First off, I appreciate the write-up. Most of my Friday fight conditioning is just snarky tweets about the data mesh. I’m not sure I’m cut out for something so substantive. (View Tweet)
Anyway, as I understand it, the core argument of the post is that we build a data OS, and then have specialized apps that sit on top.
Fair, I don’t think it’s controversial to say that some apps should be combined and some separate. The question is where do you draw the line. (View Tweet)
My view is that generalized consumption—ie, explore data to answer generic questions—should be combined, regardless of who’s doing it.
That’s because these workflows are fluid. Data scientists benefit from Tableau. Tableau users benefit from forecasts in their datasets. Etc. (View Tweet)
Plus, those people need to collaborate, and it’s really hard to build real collaboration through APIs. It just works better in a single app.
The things that get split out are things where the question is more specialized, not the means of answering it. E.g., testing platforms. (View Tweet)
I also think we could embed data in a lot more operational workflows. Think of how we use data to make decisions on Kayak, or Yelp. That’s the real future of specialized apps. (More on this later, probably.) (View Tweet)
Last point - a clean data OS and open standards are nice ideals. They seem clean and elegant. But they’re really, really hard to build, and people care about solving problems, not the craft with which it’s solved: https://t.co/7NhmrRsVfw
(h/t @besquared) (View Tweet)
Which brings me back to @devarispbrown ‘s point: These things, just like everything that’s ever been written about the blockchain, sound nice in think pieces. But that doesn’t make them practical. (View Tweet)
