subscribe: Posts | Comments      Facebook      Email Steve

On “blindfolded monkeys” and blind tasting



I love graphs like this.



It illustrates, in stark honesty, the multi-year stock performance of the S&P 500 compared with a randomly-selected portfolio that could have been picked “by a blindfolded monkey.”

Those, at least, are the words of the person who wrote the article, “A Random Way to Get Rich,” which appears in a summer, 2014 supplement on “Money” from the Wall Street Journal.

The article’s point—which investment advisors will not like—is that “If your money manager hasn’t managed to beat some random stocks…you might want to start asking some questions.” I don’t suppose it will come as a shock to anyone that managed stock portfolio accounts will, often as not, perform more poorly than a portfolio chosen by, well, a blindfolded monkey. Every time I see those ads on T.V. for companies that will “help you retire with confidence,” a wave of despair washes over me, for I don’t trust these companies any more than I trust, well, a blindfolded monkey. After all, their promises of “security” were completely undone by the Recession. But there is a point to be made here, and that concerns professional wine tasting.

You have heard me say time and time again that the only wine review you should place your trust in is one that was conducted blind (albeit, not by monkeys). This is because of the nature of human psychology. If a financial analyst (I love that term, which conjures up images of brilliant, altruistic M.B.A.s working for no reason other than your financial security) imagines that he or she is being strictly objective in the choice of particular stocks, that M.B.A. ought to take a remedial course in freshman Psych 101. The stock market is, by definition and experience, a place of unpredictability. It is so random, so utterly un-analyzable by any human device yet conceived, that the 1905 description of Brownian motion, by Albert Einstein—for which he won the Nobel Prize, and which went on to become a conceptual base of quantum mechanics—could well apply to market fluctuations.

Randomness, too, characterizes the tasting and reviewing of wine. While wine reviewing may appear to be, or approach being, an exact science, of course it is anything but; and by this, I do not mean to undercut the role of the wine critic, which was a job that paid my bills for many years, and which I thoroughly enjoyed. No, I mean only to suggest that, if you consistently taste wines blind, wrapped in their little brown paper bags, you will find results quite different from those published by the majority of famous magazines and newsletters.

Correction: let me rephrase that. You might or might not find results that might or might not be quite different. In fact, in one parallel universe, if you taste enough wines, blind, for a long enough period of time, at least one of the results will look exactly like any given edition of The Wine Advocate. But an infinitude of other possible results will be different, sometimes only slightly so, sometimes significantly. That is the Heisenbergian truth of random results: they are random precisely because they are unpredictable, an aspect of reality that is hard-wired into the fabric of the universe.

Some conservatives lament theories of the random nature of outcomes. They say that experience proves otherwise—a car, driven at high speed, will not pass through a brick wall, but will smash into it with deadly results, despite the slight relativistic possibility that it will in fact emerge unscathed on the other side. So, these anti-relativists argue, it is ridiculous to think that anything other than our well-understood cause-and-effect universe could be “real,” except, possibly, in some fantastic mathematical sense.

This is true as far as it goes in our macro world of cars and brick walls. It is patently untrue when it comes to the anything-goes micro world of sub-atomic particles. And while the human mind seems neither to be part of the macro world nor of the micro world, it does in fact have more in common with the latter; for it is a “black box” into which others cannot peer, and of which the owner himself may be unaware, in terms of its particulars. Tasting wine openly, with full appreciation of its origin, is in fact contamination of the black box’s insides: you cannot do it without discombobulating the entire process. Thus, you cannot call open tasting “real,” since it represents a serious interference with the reality of what a wine actually tastes like. In the famous thought experiment called Schrodinger’s cat, one is confronted with a paradox: “This poses [says Wikipedia] the question of when exactly quantum superposition ends and reality collapses into one possibility or the other.” Open tasting illustrates this paradox: It is utterly impossible for the outsider to know, or even to begin to comprehend, what the thought process was of the reviewer, whose mentation may (or may not) have been influenced by any one, or combination of, multiple factors. Therefore, one cannot say that the reviewer’s conclusion is “real,” except within limits, and even then subject to faith and belief. One can accept the conclusion—it may influence one’s own conclusions and behavior—but one cannot assume that the conclusion itself has validity, at least in the scientific sense of being replicable.

