Proycon's Homepage
2023-02-13

Rant against centralising e-mail in big-tech silos, and breaking the internet in the process

Universities and research institutes used to be at the forefront when it comes to development of the fundamental technologies that power the internet. As early as 1965, researchers at MIT developed a kind of precursor to e-mail. In 1978, researchers at Berkeley leveraged the already existing unix tool 'mail' (1971) to communicate over a larger network. Universities and research institutes were among the first to embrace this new way to communicate with peers across ever greater distances and eventually, with the advent of the internet, all over the globe.

I'm a rather dismayed to see those universities and institutes nowadays no longer as pioneers and innovators in this area, but instead as mere consumers of ready-made corporate solutions, following corporate interests and centralising solutions. I have two employers, both academic, and both have resorted to big-tech corporations that offer solutions like e-mail as a service. E-mail "in the cloud", as they say. In these particular instances it's Microsoft's Office 365 solution that is being adopted, but I'd be just as vocal in my opposition it were Apple, Google or Facebook.

By adopting such solutions, you, as the user, very conveniently get great interoperability within a limited set of proprietary software tools (the ones of the vendor of course). That may be great if you are indeed a user of those tools and happy with it as you are. But as soon as you leave that vendor's ecosystem, the experience becomes less than stellar and often even a burden. Using one of their tools nudges you to also use the others, because it's the most convenient. Worse, it may nudge your peers and they instead apply peer pressure on you to use also those tools.

The average user may not care, but for those of us on the deep-end of the geek spectrum, these trends have worrying implications for software freedom, open standards, overall interoperability and net neutrality. Our concern is highly political.

I don't know what the motivations of the institutes were to adopt this e-mail service, nor who made the decisions, but I'm sure parties like Microsoft know how to ply their wares enticingly. Their promise is, I would guess, most often one of convenience and security. For an institute, outsourcing your e-mail needs to a 3rd party who has the expertise is of course not a bad idea if you lack the expertise yourself. It may be more convenient as you don't have to appoint system administrators yourself and secure your own mail infrastructure. That in itself, however, is not the issue.

My issue is everyone converging on the same extremely select group of service providers. This has already been ongoing for a long time for personal e-mail, with end-users adopting mail services like hotmail and gmail, but it is a different thing when we as academic institutions now follow suit and abandon any pretence of contributing to a mature and properly decentralised infrastructure. In this way, we are willingly contributing to centralisation, which if conducted on such a scale, may fundamentally alter the nature and power-balance of a decentralised network like the internet is in essence meant to be.

This centralisation has security implications too, because not just institutes and users converge on the select group of service providers, but so do attackers. When (I dare say when rather than if) a vulnerability is discovered and exploited by a malicious actor at one point, tremendous amount of e-mails might be compromised.

Let's take the political argument further, what is fundamentally at stake here is who owns the world's computing resources, the world's communication infrastructure, who has access to those, and to whose benefit are they employed. It is all about fair ownership and fair distribution, and in the end, a fairer world. In a digitised society, computing power as well as raw data will become ever more important resources. The question who owns (or has access to) both matters and should not be trivialised. E-mail in particular is quite a treasure trove of data, if accumulated by any centralised party.

Whenever you centralize something, especially on such a scale, you grant power to some entity (a private company in this case) and you have to ask whether this trust is justified and whose interests are really served. How will it play out over the long term? What if some eccentric right-wing billionaire takes over the big-tech company and enforces his own agenda? We've seen it happen already in the social media world, we've put all our microblogging needs in the hands of Twitter and are now tasting the bitter results when such a platform goes downhill rapidly. With centralised e-mail, we'd expose ourselves once more to such risks. Or what if some state-actor takes a hold like in China and uses the data for censorship and other repression? It'd be naive to think we're immune to all this in Western democracies. Technology choices made today shape the internet, the digital society, of the future. Putting all eggs in any single basket is a bad idea in principle. We need to nurture the decentralised nature of the internet.

A few years ago, I already wrote in my post Break free! Don't be a prisoner of your software platform! that the internet is not in the best of states. It has become a hyper-commercialised place in which users flock to free social media platforms, usually advertisement companies at heart, to post their photos, blogs and videos. This trend has developed for over almost two decades, in which fundamental technologies that used to be fairly open were sealed off and turned into walled-in gardens. Users were lured in en masse by the convenience of a service free of charge, but they pay with their privacy by being tracked, profiled and fed to an algorithm aimed to maximize their stay on their service.

