.


A decentralized platform similar to reddit with more autonomy and control over subreddits is sort of what I envisioned the parent poster's comment to be referring to. Something that more easily allows a custom format, likely with easier media uploading capabilities and a higher level of segregation between disparate subreddits.


Subreddit style communication is something I was considering yep, but I was definitely referring to Discord style server based group chats. If Facebook moved their entire platform in that direction (and hopefully abandoning news feed) I'd be really happy.


Yeah the discord group chat model is definitely better for talking actively with people, but for any kind of lasting content I find it pretty difficult to figure out.


On the other hand, discord server's do exist for that kind of communication; there are discord servers I am in that contain thousands of members and are about robotics, and often will have meaningful technical discussion about programming, design, etc.


How does that work? I've only been on discords with maybe ten people and it seems earlier comments basically get lost in the flood of later comment. You may communicate to another person or other people and those people might learn stuff but a given comment effectively isn't public, post isn't equivalent to publishing. Even adding channels for official documents, things simply get lost in the stream.


How is this different than Slack? Or even worse, Facebook News Feed, where the average lifespan of content seems to be converging on zero.



I made something like this, called Aether. (https://getaether.net) It’s not federated, but distributed / fully serverless.

It has already made its way up here on HN a few times already, this was the latest one: https://news.ycombinator.com/item?id=19161171

The protocol underlying Aether is called Mim, and it’s close to what the article is asking for: https://getaether.net/docs/developers/


I will take a more detailed look.

Still, I think "federated" is important in that different pieces of the system should be able to break off. If a mod can't leave a group, what use is it?

Also, "mod election" doesn't make sense. Mod should be (able to be) petty dictators - a member doesn't like a mod, they can easily go to another group and a majority means little-to-nothing online where a random number of socks or confederates can be easily recruited.


I agree to use protocols, and not platforms (it would be better to do that rather than what a lot of stuff including Hacker News are doing). I wrote my own software for IRC and NNTP (both client-side and server-side), but you don't have to use mine and you can use your own or others too. Regardless of protocol services are still required, but you can use other services with the same protocol; you can set up your own server if you do not want to (or cannot) use theirs, for example. And, SMTP is good for email, compared with having to use a web browser. (And, there still are free Usenet servers (some require registration and some don't).)

Decentralized systems also will help. However, I think ActivityPub and so on are too complicated; NNTP and IRC is good. Also, sometimes some federation with others might not be wanted, or only partially, or whatever reason; sometimes the policy also might not match what someone else has. That is one reason I write a specification of Unusenet, which specifies a format for newsgroup hierarchies which are not part of Usenet (it is possible for the same server to feed both Usenet and Unusenet, though). Perhaps I think even should set up a NNTP for Hacker News, too. (I also invented a Netsubscribe protocol, which could be a simpler alternative to ActivityPub/Mastodon.)

Let's Make Internet Great Again


I've said this 1000 times, but garbage like Facebook or Twitter could be replaced overnight by a protocol. There is really zero reason it couldn't be built from existing distributed tooling, be secure and just as useful without lizard people getting in the way.


> I've said this 1000 times, but garbage like Facebook or Twitter could be replaced overnight by a protocol.

No, it couldn't, and saying it more won't make it any more true. That is, sure, you can design a protocol that supports manpy of the same basic functions (and such protocols exist already), but centralized governance is ultimately part of what people want, or at least essential to stop a social network into degrading into what people don't want.


Neither of you have provided any meaningful arguments to support your positions.

I would say that Mastodon is an example that supports the OP’s point. But in the real, practical world, a social network needs investment and marketing to catch on and extend to a useful amount of people. So in that sense, you’re correct as well.


Mastodon is minuscule, and has already been showing growing pains.

(I'm on Mastodon, have been for two years, and love it. But it is no majick bullet.)


> I would say that Mastodon is an example that supports the OP’s point.

I would say Mastodon’s degree of success in replacing Facebook, etc., is an example which refutes OP’s point. (Also, Mastodon is more than just a protocol, but even if you ignore that...)


why not?

it certainly is technically feasable


> it certainly is technically feasable

It's technically feasible to build a protocol that supports the distributing social status updates group chat, direct messaging, etc.; many such things are around, some are, IIRC, older than Facebook.

