I agree with all the points about what Steam could do better. Had these been in place I predict Offworld (and Ashes) would have better steam reviews and more players. Early Access should be only for devotees who will give the game a fair shake and are willing to take some pre-release hiccups in stride. That would mean less feedback, but frankly most feedback that makes its way into the game comes from devotees anyway.

If there is concern a game release goes quietly, there is a possible solution. Within a couple weeks after release allow people who have not purchased the game play for a free weekend. A "try before you buy". You may get many negative reviews during that period, but at least those people are reviewing a finished product that has had a couple weeks of post-release patches. And if those people like the game they can then buy it, boosting the playerbase.

Nicolas Gadenne, creator of "Dig or Die" posted a reply disagreeing with "No early access reviews", saying that most people aren't too critical during early access. While true for most, if even as few as 1 out of every 10 do no abide this, that's an automatic 10% drop in rating, and that may be enough to kill the game. He is also comparing his $8 2D platformer to your $40 game. People are willing to accept less of a complete game if they only pay $8. So if anything the developer should have the option to disable early access reviews, and I'd be willing to bet development teams that are really involving players in the development process are likely to exercise that option.

Personally, I think splitting the reviews is the answer.  During the early access period and for very early buying decisions, the early access reviews have some value.  However, being able to see just the post-release review scores, and having only those used in metrics would be better, both for games that have fixed their early access issues and (for the sake of consumers) for those games that got positive early access reviews because of their promise but released in a very messy state.

Good points.

Will be interesting to see if Steam sees any reason to adapt; being market leader is a comfortable position.
Technically it should not be difficult to separate early access from release reviews. One option would be to restrict considered reviews during early access to a time interval to reflect current state. Definitely separate them from release reviews unless player has spent time with this version and renews the review.

Finding the right players is really tough; agree with Soren that Steam does a bad job here with expectation management. EA is even less then a public beta, many players don't get it.
Better turn around times are key to keeping your players happy. Those who want to help shape a game take a lot of placeholders, if they feel their input has meaning and impact. If players chew for weeks on a dated version and are subsequently mostly ignored by devs, their motivation will evaporate.
So the intention of finding a smaller but dedicated group is correct; speeding up release cycles. A danger here is that these players become veterans too, with similar blindness issues as the dev team.