Any entity that controls more than 51% of network capability (i.e. owns and operates a lot of expensive hardware) can, in theory, begin making fradulent transactions called "double spends".
Any common user of the network that is unfortunate enough to get caught up in this scheme would lose their money. And the long term value of the network would most likely plummet, but in 2013 it didn't for some odd reason :)
Unless I am missing something, this paper is referring to connections between addresses when it mentions the “Bitcoin network”. Not actual mining nodes. It does not have anything to do with a 51% attack.
In that case, the select few owners of these coins could dump their holdings to buyers who are otherwise unaware of these large numbers of "hidden" coins. The owners make good money, the buyers overpay.
After that, it would be up to the core developers and the new holders of the coins to make the Bitcoin ecosystem a valuable place to exchange money. If they do that, then the coin will survive all that inflationary volume.
If the oligarchy never dumps, then they've just aided Bitcoin price support.
A coin does not equal a vote in Bitcoin version support, only hashing power does.
But de facto, I suspect wallet size is correlated with hashing power.
When I think of political control and Bitcoin I think of people who have the power to build a consensus for a hard fork. It wouldn't really make sense for someone controlling 51% of the network to hack the network. Then all their bitcoins would be useless because everyone else would abandon the network.
They could do short, targeted 51% attacks without doing much harm to the value of Bitcoin. In 2013 Ghash.io a mining pool had an employee who carried out mining based doublespends against a gambling site [0]. Very few people abandoned the network because very few people were impacted by this attack.
The Ghash double spend was a pretty big deal, and instigated a big campaign by miners to switch away from the Ghash pool. So it was costly for the pool, and it was probably not happy about its employee executing it.
Any player that directly owns 51% of the hash power, rather than just pooling, would similarly be economically incentivized to avoid such double spends, as they provide very little payout, while undermining confidence in the protoocl.
Any common user of the network that is unfortunate enough to get caught up in this scheme would lose their money. And the long term value of the network would most likely plummet, but in 2013 it didn't for some odd reason :)