How do data companies get our data?


An easy-to-read guide to how data companies get our data, from Privacy International:

“Data companies – a catch all term for data brokers, advertisers, marketers, web trackers, and more – facilitate a hidden data ecosystem that collects, generates and supplies data to wide variety of beneficiaries. The beneficiaries of the ecosystem can include other advertisers, social media sites, credit agencies, insurers, law enforcement, and more.”


Have you thought of how you would act if companies wanted you to process data?


You mean us as We don’t and wouldn’t. Our whole ethos is to architect technology that cannot facilitate the processing of data by design.


How is that possible? It seems impossible.


I’m not sure what you mean?

I’ll give you an example with one thing we’ve built: Better Blocker. Better is an app we’ve built, but we have no idea who has bought it (asides from numbers from countries that Apple gives us and uses to pay us) and we have no idea how people use it.

Better has a list of blocking rules that it passes to Safari. Because it uses Apple’s content blocking functionality, we can’t see what’s going on in Safari, we’re just telling Safari what to do. People can contact us and report things they think need blocking, or sites where the blocking breaks the functionality, but otherwise we have no information (or data) on any person or their activity.

This is an example using Apple’s ecosystem, but the same goes for what we’re playing with Indienet.


I thought you wanted to facilitate the processing of data by designing a decentralised social network, like Kubernetes with people, kinda.


I’m not entirely familiar with Kubernetes, but it looks like it’s centralised? (All on their servers?)

So for Indienet, the idea is that we build the software for a personal federated website, which can be deployed to a person’s own domain name/server. A minimum requirement is that it is hosted in a way that can easily be relocated to another server without any technical knowledge required. We would never host anyone else’s sites ourselves, and in fact are looking to encourage providers to deploy Indienet sites for their customers (with the requirement that they be interoperable so people can switch providers).

For example, the work we’ve been doing with Ghent is following the idea that they pay for their city’s citizens hosting, and have their own space to help people set up their sites, based on codes that are mailed to the individuals. (See for an example.)


Kubernetes is a container management system created by Google for scalable systems and has no particular centralisation requirements apart from those conceptually aligned to container management (or put another way: it’s only necessarily centralised in the way that any meta system must in some way be…).

For what you’re talking about you probably need to be using Erlang based micro-services, as these can scale to millions of connections on a single server and have 32ms/year technically attainable downtime which a social network will of course require (or at least close to) for a proper social networking system.


I’m not a sysadmin person, but I don’t believe that to be necessary for a federated system. The idea is that these are loads of different servers connecting to each other.


For the volume of users necessary to create a sustainable decentralised social network unless you have a much larger budget than you present as having you’re likely to need something like Erlang in order to provide a low resource-usage underpinning for what in your conception is kind of like a telephone exchange for social media, which is exactly what Erlang/Elixir was designed and has been battle-tested proven to be able to handle – with functionally close to zero downtime, which you will need if you wish to instill confidence in any live messaging services which are usually part of the core social network service offering.


Essentially what you are doing would be bringing peer-to-peer back to the age of cloud computing… which is a rather beautiful paradigm… but you’ll need to provide the connective tissue for this particular vision I would have thought.