March 2025 Wrap Up & News
I’ll be mixing up the monthly wrap-up emails, including more new and recent stuff I’m still reading, digesting, and doing…
Big new things
Datacisions.com is back in action with daily random thoughts from me on decisions, data and other stuff.
Did a major update on my “Substack Enhanced Stats” Chrome extension for Substack authors.
I’m adding a new section to my monthly update about posts (this one!) and news I’m still reading and contemplating. (you know this one you’re reading right now)
What happened here?
The Uncomfortable Truth About B2B Roadmaps: Implementation Challenges & Solutions
15 lessons for internal data product managers I learned the hard way
What am I still contemplating?
I’m thinking a lot about the AI co-scientist lately. What is it? It’s exactly my cup of tea ;) It’s a creative support system for scientists helping to generate theses. Now the amazing thing about it is that it works really well, and that it does so because it is a set of agents that work well together. It makes me think, that agents do have a future and that more types of co-Xs should exist. The system Google has pioneered here is actually pretty easy to generalize and specialize again to other tasks (in my mind that is). I suggest you read the Google blog post and the paper (including the appendix, good stuff in there) to get a good understanding of how the system actually works.
“A robot has a lifetime cost of 10k$, works 22 hours a day and lasts 5 years for a maginal cost of 25 cents.“ Yes that made me think and got me to subscribe to my first robotics newsletter.
Homomorphic encryption. I’m still a fan of “encrypt everything” and I was pushing for all-encryption databases before. Now the further AI goes, the more the need for something that’s called homomorphic encryption arises. A homomorphism is a structure preserving map (between algebraic structures, if you want to get math geeky). So the idea is…. you encrypt stuff (the new algebraic structure is the encrypted space) and still can do all the compute (like training ML systems, inference, SQL,…) on it. If we get that to work, the world becomes a whole lot safer (and Werner Vogels a whole lot happier).
Books I have open right now
I'm not saying I recommend any of those; I’m sampling, reading, and discarding a lot of books nowadays.
“Blue Ocean Strategy” one of the classics of business strategy (well recent classics) that I haven’t read. I’ve sampled it in the past and discarded it, now I’m diving into it again.
“Cybersecurity essentials” so far I’m liking the book, love the frameworks and analogies it provides.
“Superagency” not sure I’m gonna continue with this one, but Reid does have a point in general, we’ve been technological creatures for some time, and AI will only help us to change to the better and gain more agency (if we do it right… that’s kind of the caveat you always get).