The product Facebook, and why people use it, is more than just that technical underpinning, but involves active management. A protocol doesn't replace that, and a distributed protocol where nodes do their own active management, while possible, isn't a stable equilibrium. Which we've seen with open protocols in spaces related to what Facebook does many, many times already.


Can you give an example of how active management is done by Facebook (etc) in a way that a distributed protocol (with a team of developers and incentivized crowds) could not?

It seems to me the major differences between centralized and decentralized solutions are first-mover advantages, network effect advantages (big gets bigger), and advertising budget. I'm skeptical that the quality of the product is the defining reason for Facebook et al's success, or that it would save them against a viably-large decentralized solution.


I am confused, what do you mean by active management? How have I benefited from it?


Mastodon does exist and isn't perfect in all regards.


People who want to take control of their own lives, and curate the mix of knowledge / fact / falsity / opinion / propaganda / disinformation / useful-idiocy in their own heads can technically head in this direction today.

I think a key ingredient is translators that move content out of the platform-bubbles into the world you control. You can move material between RSS, email, blog channel easily, if you choose. Harder but doable is to crack the walled gardens of social media.

These can fuel development of better forms of expression than the primitive nonsense so often used today.

Many people have major barriers; not interested in taking control, not going to build anything myself, fear of leaving the mob, will accept advertising, want somebody to tell me what to do, too rushed...

That's fine, they will continue. Those who want can, and are, going down the other road.


Export options, even if mandated by the GDPR, are one avenue people could use. Though I imagine the tooling probably demands a lot of maintenance to harmonize all the different forms.


In that regard Facebook has improved recently. Their export now has a JSON format which is useful. They still exclude so much that it may be better to scrape from the UI though.


The downside of empowering the best of us with the ability to organize without constraints is that we empower the worst of us to do the same.

For whatever reference of best/worse you may have.


I care about Free Speech.

But most people don't.

In order for a protocol to be successful, it must bridge that gap.

That's what we've already done, gotten 15Million monthly active users onto GUN's network ( https://github.com/amark/gun ), by focusing on value add - whether people are idealistic or don't care, they're still using it.

This is how protocols will grow. Stay true to your values, but ship value add.


I'm not sure exactly what this is, but looks very interesting. I'm going to dive a little deeper. Thanks for sharing.


SMTP is a better example than NNTP because it is so much more mainstream.

SMTP had a spam crisis. Early on providers were loathe to filter at the server. They didn't want the responsibility. Then there was the "love letter" virus that brought the net to its knees. Filtering became a requirement for survival but e-mail deliverability became a problem, first with AOL as a big provider, later with gmail, etc.

SMTP is nominally open but if gmail wants to trash the mail you send you may be helpless.

Any new protocol has to face spam not only as an annoyance but also a dos threat. Look at how unusable the web has becomes with multiple pop-up windows, sites that say mobile is the future but God forbid you try to close the ads on your tablet, 50mbps downloads to see 5000 bytes of text, ...


This is what we are aiming for in building TruStory (https://beta.trustory.io).

The protocol is built as a distributed proof-of-stake network, where people can earn a crypto token by contributing good content, or by running nodes of the blockchain. Users and node operators can take ownership in, and govern the network based on how much of the crypto token they own (or have earned).

I recently wrote about the architecture at https://blog.cosmos.network/building-a-decentralized-app-wit....


How does this prevent a Sybil attack or astroturfing?

https://en.wikipedia.org/wiki/50_Cent_Party


We have a couple of mechanisms in place for this.

Users who have earned a certain amount of the token by being good citizens of the network will have to vote in new users. These users also moderate content, where offenders get some of their token taken away. Furthermore, network operators (validators) can vote to remove malicious content and users. And as with most crypto networks, a small fee (gas) is charged for usage.


I've long felt that we need a "post office of the internet". Some type of communication platform that's truly neutral, doesn't attempt to censor or manipulate the activities of its users aside from enforcing the law, and isn't liable in any way for their behavior. Sure, you can send a bomb or a death threat or an illegal copy of a movie through the mail, but the post office isn't liable for the actions of its customers. Its only responsibility is to filter illegal content as best it can, and that's all.


[dead]


The Metanet protocol isn't Bitcoin SV specific. Any immutable datastore could serve as the backend for a Metanet.


MetaNet




Applications are open for YC Winter 2020

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: