WIRED: I wish I knew in advance which Silicon Valley trends were to become global hits, and which ones will die shivering from neglect in a dingy internet ditch.

I'm happy that this column was one of the first media outlets to talk about YouTube, Flickr, del.icio.us and many other "web 2.0" websites that attracted millions and were sold to Yahoo and Google.

But for every hit we've covered, there's been a miss. How can you make sure you're in the first category, and not the ditch? If you're starting a company (or dragging your start-up into its second year), what trends should you mimic from these hype-heavy hits of 2007? What dead-ends should you avoid? With my proviso that this is just one man's guess from what I've seen paraded in the past year, here's my pick of what business technology trends to ape - and what to avoid. This list should also have some value if you're a tech investor or even just a customer: buy into the companies that do the right thing, and steer clear of those that break these rules.

Here's what to say no to, whether you're a company seeking a new pitch, or a customer getting yet another near-spam invite e-mail. Steer clear of being, joining or investing in "the next big social networking site". I've been on every one of these, from Friendster to Orkut, to Tribe.net, to MySpace, to Bebo, to Facebook. Each time, the ennui grows, and the average time before everyone gets bored and moves on decreases.

More pertinently, it's tacking against the great trade winds of the wider internet. The business plan of social networking websites involves sucking people into your Facebook world and duct-taping them there by keeping them and their friends' information hostage.

Eventually, someone is going to liberate them: that's what the internet does. It may not be now, it may not be by an obvious competitor - but it certainly won't be by you.

Instead, take this moment to build your own website so that it can gaily chat with users trapped inside these social networking websites. But be eclectic. Writing a Facebook app (a program that can run within Facebook) is, medium term, a waste of time: in five years, Facebook will be nothing more than a featurette on a We Love 2007nostalgia show. But you will make some of your users happy and, more importantly, teach your company to make software that is flexible to work with whatever one-hit wonder comes along next.

Flexibility is everything. Avoid the current dogma about "moving your system into the cloud" - which is to say, building a business not on your own webservers and databases, but by hiring out "virtual servers" and renting time on pay-as-you-go web services like Amazon's storage system or its new database environment.

I'm swimming against the tide when I say you shouldn't do this: but one day this tide will turn. Today you should move everything on to the big companies' big server farms. Tomorrow, it'll be cheaper to run it all out of your basement. The economics of where to run software is as seasonal as strawberries, and you never know when Amazon's business model will switch from "providing you with everything you need to run your business" to "doing everything it can to destroy your competing sector".

Again with the flexibility: make your software so that you can run it as easily on someone else's servers as your own. Open-source software like Linux and MySQL are your friends here. Amazon will let you run it just the same on their servers as you can at home or on your local cheapo hosting service. Make it so that you can switch seamlessly between them and, not only do you have more economic flexibility, but your users will see even less downtime.

Ah, users. An increasingly dark side of web 2.0 is to reverse the meaning of "users", so that you end up using them. "Crowdsourcing" is the dirty euphemism - mining the content that your users create (be it recipes or home videos) to make money for yourself. Nice gig, but with a limited lifetime. Instead, concentrate on doing at least the heavy lifting yourself, and be open enough to let others create the rest. Build 90 per cent of the application that everyone needs, and make it easy enough for individual users to make the 10 per cent you don't know they need.

And watch them carefully while they do it. The biggest horrors I've seen in this mini tech boom come from companies watching other companies watching other companies. Don't get your ideas from Techcrunch - or me. Get them from your customers.

Media like us write about companies that got their ideas from their own users. And we write about them late - just before they get big, not while they found out how to get big.

You want to make it through this boom and, more importantly, through 2008.

Don't jump on 2007's bandwagon; start sending out people to explore what's coming in 2008.