You mean something like this (pardon the aweful picture)?
That would be quite useful if made mandatory. I guess not every EA title can define every milestone until launch early on, but those should receive an extra warning from valve or mustn't be sold above a certain price.
It would be easy for a customer to see, if the development is coming along nicely and if not, the dev-team will surely put out an explanation and the customer has been warned.
No I meaned more like:
"Hello. We are making the most awesome game ever called supergame. EA build is Alpa 0.8 and is all we have for now. We are going to add features 1-5 later in developement as the game goes on. Here is our milestone list.
alpha.08 50%
Feature1 60% New UI
Feature 2 70% Gravity
Feature 3 75% Stones
Feature 4 80% Targets for stones
Feature 5 100% Throwing the stones
And we plan to continue developement of the final game after release to expand our sales."
So if supergame was 10$ EA they would get 3.5$ (50% of 10$-30% valvecut) initialy from sale and if the game has 1000 copies sold when it hits feature 1 they would get 700$ from the earlier sales and then get 4.2$ going onward from every sale.
This basicly means the dev can determine how complet they feel the EA first build is of the game, then how important they feel the promised features are and at what featurepoint they feel the game is 100% ready and they may promise to do more but want to get 100% money before adding those features to extend the games lifetime after release. If the game project would be cancelled/ended before 100% the rest of the % would be returned to customers steamwallets. Making it smaller risk for customers, and less hasle for steam.
This way we avoid DF-9 Situation what would probaly launch with 0-2 feature goals taking them up to 95% and then having load of goals after that. So we could see that theyre not confident in having the money/resources to fund the game longer before needing the money out of steam. So we could more easily see the risk curve.
This method would not force the developers to do milestones like for publishers what they hate, but would more easily show customers how commited they are hitting those features. There would be no timegoals set.