It's taken a while but in recent years societal awareness has finally increased that this may not be all for the better. Is the technology still serving us or are we now serving the technology, or the companies providing it to us?

I always maintain that technology choices and usage have real-life implications; it's probably fair to say that social media, as an example, brought an unparalleled upsurge in disinformation, hate-speech, and a whole array of psychological disorders such as attention problems. Things that eventually affect the real world.

If you are, like I am, conscientiously choosing not to use centralised communication platforms that lock you in, then you quickly hit a wall if you don't play along. I'm already excluded (self-imposed) from various instant messaging groups because they use WhatsApp (= Facebook aka Meta), which I've always refused to use. Now at work the push towards M365 effectively locks me out of my institutional mail; this used to be easily remedied with a simple automatic mail forward, a basic mail function that's been around for ages and which everyone should have, but that option has been disabled. I get the concern you may not want everybody wildly forwarding everything and prevent sensitive information from landing on insecure servers, but a complete block is unnecessarily inflexible. Now I have no say in where mail directed at me goes anymore and am forced into Microsoft's solution.

As a result, I have resorted to just telling people not to use my institutional mail at all anymore, via a silly auto-reply. This is an option I'd rather not have taken as that burdens the sender with something that should have been, and can normally be, easily automated.

Aside from the political objections I also wanted to provide some further technical arguments against M365, as they make several technical choices that go against my idea of software freedom and how e-mail should work:

  • M365 blocks various IP ranges which affects self-hosted SMTP servers. The sender will have to explicitly register mail address and IP with M365 to be let through. I get the need to combat spam, which is the major cancer of e-mail. But this strictness breaks the decentralised spirit.
  • M365 rewrites URLs in mails, it modifies the incoming mail (which in my view is a already big no-go by definition and breaks cryptographic signatures). It makes all links go through their system (safelinks.protection.outlook.com). Again, I understand this is to protect against phishing, which is the 2nd most common problem with e-mail, but it misses the mark entirely, works counter productive, and introduces a huge privacy problem whilst doing so. Every link I click would go through their system and can potentially be tracked! Moreover, I only use text mails (no HTML, so far less prone to phishing) and their URL rewriting obfuscates the URLs, making it harder for me to spot whether a link is legitimate or not. I can make that decision on my own, I don't need nor want MS to do it on my behalf.
  • The IMAP login procedure with two-factor authentication breaks more traditional mail clients without OAuth2 support. Okay, then they should implement OAuth2, fine, I don't have an issue with 2FA in general. But the way it is implemented in OAuth2 by big mail providers like M365, according to the fetchmail developer, obliges mail clients to register with the mail provider. This gives more control to the companies and less to the user. The service provider shouldn't care what client I use, the only thing that matters is that I as user have the right credentials to access my mail.
  • I'm using a rather specific mail workflow which I've optimised over the years to work for me. Using my own mailserver, notmuch, neomutt and more. It's certainly not for everybody but it works for me (and it took me years to find something I'm actually comfortable with, most mail solutions suck pretty bad, or I suck at mail, probably a combination of both). Pulling in IMAP from M365 into my mail workflow would require me to spend a few nights setting up complex solutions which I don't really need or want in the first place.

I may be a bit (okay, more than a bit) of an opinionated left-leaning idealist, but the principles that formed the backbone of the internet are grounded in interconnectivity of decentralised entities. Agreements like RFCs establish protocols that specify how communication should proceed and what is expected of implementations. Anyone is free to implement them. Communication especially should be open, secure and interoperable, or we end up communicating only with people in different walled-off corporate (or governmental) silos. Which is sub-optimal. I strongly believe in federation, but that's been under siege for well over two decades because it's 1) technically much more difficult to get right 2) socially harder because it requires communication and coordination and 3) harder to exploit commercially, as trapping users in your own platform is way more profitable.

In conclusion, I want to appeal to institutes and individuals alike to think about this and contribute to a better decentralised yet interconnected digital future, rather than be complacent in its demise: keep fundamental technologies accessible to everybody, not just big commercial (or governmental) players, and improve upon them where necessary.

If you haven't had enough after my rant on this subject, I strongly recommend you to also read Carlos Fenollosa's blogpost "After self hosting my email for twenty-three years, I have thrown in the towel. The oligopoly has won." who makes compelling arguments in the same direction and explains the technical difficulties more in depth than I did here. I know I'm not alone in this struggle.