For one thing, a lot of those tokens are either centralized or represent products that don't even exist.
For another, they are over-counted. Imagine the thought experiment: I create a new eth token and sell a bunch, pumping the market cap of my new asset to $100M. That eth that was paid for my asset was not sold - I'm still holding onto it.
Now I've just added $100M+ to the 'dominance' factor by doing nothing. People bought my token, and maybe even bought eth to buy my token, with no sell pressure being added anywhere in the eth ecosystem. This extra dominance was just sort of poofed out of nowhere.
So, 55% is not really as meaningful as some people ascribe.
Do you think there could be any other metrics that can tell a better story? (serious question, just wondering what your thoughts are)
[1] http://www.cnbc.com/2017/04/26/bitcoin-price-sec-winklevoss-... [2] http://www.cnbc.com/2017/04/12/bitcoin-price-rises-japan-rus...