Should therefore we entrust our wine reviewing to “a blindfolded monkey” or indeed a team of blindfolded monkeys? No; that would be a logical fallacy. But the realization that a random portfolio of stocks outperforms the S&P 500, which is the basis of so many mutual fund portfolios, should alert us to the fact that winetasting results from open tastings may also not be the best source of information; and this in turn should put the results of any such tasting into perspective. This is a revolutionary statement: if everyone subscribed to it—tastemakers, gatekeepers, consumers, producers, the media—the wine world as we know it would turn completely upside down. Nobody likes disruption, though, which is why the status quo is unlikely to change anytime soon, despite the burgeoning presence of Millennials who, it is said, are going to upset every apple cart there is. They should, in the case of formal wine reviewing; but they are unlikely to, because (speaking of apple carts), the fruit never falls far from the tree; each Millennial is more like his or her mother or father than one may care to admit. Still, go back and take another look at the chart. If instead of AAPL, CAT, XOM and NFLX you substituted Lafite, Romanée-Conti, Yquem and Screaming Eagle, would that cause you to reconsider your reliance on professional critics, now that you understand how perfectly random these things are?

  1. Bob Henry says:

    For Warren Buffett’s personal investment advice (disdaining stock pickers), see:

    If you wish to learn more about randomness, read Caltech lecturer Leonard Mlodinow’s book titled “The Drunkard’s Walk: How Randomness Rules Our Lives,” which includes a chapter on the fallacy of wine scoring scales.

    And see his essay:

    From The Wall Street Journal “Weekend” Section
    (November 20, 2009, Page W6):

    “A Hint of Hype, A Taste of Illusion;
    They pour, sip and, with passion and snobbery, glorify or doom wines.
    But studies say the wine-rating system is badly flawed.
    How the experts fare against a coin toss.”


    And for the science wonks, he has co-authored two books with the esteemed Stephen Hawking.

  2. I must be dreaming. Did you just say that your lifelong profession–the one that made it possible for you to get your nice new job and gave you the background to be an enjoyable bloggist–is less predictable than selecting wine to drink through a bunch of monkeys?

    Admittedly, neither stock-picking nor wine criticking are purely scientific endeavors. But you used to argue that experience mattered, that track records mattered in wine review.

    Have you now given up on those ideas?

    I hope I am dreaming.

  3. Charlie, experience does matter. My post today is meant to suggest that, if you give the same wine multiple times to the same professional reviewer, tasted blind, the results will be random. And the more often this experiment is repeated, the more random the results will be. Do you argue with that–or do you acknowledge that the scores and reviews will be different?

  4. Argue with?

    I have a different view of the world. I think my results and the results that get published in Connoisseurs’ Guide are substantially more replicable than random.

    I also think the same of your reviews. Indeed, your reviews and mine correlate way better than random. That is not conjecture. It is a fact.

    The comparison to monkeys and the use of the term random are, in my humble opinion, unnecessarily harsh and more than a little pejorative.

    Random means just that. It means that if you gave me the same set of twelve wines blind every day for a week, the results would not have any more correlation than that achieved by a bunch of monkeys.

    I cannot believe that you mean that. I can accept that you do, and that you include yourself in the category of random reviewer of no value. But I don’t agree with you about your reviews if that is your meaning. And I am willing to be tested for randomness if you are suggesting that my reviews would be no more accurate than a bunch of monkeys.

  5. One final comment I should have added.

    Your final question: “Would the scores and results be different” has no meaning as asked, and that problem is perhaps at the heart of what we are discussing here.

    Random does not equate to different. Random is what the monkeys would produce. Different by small measure most of the time is what I think you or I or Laube or Parker or Tanzer would produce.

    That does not say that we could reproduce results precisely. But the inability to reproduce results precisely does not make them random in the true meaning of that word.

  6. Charlie, when you and I review a wine, we taste it one time, maybe twice if we have two bottles. I suggest that if we, or any critic, were to taste the same wine multiple times, that scores would vary. And how much they would vary is tied into the nature of statistical analysis. The more frequently we tasted the wine, the more variance in scores there would likely be. So my position is consistent with previous things I’ve written. The take-home messages are (a) consumers should accept wine reviews for what they are, imperfect but useful, and (b) wine reviews that were not done blind have less objective value.

