I started on Haskell last year and I fall in the category of "discovering it by searching on google" and learning it myself, nobody told me about Haskell and I did not even know about it before. Let me share my perspective on this issue. I can vouch for Haskell being a very high barrier to entry for casually interested and even moderately motivated people. In the long battle most people will quit before the finish line.
I assume our goal here is to make Haskell popular, successful and influential. I believe it is already highly successful in the academic community and among a relatively small community of self motivated people who are smart enough to see why it is the right choice. What we need is to make it successful among the bulk of the engineering community who usually don't care what language they use. When I mention Haskell to anyone I know of in my engineering circle, they are usually confident that I meant Pascal. I think Haskell deserves a lot better than that.
I would strongly argue that we should not target the Haskell landing or first download page for those who already know Haskell, they won't even need it. It is for those who do not know enough and want to learn more. It is for luring and then trapping people. Specifically I would say that this page should not be targeted for students or mentored learners in general. It should be targeted to self learners. If you are a student, learning Haskell as part of a course you would have an expert mentor to tell you what to do. If you are a mentor you should know enough to not need that page and if you do then you are in the self learner category anyway. If you are a self motivated person you would anyway find what you need. We need the website to lure the mass engineering community of casual users or moderately motivated self learners usually looking for a better way to solve some problem.
In the engineering community people have little time. Everybody is busy with their day to day work and nobody has time and motivation to put huge amount of time and effort to learn something that they do not even know whether it will be useful. The way we can lure them is by showing how easily they can solve one of their small day to day problem, using Haskell, without investing a huge amount of time. That means we need a _zero_ learning build tool. Though build tool is not the only hurdle but it is an important first step. In my opinion we should not even talk about build tool user guide for beginners, it should not be required, the tool must be intuitive enough and the help available with it must be really good. Once you graduate to advanced usage it is a different matter altogether.
Another point that I want to make is that we should list only _one tool_ on the front page. Multiple choice is an immediate psychological barrier for beginners. When the users have a choice they must decide on one out of many. So either you explain the rationale right there or they will have to search the internet for reviews etc. In any case it requires an effort to choose and has a risk of turning them off. The committee should decide on a single choice after gathering enough information on what is best. They can ask for a shootout from the competing tools if that's needed.
In my opinion and from my experience learning Haskell for last one year or so, stack is a pretty good choice at the moment. It fits most of the requirements of a tool which has an out of the box experience. Though it may not be perfect it seems to be generally in the right direction especially for those who want to get started quickly and effortlessly.
-harendra