Actually, I think the real problem with MSN was that they didn't start developing their own crawler-based engine four years ago. The people in charge didn't see the value of creating a new crawler when they could just buy Inktomi results (classic MS reasoning). If they had developed their own crawler engine, then they would have the breadth of results to back up the directory. They just didn't anticipate how far behind the relevancy technology curve Inktomi would fall.
I think that people miss the point when they say that a directory search can't be as relevant as full-text search. You don't NEED to search the entire Internet to find the best possible websites for many (most?) queries. The way MSN Search was set up, MSN's editors went out and found the best possible search results for the most common queries. And so for the 40,000 or so most common queries, the search results were hand-selected by human beings.
Because let's face it, automated search engine algorithms cannot judge the relevance and value of a website as well as a human being can! Full-text search engines can't read content, they don't "understand" what a web page is about, and there is no way they can consider "real-world" authoritativeness. (Try a search for "medical journals" on Google and see if you can find the New England Journal of Medicine, for example. Now go to MSN Search.) All they can do is make a statistical approximation of human judgement.
But obviously, the problem with editors is that they're not scaleable. They can only look at so many websites, they can only ponder what are the best results for a limited number of queries.
That's why if MSN Search had had a great full-text engine to fall back on, instead of Inktomi, and if MSN Search hadn't abandoned the upkeep of indexing their directory several years ago, they would have had the best search engine ever.
I strongly believe that the best possible search engine would leverage both the superior discrimination of human editors, with the superior breadth of full-text search.
To return to your question, yes I do think that the future MSN Search will operate without a directory. Microsoft has always been uncomfortable (and dropped the ball) with content. If they can get someone else to give them content for free, or pay MS to use their content, then they will do it. Otherwise, no.