myna

Myna covers a reasonable subset of Unicode but one can indeed use Julia as a fallback for Myna to cover Unicode if one wishes.

The Myna page explicitly states it uses multi armed bandit algorithms, the algorithm fits the properties of the following claims.

At Myna we normally squish data through the logistic function to rescale (if necessary) and then use a beta model.

I agree with your general comments on bandits, and love what Myna is doing.

If that isn't an option: - Our main approach to managing latency right now is making Myna really really fast (avg response time 2.5ms, 99% ~6ms).

If you use the _digital_ MyNa card (e.g.

- If that isn't good enough, you can host your own Myna server(s).

Reproducing, say, the development process of Myna is practically impossible not to mention the prohibitive cost of finding devs as awesome as Dave and I;-) To apply the scientific method to software development you need to apply methods from the social sciences.

Our product (Myna; mynaweb.com) provides a JS client that our customers embed on their web sites.

Myna is my attempt to break some of these conventions to make things look a little bit even for programmers.

I work on this stuff in my day job (Myna -- mynaweb.com -- go sign up now!