The Vine Nerds
May 10, 2015 11:37 PM   Subscribe

But to a growing cadre of A-list winemakers, there’s actionable intelligence in the data. Many of Fruition’s clients are altering their irrigation techniques, turning laggard vineyards into top performers and using far less water than they ever imagined. Along the way they’re extracting lessons that could extend far beyond this rarified corner of agriculture. By gaining insight into the relationships between water, sunlight, yield, and taste, Fruition Sciences is showing the way for farmers of all stripes to increase productivity and quality in a world of shifting weather patterns and decreasing supplies of freshwater.
posted by the man of twists and turns (18 comments total) 15 users marked this as a favorite
 
Final project in the Stats 315A class at Stanford (Elements of Statistical Learning: The Book: The Class, taught by one of the authors usually) is basically, "here, have this big ol' dataset of wine data, here's the model DV we want you to get, go to town with the machine learning". I remember pulling out all the random forests.

Classic sort of p >> n dataset (dimensionality way bigger than the cardinality of the data), which was in large part the impetus of the book. And I suspect these guys with the actual wine problem, their problem is a big ol' p >> n sort of dealio also. In which case, sparsification and/or smoothing are essential parts of their problems: compare to the sparsification of the data the Cybersyn people recommended for the p >> n problem of an economy (point 4 in that article).
posted by curuinor at 11:57 PM on May 10, 2015 [2 favorites]


"actionable intelligence"

I'm not sure if there's any way to use this phrase without sounding completely tone-deaf. What editor said "yeah, sure, let's associate agricultural research with tragically ignored terror memos?"

The past 14 years have been bad enough, but it's especially galling to see unironic absorption of the vocabulary of incompetence.
posted by lumensimus at 12:26 AM on May 11, 2015 [2 favorites]


Are we doing the "words can only have one meaning and that meaning never depends on context" thing? Business intelligence isn't exactly a new concept (wikipedia claims first use was 1865).
posted by effbot at 12:38 AM on May 11, 2015 [1 favorite]


A quick search through the Google Books American English corpus for the exact phrase "actionable intelligence" shows the following results:
1810 0
[...]
1970 63
[...]
1990 43
[...]
2010 842
Similar corpora show a similar bias.

Business intelligence is not new, but this phrase entered the contemporary vernacular through a single vector. Just an editorial comment -- I have high hopes for fancy informatics!
posted by lumensimus at 12:52 AM on May 11, 2015 [6 favorites]


The milspeak "actionable intelligence" irritated me right into RTFA before I even opened the thread, so there's that. I think maybe the reporter has grown so accustomed to the struggle to describe application of evidence in those fields where observation is the only way to collect data ("sports, advertising, medicine, policing, and real estate" - shout out to my field epidemiology FTW), he forgot that we already have words to describe the standard use of the scientific method. These geeks are conducting experiments and applying their findings. In real-time, if you absolutely must.
posted by gingerest at 12:54 AM on May 11, 2015


[Okay, folks, let's call the terminology annoyance fully registered, and forgo more derailing on that? Thanks.]
posted by taz (staff) at 1:02 AM on May 11, 2015 [5 favorites]


That's an interesting measurement. I wonder how universal it is that if I have a little heater and a temperature gauge some distance away, and I know a) the current dampness, and b) the time for a certain rise in temperature to propagate, then I can reset, let things settle, then turn on the heater again marking the time it takes for the same temperature change to occur, and from that figure how much less or more dampness there is from baseline.

Would that work, for instance, embedded in the concrete foundation of a home?
posted by StickyCarpet at 3:35 AM on May 11, 2015


This is the ideal crop for new methods, it would not be the first time winemakers have changed their watering methods. The elite "fine" wine The growers know quite well that differing techniques can improve the flavor of a grape. Many of the best grapes are grown in stressful conditions. High volume growers may not be as interested unless the irrigation techniques are demonstrated to be profitable.
posted by sammyo at 5:26 AM on May 11, 2015


I anyone else sick of the usage of 'nerds' to signify people using any kind of information technology in a way that, to the journalist, seems hard to understand or explain? I know that it's supposed to have been 'reclaimed', but I think it's way overused and still has a stigma attached, and is basically just lazy writing.
posted by signal at 5:40 AM on May 11, 2015 [3 favorites]


We've been making, trading and drinking wine for at least ten thousand years, across all manner of climate change and every conceivable type of economic, cultural and political upheaval. Lots of innovation has gone into the practice throughout, because by golly we do like the stuff.

