Ohad,
Instead of using Agoras as a blockchain to support TAU, would it be preferable to port it on top of EOS instead? We would end-up with a truly powerful langage using the latest (scalable to over 1million tx/sec) blockchain platform with not transaction fees ? Then, the Agoras token would be dedicated to create the TAU/AGORAS network effect. Man, I would like to see this happening

Then I suggest using ethereum
lightning network technology that can handle billions of tx/sec.

Just kidding, asking Ohad to make an EOS dapp is like asking him to give up his years of works and start something else from scratch again.
Well, sharding comes with severe limitations when it comes to contracts transacting with other contracts. TX/sec is an incomplete metric to evaluate a platform capacity to handle a project like AGORAS.
In my understanding, the TAU Langage is a project in itself that could be implemented as a smart contract (DAPP) on EOS, taking advantage of parallel processing and other EOS features to create an AGORAS market place based on the TAU Langage. If so, other TAU based Dapps (like AI Dapps) could also be created on EOS. I think that Ohad years of work are related mostly to the TAU Langage. The AGORAS Blockchain is only one implementation of this very powerful paradgim. I would appreciate Ohad's opinion on this.
the most important part of tau is the ability to change itself with time. therefore it cannot rely on a system which doesn't have this property