I agree that the product/solution will change drastically after it is launched so having a site oriented around product features is pointless. What this articl

I agree that the product/solution will change drastically after it is launched so having a site oriented around product features is pointless. What this article is suggesting is that you should have one for 2 main reasons - firstly to explore the problem (you can do this with a blog or other info on the site) and secondly to start gathering names for a mailing list. The product will change but you will have a sense of the broad problem domain. There are many examples of startups doing a great job on this. Mint for example launched a blog and collected thousands of email addresses well in advance of releasing their product into market. When they did launch they already had an established audience and some traffic built up, which takes time no matter when you start.