Voting Power
How much influence an address has in a governance vote, usually determined by tokens held or tokens delegated to it.
Voting power is normally measured at a snapshot block chosen when a proposal goes live, which prevents tokens from being borrowed during the vote itself, though tokens can still be borrowed before the snapshot. Designs differ: one token one vote is the baseline, vote-escrow systems grant more weight for longer lock commitments, and some experiments weight votes non-linearly to reduce the advantage of large balances. Delegated tokens count toward the delegate rather than the owner, so the distribution of voting power is not the same as the distribution of balances. Analysts measure concentration as the share of voting power held by the largest addresses and by the largest delegates.
In practice
Flash-loan attacks on early governance systems, where an attacker borrowed tokens for a single transaction, are why proposals now use snapshots and timelocks.
The common misunderstanding
Voting power is not the same as tokens held: it includes delegated balances and, in vote-escrow systems, depends on lock duration rather than balance alone.