Removing the names from the problem
How I actually approach something before I understand it.
If I call something a platform, a company, a market or a community too early, the name arrives carrying assumptions I have not examined. Worse, it arrives carrying other people's solutions — call a thing a marketplace and you have already half-decided it needs listings, search and reviews.
So I start by taking the names off.
- What are the parts?
- What relationship exists between them?
- What rule changes that relationship?
- What behaviour appears?
- What happens when the rule changes?
- What feeds the result back into the system?
- What stabilises it?
Then I can rebuild the explanation, and the name I end up with is a conclusion instead of an assumption.
The loop
Observe without explaining too quickly. Separate the parts. Find what connects them. Find the rule. Watch what it produces. Identify what amplifies and what stabilises. Find the threshold where behaviour changes character. Change one condition. Compare. Keep the explanation only if it survived the change.
The first explanation does not need to survive. I would rather have coherence than a defended earlier conclusion — which is why the things on this site carry dates, and why the ones that changed still show what they used to say.