If there's any new technology or technique that could be applied to viniculture but isn't - well, there's the story.
posted by Devonian at 5:50 AM on May 11, 2015


Many of the best grapes are grown in stressful conditions. High volume growers may not be as interested unless the irrigation techniques are demonstrated to be profitable.


As you imply, high volume growers don't want stressful conditions--they want maximum yield. The same analytical techniques should be just as useful as they are for maximizing quality though, as they give the grower a better idea of the state of the plants in both cases. The difference is what the grower chooses to do in the face of that information: add more water to plump up the grapes, or restrict water but keep the plants on the right side of death. If anything, it should be easier to maximize yield with this technology, as what you are trying to achieve is more quantifiable. Maximizing quality relies on variables that are much more subjective, like flavour, and scientific data will have less value in making those calls.

Assuming it works, this technology is like a meat thermometer: it gives you more information that can be used to get better repeatability on your results. There is still plenty of room left for art in the way you prepare things.
posted by cardboard at 6:49 AM on May 11, 2015 [2 favorites]


I anyone else sick of the usage of 'nerds' to signify people using any kind of information technology in a way that, to the journalist, seems hard to understand or explain? I know that it's supposed to have been 'reclaimed', but I think it's way overused and still has a stigma attached, and is basically just lazy writing.

I take it as a signal that either the writer is dumb or he thinks his audience is or both. Bazinga!
posted by srboisvert at 7:03 AM on May 11, 2015 [1 favorite]


But to a growing cadre of A-list winemakers, there’s actionable intelligence in the data.

Each successive noun and adjective in that sentence raises my blood pressure by ~5%.
posted by Behemoth at 7:10 AM on May 11, 2015 [5 favorites]


many many very high quality grapes are dry-farmed anyway. Actually in France nearly all AOCs expressly prohibit irrigation, although I think they have made exceptions for years so excessively dry its not a question of yield so much as survival.
posted by JPD at 7:49 AM on May 11, 2015 [3 favorites]


This is fantastic. I found some information about water-stressing vineyards and dry-farming in a college course where we were asked to investigate ways to reduce water usage in a field of our choosing. As JPD says, it's not novel, but I get the feeling that a lot of growers in the US aren't as confident in dry farming as European counterparts who have been running the same vineyards for hundreds of years.

So in the US, you'll see the shift from cattle to grapes significantly impacting groundwater resources, made worse because private wells, as are often used for ag, are not monitored.
posted by filthy light thief at 7:57 AM on May 11, 2015


In looking around for how this company's sensors work, I found this free article on how to build sap flow sensors that work on the same principle, and what that principle entails. Cheers!
posted by wormwood23 at 4:15 PM on May 11, 2015 [1 favorite]


The economics of this are pretty interesting: all of these precision ag technologies have diminishing marginal utility (sensors, spectral imaging, just going out to look, etc.), but still need to result in growing marginal profits, so there's going to be an equilibrium for any given crop/location/set of technologies where it's just not economic to do things any fancier. Fancy wine is so value added that vintners can throw a lot of tech even at individual vines to really get the best performance out of them.

Meanwhile, the same utility curve is at work on the far end, too, where even very cheap precision technologies can make a poor farmer a lot more profitable (or at least spend a lot less time wasting effort for the same amount of profits).
posted by wormwood23 at 4:29 PM on May 11, 2015


I was surprised to read how much some of those vineyards are relying on irrigation. Around here most use irrigation to get the vines established and then use either no or very little irrigation in order to keep the vines stressed. There are juice vineyards in the area, and they of course irrigate like mad, but that is because they want maximum yield, not maximum flavor.

The sensors sound neat, but there have been ways of measuring much the same thing using soil sensors and field weather stations for a long time; it sounds like a good part of what they are offering is not so much the sensors themselves but the data collection and interpretation services. We are on the cusp of starting to fully apply cheap processing and cheap sensors to a lot of agriculture questions that have been traditionally solved in ad hoc ways, but with so much money at stake farmers want the quantifiable results and A/B testing that data allows.
posted by Dip Flash at 5:49 PM on May 11, 2015 [1 favorite]


« Older “There are amazing images of all these tortures of...   |   Librarians as privacy warriors Newer »


This thread has been archived and is closed to new comments