Posted 16 November 2012
“A medium-bodied wine with aromas and flavours of strawberry, cherry and marzipan. It has a smooth acidity and medium soft tannins.”
“Mendoza, Argentina. 100% Malbec. 2011. 13% alcohol. 4g/l residual sugar.”
Introduction
There’s no such thing as a ‘wrong’ tasting note, but which of the above is most useful? In a digital world, qualitative information is of minimal use. The first note gives us a personal experience (which I believe is too personal to be of use to 99% of other people – see my previous post), whilst the other is data-driven, quantitative data – from which a computer could reference a database, run an algorithm, and return similar wines as recommendations (more about the benefits in a later post).
The power and value of being able to successfully categorise got me wondering whether it’s possible to turn wines into data. Obviously all online stores already do this to enable us to search their catalogues: colour, price, country are the most popular. They are also, in my opinion, the most useless. What do those factors really tell us about the wine? Not a lot. So, which factors are most important?
My top wine factors
Below are the handful of quantitative factors that I think best desribe a wine – along with my reasoning. I’d love to hear if you agree.
Region/Appellation
Someone recently said that country of origin is the most important description of a wine. If we change this to region (or even sub-region) then I agree that it is one of the most important, but not necessarily the most important.
Grapes
Grapes have to be up there: a wine might be from Champagne, but is it made from Chardonnay, Pinot Noir or Pinot Meunier – or a blend? However, there are exceptions such as when the region tells us enough (e.g. Sauternes).
Colour
Although a wine’s region and grape varieties will normally tell us the colour of the wine, there are rare occasions where it needs to be clarified. For example, I know that a Burgundian Pinot Noir is going to be red, but need more information on a Champagne made from Pinot Noir and Pinot Meunier: is it a blanc de noir or a rosé?
Vintage
Vintage informs us how much the wine has aged but also tells us about the weather patterns of the grapes used to make the wine. Although I doubt we’ll ever see a universal coding system to describe the micro-climate that affected a particular harvest, the vintage does provide us with a summary for the region. For example, I know that a wine from Penedès in 2003 was excellent with the exception of Syrah-based wines due to high levels of heat which brought harvest forward. Sure, it doesn’t tell me whether the particular vineyard was thwarted by frost or hail, but it gives me quite a lot of information.
Alcohol level
Frequently a hotly debated topic (especially if Wink Lorch is in the room): many people arguing that alcohol levels aren’t important (nor accurate). But, if the only two things I know about a wine are that it’s made from Pinot Noir and it’s 15%, then I’d be able to guess that it’s a full-on Aussie Pinot (probably Tasmanian) rather than a delicate Burgundy. Therefore, you could argue that alcohol levels are more important than region [cue fierce response!].
Residual sugar
Sparkling wines and any wines with sweetness need to disclose the levels of residual sugar for us to get an idea of what the wine is going to be like.
Conclusion
Nothing beats a personal tasting note to record your experience of a wine, but categorisations have incredible value. Assuming that a database has information within it to interpret the data it is given, then establishing a wine’s key factors allows us to compute recommendations based on the original wine. Some of those factors would definitely include the ones mentioned above, but are there any others that you think would be beneficial?
Of course, the next step is to reduce the key factors into shorthand: “Mendoza, Argentina. 100% Malbec. 2011. 13% alcohol. 4g/l residual sugar” could easily be changed into Mz/Mb/2011/13/4. But don’t expect to see that in Decanter soon!


The trouble with this is that it assumes that people would know what a Pinot from Limari in Chile was like compared with one from Bio Bio. And that all wine was made the same way from these regions. And that people understood how residual sugar translated to the dry wine they had in their glass. So I’m not convinced hard data translates into the experience you might get when you drunk the wine. I’d want something like “crisp”, “soft”, “rich” to give an indication of what I might get. But then it gets subjective. A nice theory though.
Thanks for your comment. And, yes, you are correct.
My next (or next but one) post will go into how it would be put into practice, but it just requires people to say “I liked wine x” and “I didn’t like wine y”; the power of algorithms will do the rest…
I’ll be interested to see how this develops. Can you have I liked it, it was ok, I disliked it? I find black and white very hard. That’s my issue with Wine Demon (apart from them recommending me white zin…)
Very interesting topic, thanks for posting.