We had a real-life web-lebrity in the Guardian's offices last week: Martin Stiksel, one third of the brains behind the music recommendation site Last.fm. He's laidback and rather unassuming, but has the kind of enthusiasm and insight about his company that is often missing in serial executives. He'd probably say that's because he always did it for the music not the money, but seems to have ended up doing rather well with both.


Last.fm has just signed Universal - the last of the four major music labels to partner - so the site can offer live streaming of full-length music on the site. It's the latest and most logical part of the Last.fm pie, and a rare example of a site that has legally, and successfully, combined music sharing with discussion and community.


Being bought by the firm last year was an enormous boost to the start-up community in the UK. (Let's remind ourselves the purchase price was £140m, lest we soon decline into an economic abyss and have to wait another six years for the internet to get hot again.) That deal not only cemented the reputation of British start-ups and our awesome music scene, but also reinforced the power of recommendation as a business model. And it's recommendation that is the secret source of Last.fm's success.


If web 2.0 could be summarised as interaction, web 3.0 must be about recommendation and personalisation. While the Tim Berners-Lees of this world work out how to make the language of the web function more effectively behind the scenes, our front-of-house task is to get stuck in and intelligently work these technologies into our businesses. It is not enough to understand the strategy behind these new applications, such as Twitter and Reddit - they rely on participation. Tokenism won't do.


Recommendation is nothing new, of course. Amazon has been pushing "people who bought this also bought this" for years, and tools like eBay's trader ratings system are staple. Things get more interesting as the technology gets cleverer; hence we get automated recommendation and personal recommendation.


So Last.fm uses "scrobbling" to tally up every song you play every time you play it, and then processes that data from all its users to work out what's musically hot, and what's not. What are the implications for editorial? It's no longer about what one half-cut muso recommends, it's about what several hundred thousand music obsessives have actually listened to. The rising stars chart on Last.fm lifts the lid on discovering new music - it's the new A&R.


Also in the technology camp we have Facebook. Facebook is fascinating because it is pushing in these new areas that will become critical to our businesses. It got a lot of stick for its Beacon ads programme because of concerns that it would share too much user data outside the site; your Facebook friends might get a targeted ad showing what you bought on Amazon, for example.


Perhaps I should be more concerned over how my data is used, but to me the road bumps in Facebook's development just prove how experimental the site is being. It's doing R&D for all of us. If we're honest, we'd surely all rather have adverts that are relevant to us.


And among our social networks, comment and opinion are being more efficiently aggregated than ever before. There's a swathe of local reviews sites such as welovelocal.com, which mixes listings with reviews; recruitment sites such as Zubka that pay out if you recommend a good candidate for a job; and bookmarking sites such as Del.icio.us that help rank online news.


Above all, the most reassuring trend is that the values of credibility and trust are more important than ever in the ocean of information we have to navigate every day. The technology is not enough on its own, and that should be a comfort to editors everywhere.