It is impossible to remove transaction costs - even if the technology charges nothing for transferring EOS from one user to another, there's always the cost of transferring EOS to a fiat currency and vice-versa.
Imagine having to buy something from EOSBay (I should copyright that :) ) - you would either have to buy EOS with dollars or exchange another cryptocurrency for EOS. In either case, you will pay some cash to the exchange you are using and to the miners of the other cryptocurrency.
Also, if there are no transaction costs, the developers and sysadmins behind EOS will demand payment for their work and hosting services, which will have to come from somewhere.
Additionally, most people have no idea what Bitcoin is, let alone exchange their money for it, or even manage converting it to EOS, to buy something from EOSBay, when the normal eBay is still fully operational and sufficient.
This is why I highly question the feasibility of this project - even if it is technically sound, to begin with, and not another "pump and dump".
And I agree with you about Steemit - that was my point, we don't need another one.
I'm not sure man. It's up to you but I'm really putting money to this. Just like in every other seemingly unfeasible project out there, I'll take a risk. I've also been absorbing references about EOS and how disruptive it could be - the scarcity that it will try to fill in and I must say that I'm on the right position.
Let's hope you are right. I may invest in EOS after the hype is over and a couple of apps are up. If they are useful, its price will slowly increase over the years. But I have the feeling that Dan will create a new cryptocurrency in 2018.