“Again with Twitter?!”
Yeah I know… Twitter is here, get used to it.

Alright, so what does Twitter need? First one who says “a business model” gets a cookie. Yes, they do need that, but let me take a detour through the user side of things for a moment and I’ll get back to that in a minute.

As I have said before, Twitter’s simplicity is its strength. I don’t think they should give it up for anything. Others have tried to “enhance” the user experience by adding threading and things like that. An interesting idea on paper, but it does take away some of the product’s usability. Don’t get me wrong, I’m not a hater; let’s just say it’s different and the crowds seem to be gravitating towards more simplicity.
So Twitter has to stay simple, but there is still room to expand. And I think topical conversations is indeed something that could enhance the user experience tremendously. If done right, and in the spirit of the product.
In comes the #hashtag. We all know that you can use these to specify what topic you are talking about. This is incredibly useful and I strongly believe Twitter should make it its next big push. So to put things clearly:

@notpatrick thinks @twitter should embrace #hashtags and make them the second arm of the community.

Here’s how it would work: you follow @people the way you always have, and you can also follow #conversations. These would work pretty much the same way: if you follow #grammys, anytime anyone says something in that topic (with a #grammy in the message), you see it in your feed. Thus following the global conversation that you’re interested in.
Of course, this has a huge potential for overflowing / making your feed unreadable. Just separate the two: add a tab on the front page for your conversations feed, and make it an option to merge the two.
It also has a potential for spamming: the one internet vice that Twitter is spared from because you choose who you follow. If anyone can add a #something and be on 20.000 people’s feeds, it will happen fast. But there are ways to control it (you need to sign up to be on twitter) and a clever use of crowdsourcing (a “signal user as spammer” feature) would pretty much smash the issue on the spot.

And just think of how much we would get out of it though: the Twitter dynamic mixed with information you care about and opinions from around the world. Discover interesting people, learn new things… A new world of possibilities opens. Now that the Fail Whale is nothing but a distant memory and that Twitter is a reliable tool, I think the team should concentrate their efforts and get this integrated into the service and API.

So what’s the upside for them you ask? Well, not only would that be a great for the users, but I think it would also hold the key (ok, a key) to the long awaited business model for the service. Sure, they could make the general API a pay-for acess as other have suggested, but I think this could also be an avenue worth exploring.
We have been promised no advertising in our regular feeds, not now, not ever. And I agree that it would be both an inconvenience and a sort of invasion of privacy. Not so in the #conversation feeds. These are general conversations that no one really has ownership over, hence no sence of privacy. As long as your regular @people feed is protected, I think most people wouldn’t mind seeing an ad an hour in their #conversation feed.
Now if we agree on this, think about how insanely tailored those ads would be! Not only do they already know all about you (your tweets, the people you follow), but now they would also know precisely what you’re interested in (the conversations you subscribe to). This would be Google Adsense on steroïds and would sell like virtual hotcakes.

There you go Twtter, problem sovled. Get on it.

And baring that, third party clients like the wonderful Twhirl from the Seesmic team should integrate that functionality. Like, right now.

[Edit: People are noting that this is already possible in some third party apps, which is great. But this is only a stopgap; having it integrated in the stardard site and API is the only way to make it truely universal and push the feature’s use. And the only way to allow for monetization of course…]

[Edit 2: Another possible monetization: I would actually pay a buck or two a month for an added service like that, and I’m sure a lot of other people would too.]