Terroir is thought to be an expression of "place" that is considered a characteristic of wine from a particular region, expressed as a flavor of the soil. In theory, wines from the same region will have defining characteristics that will allow you to blindly pick them out from other wines. Oddly, it is a term used for old-world wines (France, mostly) but almost never for new-world wines. The French are fond of pointing out that they have a monopoly on terroir: 'Very good wines are produced in Chile, for example,' says Denise Capbern Gasqueton of Château Calon-Ségur in St Estèphe. 'But they can lack terroir, and terroir is what makes everything. A wine that is well-produced is a good wine, but lacks complexity and other elements to which we are used.' (from Decanter)
That's right, new-world wines are lacking a certain, as the French say, "I don't know what". And they aren't kidding either, they really don't know. They can call it terroir, complexity, minerality, etc. but apparently when you try to measure it, as these English economists have shown, you can't find it. Maybe Terroir is subject to Heisenburg's uncertainty principle of quantum physics. Therefore, I now put forward what I will call Huge's principle of Terroir:
"The more precisely Terroir can be tasted, the less precisely it can be measured"
Tom at Fermentations has had a number of good posts on this lately and I agree with him. I think that a region's wine's inputs must be consistent between too many variables that can overcome "terroir". What I mean by that is that for wines from the same region to have defining, consistent characteristics, you must have: Similar soils, similar drainage, similar trellis configuration, similar water inputs, similar timing for all viticultural activities, similar row orientation, similar slope, similar additions to the soil, similar harvest time, similar maceration, similar barrels, similar aging time, similar blending, similar additions, similar fining, etc. etc. etc.
If you can keep all of the above the same from winery to winery and vineyard to vineyard then you will have something to compare it to. My belief is that the above variables tend to be more homogenous across a given region of the old-world and while that does allow for a better comparison, it is still difficult to isolate which of the above elements contributes to any commonality and which elements are an intrinsic expression of "place".
As Tom points out in one of the above articles, it is not universally accepted that soil characteristics are expressed through the vine and into fruit. What I found highly interesting was an article on an experiment performed by Randall Grahm of Bonny Doon where he basically macerated his wine with three different sets of ground-up rocks to directly express 'minerality' in his Le Cigare Volant.
I do believe that since many new-world winemaking regions tend to produce riper fruit (through winemaking style or through Mother Nature herself) that it can be more difficult to isolate nuances that can get lost in a monster Cabernet from Napa or a Shiraz from Australia. However, not all new-world wines are made this way, and I still believe that if terroir exists, that you would see it expressed when tasting two different wines made by two different wineries from fruit of the same vineyard. I have had the chance to do this before and the difference was so remarkable that I would have never guessed them to have the same origin (or even the same varietal!). Thus, I have to agree with the English economists, that winemaking style (and viticultural inputs) can too easily mask whatever terroir really is.