I've talked a lot on this blog about the idea of "Progressive Enhancement".  In case you haven't been following along, that's when you regularly release code updates rather than storing everything up for one massive release so that you can make smarter decisions based on the real life data you can collect through your earlier releases.

Here's the problem: if you are aware of your own inability to predict the future and you put off important decisions, there's never any way to tell if you made the right decision.  It's always easy to know when you get something wrong, but what if everything is smooth sailing?

We've been doing this all the time at work and I think that's directly responsible for the dramatic increase in the quality of our product over the past six months, but when someone new joins the company, how can I prove to them that it works?  We don't know the contrapositive.

So here's the new idea.  Even though I don't think it makes sense to act on important decisions based on assumptions (rather than real experience), maybe it's still worth figuring out what your decision would be.  You don't act on it, but you figure out what you would do, write it down, and make a calendar event for 3 months from now reminding yourself to review how your decision would have panned out.

For example, I posted about how progressive enhancement worked out well a few weeks ago.  It probably would have been worth it to come up with an exact plan for what we would have done if we hadn't deferred the decision.  That way I could look back and easily tell if we would have been right or wrong about all our assumptions.

I'm not suggesting this should be done on every project (that sounds like a waste of time).  Maybe one out of every five projects should be fully planned out even though the plans will be ignored.  That will help validate the effectiveness of the new strategy.  It's easy to get complacent when everything goes smoothly.