When did you change your mind on this subject?

Another worrisome aspect of the blockchain statistics is that the "effective fee" for bitcoin payments is about 4% -- since the mining community earns 4000 BTC/day to process transactions whose estimated output volume (minus changeback) is 100'000 BTC/day.  Since mining is still a substantially free market, the actual cost must be not much below that value.  

Currently users do not pay that 4% fee, because it is paid by all the long-term holders of bitcoins, that lose 4000*365/12'000'000 = 12%/year of their value because of mining inflation.  (And, of course, that loss is more than offset by the price increase due to speculative demand.)  But since that fee will some day become a real transaction fee paid by users, it seems to be a bit too high for a service that is still often touted as "free" or "much cheaper than bank transfers".

But it gets worse if that 100'000 BTC/day blockchain traffic volume is indeed partly "fake" (transactions with no change in ownership).  In the final future when the mining costs will be paid by the users, the fake transactions will disappear and the cost will be borne by the real ones only.  But if 30% (say) of the volume now is fake, the "effective fee" now is not 4% but 4000/(0.70 x 100'000) = 5.7%.  If 50% of the volume is fake, the fee is 8%, and if 90% is fake, it is 40%.  So what is the "effective fee" now, and what will it be in that final future?

Putting on my triple-layer aerodynamic tinfoil hat, I would suspect that the blockchain traffic is being inflated with fake volume by The Powers That Wanna Be, in order to keep that calculation at 4% -- and thus preserve the illusion that bitcon can be a competitive payment method in that distant future.