One of the reasons social systems are so fun to work on is that they involve a lot of unknowable dynamics. No one really knows, and no one really can know, how e.g. a particular implementation —from icon to text to placement to “rules” around usage— of “reshare” functionality will work for this particular group of users, who will in part process it in the context of their highly varied senses of your “brand,” itself quite mysterious, at this precise moment in their lives / Internet history. Will they all dunk on each other? Will they mostly “signal-boost”? Will the very idea of the feature seem suddenly, surprisingly dorky, leaving you like the last dipshit to open a disco?

The epistemic situation with social systems development gives it an “everything matters / nothing matters” quality that I seek out in every moment of my life.

• It all matters: tiny details can create incentives that suddenly go awry, or can transmit messages about “what this is all for” that inform some % of user behavior; every single post becomes part of “what gets posted there,” and cultural-memetic debt is accrued as quickly as enticing hits are shared out; every single news article sways someone, every single surge or drop in usage leads to different benefits or problems. So many decisions can be, and often are, consequential that one gets addicted to “often working on things that might matter to a lot of people.” It’s compelling.

• Nothing matters: so little of how anything will play out is predictable that if a given decision doesn’t go your way, it’s hardly worth agonizing about. Every week, some new, exogenous input resets all the calculations anyway. You have no choice but to loosely hold your ideas; everything is far too fluxy, non-deterministic. Stuff that “everyone knows will work” fails, and stuff that “everyone knows won’t work” succeeds. It’s real mayhem.

So: will the fact that short-form on Substack is immersed in, surrounded by, integrated with long-form make it different? Will the underlying financial model make it different? Will the users (today and in the future) make it different? Will other details or policies or accidents of chance and history make it different? I think it’s likely, although I can’t say how precisely, and that keeps work very interesting!!!