But in the end, it doesn't really matter how many coins you have, the idea would be that if you have a stake in it, you should be motivated out of rational self interest to keep the network decentralized (as much as possible at least). Donating, say, around 0.1% of your BTC worth to a cause like this seems reasonable to me.
So, say if you're sitting on around 100 coins, that means donating 0.1 coins. At a total of around 1000, 1 coin. I'd expect we could get a pretty decent amount of donations together like that, if people are convinced it's a worthy cause.
That all sounds great. I would donate maybe 0.5% - 1% of my stake to some concrete p2pool improvement proposals, that had been suggested (and voted on) by miners and GHash users. I would not donate any directly to miners though as I don't see that being a long term solution.
It probably bears waiting to see about this scalability issue before actually soliciting donations though.
Valid point. On the other hand, as I argued above, there is currently a "bootstrapping" that any smaller pool faces: while the pool is still small, variance is high for participating miners, so they are inclined to join a bigger pool, which in turn is less affected by variance, thus attracting more miners, etc.
That's why I said above, in the *early* phase of supporting p2pool, it might be necessary to counter that variance disincentive by additional donations, so that more miners join p2pool, which in turn will mitigate the problem and will (long term) make donations less important.
EDIT: Latching onto what calian and holliday discussed... is p2pool integration into the reference client, in one slick package, an option? Something the devs are considering? That might gain some support from donators.