Linked article (which makes good points about too much centralisation) says:

> Frankly, this distribution is closer to the Dirac delta function than a power law.

This may be a throwaway line and I may be showing excess pedantry, but here goes:

My analysis shows that it actually *is* quite close to a power law, very approximately:
active_users = 30000 * x ** -1.3
where x is the rank of the instance (x=1 is the largest instance)

data from: (with my API token, retrieved tonight)

piped through:
`tr "," "\n" |
grep active_users |
sed "s/.*://g" |
sed "s/}.*//g" |
sort -nr |
grep -v null > instances.dat

log-log plotted with gnuplot, the line of fit is guesstimated by hand because I still haven't figured out how to get to do with (in?)appropriate weighting of data

Sign in to participate in the conversation

Welcome to, an instance for discussions around cultural freedom, experimental, new media art, net and computational culture, and things like that. This is part of a family of services that include mailing lists, group chat, and XMPP.