  7. Bob Henry says:


    When I critically evaluate (say) a red wine for (1) a wine store client’s merchandise mix or (2) a restaurant client’s wine list, I follow a rigid guideline of evaluating components originally introduced to me by the U.C. Davis 20-point scale (which I have revised to a 100-point scale):

    Appearance (up to 5 points)
    Color (up to 5 points)
    Off-odors (up to negative 10 points)
    Total acidity (up to 5 points)
    Sweetness (up to 5 points)
    Bitterness (up to 5 points)
    Aroma (up to 20 points)
    Bouquet (up to 10 points)
    Body Weight (up to 5 points)
    Overall Acidity (up to 10 points)
    Flavors (up to 15 points)
    Overall Quality (up to 15 points)

    Note that “Aroma” and “Bouquet” out score “Flavor.” That attests to the importance of the olfactory sense in wine evaluation.

    By adhering to this rigid “check off the component” guideline, my evaluations vary narrowly upon repeat exposures. My procedure doesn’t allow for caprice or whim.

    And I welcome the opportunity to be tested on my ability to replicate scores tasting single or double blind, using my 100-point scoring sheet and the Impitoyable Le Taster wine judging glass (likewise used by Robert Parker and James Laube).

    I leave you with this quote from Robert Parker (source: an issue and page I can’t recall from The Wine Advocate circa 2002):

    “ . . . Readers often wonder what a 100-point score means, and the best answer is that it is PURE EMOTION that makes me give a wine 100 instead of 96, 97, 98 or 99. ”

    ~~ Bob

  8. Which is why I use group scores from Regular People. I once ran the same wines by 16 members of my wineclub in three different flights varying the sequence each time, and took the average/median. Individual scores did vary a bit but the group score didn’t (maybe 3.4 vs. 3.5). Much more reliable.

  9. Randomness requires no bias, and no preferential force of any kind. When it comes to tasting wine, one’s palate dulls quickly and in fact the more you taste the more random the results become because after a while, it all tastes similar. My palate is discriminating for about 4-6 tastes and after that it all quickly dulls. Preference has been removed by repetition.

    We tend to naturally fill in the blanks with other visual clues so a blind tasting (better yet double blind) is always needed to make this about the wine itself and nothing something external. To that we all agree

    Variations can still be held to a certain range. It’s never purely random. It’s always kind of silly to state that a wine ranks 91 as opposed to 93 or 88 because it varies from day to day, bottle to bottle, even for the same taster. But a wine probably wouldn’t vary from a “D-” to an “A+” for any given taster. Perhaps then that’s a more accurate scale. Ball park numbers are not lost in useless precision.

  10. Mr. Merle–

    This must be my day to disagree with my friends. When you say that individual scores varied a bit, you are making my point.

    “Vary a bit” is not a set of random results. Random results would cover the waterfront. It means that if you gave a qualified taster twelve wines twelve times that every wine would average the same score as all others. That is what random results do when tested many multiples of times.

    “Vary a bit” is nowhere near the randomness that Steve has postulated with which I disagree as an overstatement of the likely result. Remember, Tom, that random is random.

    Steve’s clarification, which represents an extension of his comments today, is much closer to what he means than what he wrote–in my repectful opinion, of course.

  11. Bob Henry says:

    A salient quuote from that same Wall Street Journal essay by Caltech lecturer Leonard Mlodinow:

    “I [Mlodinow] did email Mr. Parker, and was amazed when he responded that he, too, did not find [Fieldbrook Winery winemaker, scientist and retired statistics professor] Mr. [Robert] Hodgson’s results [on California State Fair wine judgings and medal awards] surprising. ‘I generally stay within a three-point deviation,’ he [Parker] wrote. And though he didn’t agree to [Falcon Nest vintner] Mr. [Francesco] Grande’s challenge [to submit to a controlled blind tasting], he sent me the results of a blind tasting in which he did participate.

