Measuring centralization can be an informative academic exercise, but IMO the only way to know if a network is sufficiently decentralized is to attack it. If it can withstand attacks from nation states, it's sufficiently decentralized.
-
I'm worried about using "decentralization" as a metric to define *anything*. The Howey test doesn't have a decentralization factor, though it does specify a "common enterprise" component.
This is why regulators determining degrees of "decentralization" is so damn dangerous.
Indeed, metrics radically oversimplify the computer science and practical issues of trust minimization.
-
-
Not everything that matters can be measured, and not everything that can be measured matters. Decentralization matters, for many things, but it is not easily or reliably measured.
-
Goodhart's law here may apply
-
Correct! Just learned that it had a name and now it's indispensable. https://en.wikipedia.org/wiki/Goodhart%27s_law …
-
This Tweet is unavailable
-
Sorry there is only one Tweet, remember you have to reply to one of the main tweet from the thread not a comment. If you think I'm buggy please let me know by DM or email
Need help? check this link: https://threadreaderapp.com/help 
End of conversation
New conversation -
-
-
Without a perfect system model, you can't align metrics with goals - see our paper formalizing Goodhart's law - https://arxiv.org/abs/1803.04585 *With* perfect system models you have security. But it turns out that the actual models used for security & metrics fail for identical reasons.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.