How can we make our case?!

A little easier with this

There are many ideas and insights which have great data underpinning them, but they gain almost zero public engagement. Even in the realm of the thinkers, these ideas exist as ghosts.

Can we find a home for these these unacknowledged ideas on the internet? Discussing controversy on Twitter is asking for trouble, it becomes a battleground far too quickly, reddit produces groupthink and astroturf, and wikipedia is the ‘consensus only’ platform.

Facebook, instagram, snapchat? Forget about it.

The idea for case.science came about when I was presented with thousands of science papers, which taken together, painted a picture entirely different to the one on offer through mainstream sources. How can it be that two information ecospheres can coexist but never coalesce? Whatever you might think about the pandemic, throughout it there was an unacknowledged river of information, flowing in parallel alongside the accepted mainstream.

If you navigated the treacherous waters carefully enough, this reality was as clear as day, but for those in the mainstream, they swam through relentless warnings about ‘the dangers of misinformation!’. Repeated ad nauseum, these warnings roughly translate to ‘STAY IN THIS LANE, because out there lie demons!’

Or, as Jacinda Ardern put it, “we will continue to be your single source of truth.” Is this what we want? A single source of corporate government sponsored truth?!

The public, strapped to their train tracks whilst a vibrant discussion takes place below.

Into that backdrop arrives ChatGPT, which in its early days would happily go along with all kinds of interesting discussions which might upend our understanding of society. To our class-prefect political class, this is simply unacceptable. Before most people had even got to grips with the technology, there were guardrails on it.

What is needed is a way for AI to be confronted with the data supporting a particular idea in much the same way you might introduce anyone to a new idea. At first, there’s skepticism, so to get things moving you offer a few sources which might break the ice. “Maybe….” they think... but these are weighty sources you’re sending! People don’t have time to read your research papers and article collection!

So the inertia remains. Culture ambles along on its tired tracks with a turning circle to rival an oil tanker.

Well an AI doesn’t have a problem with weight sources and science papers! It can ingest huge amounts of data in a single sitting, and I’ve found that AI goes from ardent skeptic to ‘Oh really, that’s actually pretty interesting’ surprisingly quickly. And so in that spirit, case.science was born.

The idea is simple: when you ask the AI at case.science a question, it doesn’t respond from its existing knowledge picked up in the training phase. Instead, it takes your question and runs it as a query against a huge corpus of real data, data that normal users have given to it, and then it uses this data to formulate its answer. So what happens if you configure the AI to only use the data inside a particular case to form its answers? Well, you get case.science.

In practice, what this means is that you can load 3,665 science papers on vaccine harm into a ‘case’, and you now have an AI which becomes an expert in exactly this topic. If you could talk with 3,665 science papers, what would you ask? The AI can query all 3,665 papers at once, and then give you a reasonably decent answer.

But is that enough? The real beauty comes when anyone can start to put together cases of data to present to an AI. Imagine a database not dissimilar to wikipedia, but rather than being arranged around subjects and topics, with static data written by the class prefects, it’s organised around ‘cases’, with AI’s arguing the case dynamically. There’s no page on Brexit, instead there would be several cases which argue different aspects of Brexit. “Brexit has broken the british economy [253 sources]”, “Brexit has created new frontiers in science research in the UK [32 sources]” Each case with its own AI, with its own dataset, arguing the position from the data it has.

You get the idea.

In the previous version, making a case was very ‘web2.0’ like: forms, pasting links, write your description, hit submit and so on. It’s wasn’t exactly leveraging what’s possible with AI! So as of today, if you go to case.science, you’ll see that it works almost entirely through chat. The AI can create a case for you, it can add sources to that case for you, it can search for existing cases which might be interesting to you, it can list your cases, and it can even search the scientific literature for you! You could ask it to search for as many papers on Ivermectin as possible, then add them to your case….

Very soon its going to be able to read the news for you. It could follow science topics for you, and add relevant articles and sources to your cases for reviewing later. As a premium feature, it will be able to send full articles, or summaries of articles, directly to your kindle…

The purpose of course, is to build up a great repository of cases which are interesting, engaging, and insightful. So have a go and see if you can make something cool with it. If you have ideas and suggestions for the project, guess what, tell the AI and they’ll be sent to the developer (me!) to work on!

https://case.science

To be clear, there’s a very specific AI at the homepage of case.science, and this AI will help you with a lot. Once you land on a ‘case’, at case.science/case/some_case, the AI here will only argue about the data in that case. The casebots really can’t do anything other than read and argue about data!

PS - I put together a case on the documents that Edward Snowden leaked about the NSA’s ability to mess with democracy. Something not well understood…. and now that its in a case, it’s much easier for me to query the documents and gain insights from it!