    The tasting was at Executive Wine Seminars in New York, and “consisted of three flights of five wines each. The participants knew they were 2005 Bordeaux wines that Mr. Parker had previously rated for an issue of The Wine Advocate. Though they didn’t know which wine was which, they were provided with a list of the 15 wines, with Mr. Parker’s prior ratings, according to Executive Wine Seminars’ managing partner Howard Kaplan. The wines were chosen, Mr. Kaplan says, because they were 15 of Mr. Parker’s highest-rated from that vintage.

    “Mr. Parker pointed out that, except in three cases, his second rating for each wine fell ‘within a 2-3 point deviation’ of his first. That’s less variation than Mr. Hodgson found. One possible reason: Mr. Parker’s first rating of all the wines fell between 95 and 100 — not a large spread.”

  12. Charlie,

    I don’t wish to wade into the nuances of the meaning of randomness. Rather I prefer to focus on the consistency of scores. Because of a small individual variation from flight to flight, I find the group score superior in part because you can identify wines that please a broader array of palates and therefore deserve more interest, but also because the ‘wisdom of crowds’ identifies results remain constant in multiple tastings.

  13. Bob Henry says:


    A clarifying question to this statement:

    “I once ran the same wines by 16 members of my wineclub in three different flights varying the sequence each time, and took the average/median. Individual scores did vary a bit but the group score didn’t (maybe 3.4 vs. 3.5). Much more reliable.”

    Did you pour from the identical bottle in each of the three flights, or from the same “nominal” wine (same producer, same vintage, same grape variety)?

    If you used different bottles of the “same” wine, then you introduced an uncontrolled variable into the tasting “experiment” that could account for an array of scores (place rankings) within each of the three flights.

    I once hosted a wine party in which I brought four bottles of the “same” wine: 1990 Mt. Eden “Old Vines” Cabernet Sauvignon.

    I owned the case that comprised bottles #1201 through #1212. Each bottle was serial numbered by the winery. Each bottle ostensibly filled sequentially while going down the production line.

    (I stored the 12-pack at home with the bottles laying horizontal to keep the corks moist. I “liberated” my first four bottles the night of the tasting event.)

    I brown paper bagged bottles #1201 and #1202 and #1203 and #1204.

    Served them in that same serial number order to some very experienced wine industry professionals and wine collectors.

    And asked them just one question: “Are the wines the same or different?”

    The collective response: They are all northern California Cabernets, all seemingly from the same vintage, but all subtly different.

    (And when I tasted the wines after the “vote” so as not to influence the discussion, I concurred.)

    Winemakers have puckishly observed that 12 bottles of wine in a single case is like having 12 separate children.

    The corks vary in the tightness of their seals (and some might be afflicted with TCA).

    The wine might be different if the winery brings in the harvested fruit by different vineyard rows (“terroir”) and barrel ferments and stores it separately.

    The wine might be different if the winery bottles from one barrel at a time (the wine in the some bottles could be a production line cuvee from two discrete barrels, as the first barrel is drained and the second barrel “tapped”).

    Lots of variables that are uncontrolled . . . leading to variations in a judge’s “take” on a “same” bottle of wine.

    ~~ Bob

  14. Mr. Henry,

    Same bottles, same day, different order. I would also like to do a similar arrangement with the “same” wines (different bottles) on three different days. But keep in mind the kind of findings you arrived at came from experienced wine lovers–i.e., enthusiasts, connoisseurs. They can pick up subtle variations. My gang are just wine drinkers who usually have vino with dinner.


  15. bob- wow. That’s a broken scoring system. You’re giving points for color? You have 2 categories for acidity? Please differentiate bouquet and aroma.

    Steve- compelling, but 2 questions for you-a) do you taste within a category and note prior to tasting? I.e. Vouvrays, barolos, etc. b) can you discuss the effects tasting wines has on each other and quirks that may make a wine stand out given equal footing, but that may not indicate quality? I’m looking at a stylistic outlier unexpectedly redefining a category.

