I did the same analysis for

86.4% of the users are on 3 instances.

Used the stats from fediverse.network/pleroma

@njoseph could we at least up it to 8 so we its just slightly to large for them to act as a meaningful block.

@njoseph why does the author use instances.social rather than fediverse.network - instances.social is NOT a "comprehensive list" of intsances, it represents a small minority of instances missing most of the the small ones - you know, the ones that would tip this scale. id love to see that stat compiled based on fediverse.network information

@cosine More small instances wouldn't affect the argument. Even a couple of hundred 100-person instances wouldn't affect the look of that graph much. The statement that most people are on one of 2-3 instances still holds.

@clacke yes, that's correct i think

i have my own reasons that i think the argument misses the point but that is fair

@cosine @BestGirlGrace @njoseph I think that one point it misses is that federation, if not undermined from the get-go as in the given examples, has the power to keep a giant honest. It provides an escape hatch for disgruntled people, and has the chance of fostering a loyal opposition.

It *is* unfair that self-hosting is only available to those with the skills and resources (and we should work continually on lowering the level of skills and resources required!), but people with skills and resources provide a better check on the power balance than no people at all.


What if every instance admin voluntarily agrees to limit the growth of their own instance:

When you get to xxxx users - pass your skills onto another and help them start a new instance


#Federation #Centralisation #Decentralisation #DataSilo #WalledGarden #Mastodon #Pleroma #Fediverse #ActivityPub

@falgn0n @njoseph Yes, I think this kind of culture would help. Many people suggested that MastoSoc should have closed registrations after the first wave.

It also happens to be what I think companies should do. Richard Branson and Monty Widenius have thought like this, at varying scale depending on the nature of the particular business. That's why there are several Virgin companies operating airlines.
hbm.mariadb.org/ "The Hacking Business Model"

> The Company will be a distributed entity and strive to be small and efficient. If growing too big, it will split into separate business units or companies.

> In Virgin's early days, whenever one of our companies topped 100 employees, I would ask to see the deputy managing director, the deputy sales manager and the deputy marketing director. I would say to them: "You are now the managing director, the sales manager and the marketing director of a new company." Then we would split the company in two.
@falgn0n @cosine @njoseph

Now, Monty Program AB later merged with SkySQL Corporation AB, and several Virgin companies have merged with other companies, but as far as I know, none of these companies merged with a company they had previously split from.

Then again, "The People's Republic of Walmart"(0) claims that big command economies can actually be efficient. We're not talking about efficiency, we're talking about reducing risk and single points of failure, but the signal here is that a decentralized network may not be able to compete with a monolith that enters it?

(0) boingboing.net/2019/03/05/walm…

(was it a post by @Angle that linked to the PRW article?)


@cosine @njoseph

I don't think so? It's possible though, I link to a lot of stuff.

It's funny, actually, I was thinking about a rule that enforces basically this. Any time a company, or a fedi instance, or whatever, grows too big, it splits. I'll read more on the Virgin article, that sounds interesting.

In general, I think we need more checks and balances. Keeping companies, fedi instances, and the like small is definitely a step in the right direction.

Yes, I know I plan on limiting my instance to around 150. The Dunbar number is a good rule of thumb.

@falgn0n @njoseph This is basic (or should be) for every admin. It helps decentralization and avoids admin burn-out.

some intersting and valid points

author seems a bit too in love with their own opinion (and its "right-ness")

paints anarchy as a single (nasty) thing

future is never as clear as it seems


@js0000 @njoseph Yeah, the Anarchy Bad take (or Anarchy bad take) is going to ruffle a lot of fedi feathers.



Wow, Claes, the Hacking Business Model guide is fantastic.

I'll add it to our ongoing #humanetechnology community discussion about @aral cool vision of #SmallTech or create a new #sustainable #businessmodel topic for it.

Thank you for #sharing!


@js0000 @njoseph

@humanetech Oh! I'm glad to hear it, I felt I was maybe rambling away a bit too far on a tangent. 😀


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:
instances.social/api/1.0/insta (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 #gnuplot to do #loglog #curve #fitting with (in?)appropriate weighting of data

Sign in to participate in the conversation

This is a general instance supporting toots in English and తెలుగు.

Hero image credit: Sean O'Brien (CC BY-NC-SA 3.0)