HN user

akhavr

26 karma

https://medium.com/@akhavr

Posts14
Comments7
View on HN

Yes, you're right. That's why I've called the model "simplistic".

It already has lots parameters to deal with and, more important, it consistently underestimates your userbase - which is good while you're planning.

Though, point taken, soon I'll post an updated version.

The point is that even with simplest risk model, the padded estimate, the price for the risk, is lot larger than the work itself.

I am still quite can't imagine a rational customer agreeing to pay such premium. Perhaps this is my chance to make a fortune :)

It should, but it almost never does.

The only spec that's is not subject to interpretation is the software itself. Otherwise it all boils down to relationship between the customer and the developer: personal trust, etc. And, of course, this creates immense place for psychological games, which mostly are lose-lose kind of game.

It's slow just because it's hosted on quite low-end server, while we are in the early stages. Now we're profiling and optimizing out slow parts, as well as thinking about hosting upgrade/scaling on several servers.