Very concerned about the long-term viability of this experiment.
Don't be concerned. Even if the "unknown" increases and reached 100%, it is extremely healthy, and was the
normal state of affairs before mining pools were invented! Unknown miners are unknown to each other and can't collude. Yes, you can have a big secretive miner, like 18xf3MQ, but several exist today anyway.
You don't know and can't know that they are unknown
to each other only that they are unknown to the person constructing the chart. For all we know "unknown" might be a single miner, or 90% of it might be a single miner. In fact that isn't extraordinarily unlikely given that most truly independent miners use known pools.
We can't know. We can only think in probabilities.
P(mining pools collude) > P(unknown miners know each other) * P(unknown miners collude)
The odds that DF contacts Ghash (they have public email addresses) and they scheme up some unconfirmed tx censorship or double-spends, is highly unlikely as news of the stunt would crush both their business models, but is still more likely than the risk from unknown miners.
Assuming these total 50% which is not even the case now, hashpower is well spread out lately.
the 2 *olfi's around here seem incapable of noting trend in the mining distribution charts, ie, moving towards decreased centralization as mining becomes commoditized.