1) A unified, coherent design language and user experience for the site. This would really make Yahoo stand out, as even Google is clunky with this ("more" and "even more"). As it stands currently, it's too visually cluttered and hard to navigate. Example: I used to use Babelfish a lot, but trying to browse to it from the front page is well-nigh impossible. Another thought: sort the list of Yahoo sub-sites by popularity instead of alphabetically. It's fine to have umpty million different sub-sites, but they need to be coherently organized and findable. Yahoo really needs a first-class UI, even more than Google or Bing do, because Yahoo actually is a portal, and a portal is different things to different people. It needs to be cleanly navigable and this includes cleaning up the 1990s looking My Yahoo.

2) Back to basics: search. As soon as possible, get rid of the Bing mess and get back to being a premier search provider. Type nearly any topic into any search engine and you get a Wikipedia page first thing. If I type "hacker" into Yahoo, out of the top ten results, there are five junky results: 2 Wikipedia pages, one Answers.com page, one Dictionary.com page, and one WordIQ page. If I want to get a definition or an encyclopedia page, I can go to those sites. Web search is supposed to pull up awesome cool links from the web, not the standard junk pages that have shallow content and are not hard to find. Even Google is getting infested with these kind of crap results. Nowadays, if I find an awesome link it is typically via a site like HN or somebody's blog, but I remember the days when I used to love going to Google and digging up fresh new links that were interesting.

3) Social is still "ripe for disruption" as they say. The flurry of interest around Google+ is evidence of this. People want more control, and more privacy, and a greater ability to differentiate their online identities without any one identity compromising the other. Eric Schmidt's response to this is "well if you have something to hide blah blah blah" but it's not about having something to hide. It's about the fundamental human need to interact differently in different social and economic settings. Maybe the guy who works as a cube jockey in some corporate wasteland needs to have his LinkedIn profile for trying to make it to the next level, but what about his sideline trying to network his way into a graphic design job? Maybe he needs a couple of profiles and wouldn't like to have to use his real name on either. Innovative approaches to allowing flexibility and growth on social sites (isn't that what the web is supposed to be about?) instead of locking people into one idea that they damn well better stick their RealName(tm) on... that is an area where Yahoo could step in and fill a void. Particularly if it was done right, and done in a way to leverage their existing myriad of user bases across the niche sites.

4) Quality. (Echoes of Steve Jobs and his quest for the perfect shade of yellow in an icon.) An enterprise-wide push for quality would help. Instead of being meh at 100 things, Yahoo could easily be awesome at those same 100 things. They already have the infrastructure in place, they just need to clean house and remodel. (This kind of ties back to point 1.) Flickr is a prime example. It's big, it's a brand name, it has a loyal user base, but it isn't awesome. It needs to be awesome.

5) Develop an identity. If the preceding four steps are done right, this will follow. Google's identity is "search plus a bunch of other crazy shit, and they have really cool offices to work at" right now. It works for them because search (advertising) makes them so much freaking money. But it's not really a coherent identity. Yahoo's identity could easily be "the coolest web portal with the best links". If Yahoo became synonymous with premium quality, they would be unstoppable.