The false walls of the walled gardens

All of the information on the Web the public wants to see is already in its hands. The question is merely how to keep it.

It was early January, 2019, as I sat in the dining room of a restaurant at Plaza Semanggi in Jakarta, hot off the heels of yet another exhilarating conversation about the Web and censorship with my husband back home. Unsurprisingly, I don’t recall what the flashpoint was or why people cared, but it was clear to both of us then that something was seriously wrong with the flow of information on social media at large. It must have been regarding YouTube, because it was in that dining room that I hatched the initial idea for something I came to call HALO. A new architecture was needed for video sharing, and it was high time to learn from the mistakes of past attempts at dethroning YouTube to succeed.

Two months beforehand, I had personally sat in as an observer of two executive board meetings — one between MNC Media executives, Suzuki Indonesia and our family’s consulting agency, and another at our home office with RCTI. We were trying to create a promotional penalty kick video game in time for the launch of the AFF 2018 championship, and although we ultimately failed in that endeavour we learned some important information about the media landscape. More exactly, we learned that Google Indonesia — the headquarters of which I would pass by almost every day on my way to Pacific Place — was a total pariah. We learned that companies like MNC do not like dealing with Google, but they simply must, due to their control of YouTube.

All the while my husband and I also observed the rise and fall of an upstart service called Vanillo. They had some headstrong programmers and some considerable family support that afforded them not only the time to freely work on this, but also a licensure deal with Sony BMG. They ended up failing for technical reasons, as it seemed they simply did not have the chops to scale their video service. Downtime plagued them relentlessly and eventually put them to bed.

In a different direction as well, we also observed the launch of LBRY, a video sharing service backed by an eponymous shitcoin, and their inevitable failure due to poor marketing. They had the inverse issue that Vanillo did, caring way too much about cryptographic technicals, to the point that it would end up creating artificial problems of scaling for them in the long run. Too much math binds you, so imperfection is necessary to keep the service in bounds.

Years before Mastodon came to the forefront, my husband had personally known its creator, Eugen Rochko, as they were both members of a different community that Eugen had come under technical stewardship of for a time called the Colorless. It was thanks to this history that we came to understand a lot better how incidental Mastodon’s success was in the grand scheme of things, as he knew better the real interests and motivations Eugen had for creating it. He was not so much interested in changing the world or decentralising everything outright, although he was sympathetic to those ideas. He simply wanted to create a new community for him and his friends to use, and he had outsized technical talent and grit to do that.

Mastodon suffered immensely from the lack of integration with Twitter content, because Eugen saw no use for it once his friends had all moved to Mastodon. Ultimately it failed through the federation wars, which was an inevitable consequence of the poorly fleshed-out information sharing model in use. Pawoo came into existence by the hand of Pixiv, as Twitter was indiscriminately banning the artists of hentai, and it grew very quickly and dwarfed the “mothership” Eugen created. Given the sensitive nature of their content, that mothership led the charge for de-federating from Pawoo and many other places followed suit, leading to their social stagnation and eventual deaths as communities.

Finally, I had come to discover NewPipe after finding the old YouTube Vanced to be defunct. The app never uses platform APIs and provides usable video streaming services of YouTube, Soundcloud and more. Sure, sometimes it would break, but as long as the developers keep up with it and maintain, it will always work. Theoretically, there is nothing YouTube can do to stop them so long as they allow people to watch videos without logging in.

So, with all of this in mind, we came to realise many major concepts that were necessary for freeing the information abound on the Web: through YouTube and MNC, we realised that people don’t pay enough mind to the corporate and state reality of much of the Web’s information; through Vanillo we realised that people don’t pay enough attention to the technical challenges; through LBRY we realised that people aren’t pragmatic enough about the technicals; through Mastodon we realised that people are not politically-minded enough about how information is distributed and kept flowing; and finally, through NewPipe we realised that any adversarial information landscape will require systematic scraping of content.

Moving into today, I have come to look long and hard at Twitter and its contriving of the public discourse. While it’s neither here nor there what they ultimately do or how they do it, this much is true: a scraping-based client would completely ruin their controls over public discourse. Ever since Twitter moved away from outright shadowbanning people to an algorithmic deboosting moderation model for the sake of better PR, it has always been possible to see every piece of content if you really want to – Twitter just makes it hard for you if you’re using their client. A scraping-based client can trivially get around this, and since it has no API keys to opaquely revoke (as Twitter has done to all third-party clients in the past), there is nothing they can do to stop it.

Much could be accomplished with a tool like this. Blocks would be neutered, only preventing replies, likes and retweets. Content would be cached by default and a user could simply be notified of deletion or changes instead of having it flushed from their devices right away. Users could freely access the vast and deep geometries of replies and quote tweets and make fully arbitrary selections thereof. Such selections could be rendered into smart screenshots of sorts that show a customisable selection of information about the posters and their content. All of the information can be freed, and Twitter would have to self-immolate in order to stop it.

But what about beyond Twitter? There is so much more to the Web than that place, after all. Smart screenshots are a powerful idea, because they can allow the transmission of fully arbitrary amounts of information in a human-readable way, and it is not difficult to use ML to reverse the text from such transmissions for modification and accessibility. YouTube comments, instant messages, and so much more could all be freely intermixed, so long as we have these tools that chiefly care about the transmission and comprehension of the data we share.

In general, platforms will surely be uneasy about something like this. Unfortunately for them, there is little they can ultimately do about it, and intelligence agencies like MI5, the NSA and Mossad probably already have systems like this that are far more sophisticated than what I describe. But this is alright, because it means the public are no longer sitting ducks for phoney propaganda campaigns and systematic lying and abuse from authorities. The platforms will be forced to take an adversarial position and will probably limit the types of information they share to combat this, which always comes at a price. The more you outright limit information, the more you put out solid proof of censorship, destroying your own plausible deniability and bringing accountability upon yourself from elsewhere, be it Congress, shareholders, Party apparatchiks, or even just the public. If people can get the information they want to see, they will get it.

But still, we must be realistic with how much information can be shared or saved. A cryptographic proof of every like or RT is a total non-starter for any real-world social media platform, and working around the insistence of something like that is silly. It would be more pragmatic to escrow important things like polls, and and only bother cryptographically proving stuff people actually care about. It would be ideal to have a totally disjointed “proofnet” where people only look for as much proof as they need. We should use webs of trust and Merkle trees to distribute the load of such proofs, and ultimately important proofs can be placed on strong L1 blockchains where more minor ones live off-chain or even just in the deep web outright, referencing more critical proofs that are easily accessible and trustable.

We live in a world of trade-offs and compromises. We also live in a world of profound technological progress, which is available to us today. If we want to have the brighter future made possible by these progressions, we need to regain control of our own information to the best extent we realistically can. We’re the ones posting it, after all! People need to have the power to see honestly into what is in front of them, and it will be enough. Let there be light.