AI-powered second brain? Yes please. But not on _those_ terms. How a data union would solve the problem.

I'm queasy at signing up for UseFindr's otherwise nice-looking Second Brain plug-in. My First International Data Union project is there to solve this sort of problem.

Everyone will need a second brain (Dalle)

Back in December, I wrote about the coming enshittification of LLMs, and I had this to say about how we could stop an endless recurrence of the same old platform playbook:

Well, soon is now.

I decided, after that post, to try to set up a data union, to see what is possible within current UK law and regulation, and, if changes are needed, to do what I can to change that too.

So how exactly is this meant to work?

Let me take a specific example. I love what LLMs are doing to my productivity. But I am processing more and more information, and my neurons are not getting any younger either. So I would like my very own “LLM-powererd second brain”. This would be aware of everything that I’ve seen in my Browser, would store local copies of it, would create a RAG index of it, would automatically generate situation-specific contexts and would fire them off to any of the LLMs in the background so that I could be reminded of relevant stuff that’s been in my attention-field at some point.

How would I use this machine? Well … say I was in Google Docs writing about some topic in energy policy; the RAG would identify the documents I have seen recently and find which were most relevant to what I was typing. The second-brain would take my current document and the reading I had previously done as a context and would, in the background, fire off a question to the LLM API along the lines of “Try to figure out where this piece of writing is going, and see if there are interesting connections and examples in the corpus attached; if so, return the title, publication and a link to the source document”. I would have 2nd-Brain keep a list going on the right of my screen with these reference suggestions. Maybe there’d be a slider allowing me to either go more focused or more creative.

Or another, more fanciful example. Say I was sitting down to supper with a friend. We start talking about politics. With my friend’s consent, obvs, I would have my phone listening in and my 2nd-Brain doing much the same as when I am writing - it would be prompting for things I might mention relevant to the current conversation from my recent reading, and would again allow me to set the slider to serious/humourous, focused/creative, etc.

Now … as is usually the case with ideas these days, it turns out I am not the first with the idea, nor the quickest to get it out: the guys at UseFindr.com have more or less already done this -

Not exactly, but close. With their app, you can connect your gmail, your google drive and you can run a browser plugin that gets all your browsing.

So… do I just go ahead and plug it all in? And what does this have to do with Data Unions, anyway?

The 2nd-Brain is something I need, but as my mouse hovers over giving UseFindr permissions to all that stuff, I do start worrying that it is all a bit Big Brother. The thing is - I want my second brain to be my brain, but in order for it to be “second”, I also have to hand over some control to others. I am sure that the guys at UseFindr are thoroughly decent and all that, but what are they going to do with all that info about me apart from providing me with a 2nd-Brain? Well, I Am Not A Lawyer (IANAL), but their privacy policy reads to me like they can do more or less what they choose with this data. Do I care? And how much, compared to the clear attractiveness of their product? And what if they get hacked and some seriously nasty people get hold of that treasure-trove … everything you need to impersonate me in my email and my Google Drive, if you care to look…

In the end, I did not sign up. Yes, I have given Google those sorts of rights … and there are other plug-ins I use that get close to that sort of data (eg I have been a user of diigo for years - since 2005, I think, in fact - and it is the closest that comes to being my current 2nd-Brain) … but if I actually pause and think about it, am I really going to hand over all that data power without a second thought?

And then there are my bigger picture concerns about the data ecosystem. Looks to me like UseFindr are going to do what everyone else does with data - sell it on to advertisers who will do their utmost to sell me stuff. I like some of that advertising … but I really don’t like the system that it has created in cyberspace: the attention-capture everywhere, the addictive apps … and yes, also the manufacture of desire and the creation of a toxic choice environment in which the angels of my better self struggle to get a look-in. I want a cyberspace in which my data is used for good - public health, transport planning, that sort of thing - and for a bit of monetisation. But by signing up to another app like UseFindr, I am contributing to a cyberspace which is all about attention monetisation and very little about data for good.

So … if I had all the time in the world, this is what I would ideally love:

• To tell apps like UseFindr what my data preferences are

• To have UseFindr reply with their terms of business given my data preferences (eg - “oh! We can’t make enough out of you if you are that restrictive … all those public health people, they don’t pay enough for the data … how about you pay us £5.month for our service?”)

• To decide if there’s a fair trade

• To monitor that UseFindr are keeping up their side of the bargain

But life is way too short for that. And what’s more, if it was just me doing it, I wouldn’t have a hope of changing the sort of cyberspace we have - that requires collective action.

This is where the Data Union comes in. A union that has my back, is working for me, and is doing that grown-up negotiation that I do not have time to do. The Union does one big negotiation with UseFindr, as well as with any other app that wants the attention and custom of its members, and shares the cost of negotiation across all members. And if there is a money upside in the transaction - which often there will be - the union shares the benefits collectively, just as happened with pay deals under the previous version of capitalism.

We collectivise the effort and the benefit. That is what I need. If the union gives UserFIndr a big tick for me, then that is all I need to know. I install. Indeed, perhaps it is even better - the Union will have figured out that UseFindr is right down my street and satisfies all my preferences and all the union’s data hygiene checks … so it pushed the app to me.

Let me generalise from this miniscule example of not installing UseFindr. If Data Unions can become widespread, this is the cyberspace we can hope for:

• Our personal preferences for data use are respected

• Our collective efforts have a chance of shaping cyberspace because large numbers of us are exercising our power together

• There are large pools of consented data, with the unions holding the consents, for public good uses of data

• The platforms face a countervailing power in cyberspace, because without consented data they cannot operate, and we now have a collective means of withholding data - going on data strike; we can use that power to make a better cyberspace (and to get paid for our data work)

• The period of creative destruction that we are in today with LLMs is extended indefinitely, because many of us will choose to share our anonymised interactions with all LLM developers, and not just with the model that happened to capture the interaction.

So that’s the vision and the logic. How do we get there?

That, of course, is the hard part. But I am very pleased to share that the First International Data Union (FIDU) - is now a social enterprise in the UK. I will be recording here the progress that I make in setting up the data union.