  16. Dear Adam, first of all, I no longer review wines [except those of Jackson Family, for internal purposes]. And I never professionally reviewed any but California wines. Having said that, I did used to taste “within a category”, e.g. Napa Cabs, coastal Pinot Noirs, etc. I do think it’s vital to taste in flights of similar wines, because this puts each wine into context without which it’s very hard to gauge quality. I don’t think there’s any firm dividing line between style and quality, if I understand your question correctly. What lifts a wine above a certain standard into greatness is very hard to put into words.

  17. Jason Smith says:

    So is the wine red or white? We don’t know until we let the bottle out of the bag.

  18. Bob Henry says:


    Color is a component of the historical UC Davis scale. Everyone gets it right these days, so that’s a “gimme.”

    Likewise appearance. Another “gimme.”

    There is “overall acidity” distinct from (in red wines) astringency (tannin acids).

    On my 100-point scale for white wines and sparkling wines, there is no astringency component (though it can be “noted” in rosé sparkling wines and Champagnes), hence no separate points.

    Aroma is the smell contributed by the grape varieties (e.g., vegetables or fruits).

    Bouquet is the smell contributed by fermentation (e.g., yeast; “butter” from malolactic fermentation; “hazelnut” from sur lees contact; “toasty” or “charred” from oak barrel contact), and then there’s the “tincture of time” . . . as old wines smell distinctly different from young wines.

    Hope that clarifies for you.

    ~~ Bob

    Some great reference books:

  19. Bob Henry says:


    I saw your comment this morning just as I was running out the door to service a wine industry client, and thought it was worthy of pausing long enough to dash off a reply.

    So not having the time at the moment to make this elaboration, let me do so now.

    The bitterness component in a red wine is also the astringency component, attributed by tannic acid.

    Distinct from the judgment about overall acidity.

    So yes, there are two scores assigned to acidity in a wine. (See U.C. Davis scale academic monographs.)

    This is the first time I have seen your name affixed to a comment on Steve’s blog. So welcome to the discourse.

    Tell me more about your interest in wine.

    Are you a member of the wine trade (as I learned recently about Bill Haydon), or a wine enthusiast consumer with a passion for the beverage?

    ~~ Bob

  20. Bob Henry says:

    Adam, et. al.:

    Steve blogged recently about “experts [read: wine critics] getting stuff wrong” and encouraged them to fess up to their misteaks . . . um . . . mistakes.

    My early wine mentor (Los Angeles Times wine columnist) Robert Lawrence Balzer instilled in us wine appreciation course “children” this philosophy:

    “Confession is good for the soul.”

    (Robert was trained as a Buddhist monk in Cambodia. Obit:

    So here is my mea culpa.

    Nothing that instills humility faster than in seeing one’s slap-dash or post-midnight brain-dead innumeracy and wayward comments misstate the facts.

    (Unlike the character Miles in “Sideways,” I can’t mitigate my blunder by confessing I was drinking and dialing. But let me observe that academic studies have found that sleep deprivation produces diminished motor skills and fuzzy thinking akin to intoxication.)

    So once again with clarity, let me proffer this correction by volunteering a link to the U.C. Davis 20-point scale:

    The U.C. Davis scale assesses acidity THREE ways:

    Volatile Acidity (vinegar)
    Total Acidity

    On my modification of the Modified U.C. Davis scale, I have reworked it to award more points to the hedonic aspects of a wine — aroma and bouquet and flavor — and reduced or eliminated some “academic” components that one rarely encounters in contemporary wines.

    Such as:

    Volatile Acidity (vinegar)
    Bitterness and Astringency in white wines (*)
    Sweetness in “dry” red wines [a comment that no doubt will bring out the “boo birds” citing “fruit bomb” California red wines]

    It is the hedonic aspects of a wine that garners one’s “vote of the wallet.” Not the “gimmes” like Appearance or Color.

    I don’t expect “dry” red wines to be sweet, so I have reassigned that “Sweetness” 1 point on the 20-point scale (5 points otherwise on a 100-point scale) to other hedonic components.

    In the U.C. Davis scale as well as on my modified scale, a red wine’s “Bitterness” is judged separately from its “Astringency.”

    I hope that this is a better clarification.

    ~~ Bob

    (*Some pink or roseate-colored “white” wine grape varieties — e.g., Pinot Gris and Gewurztraminer — have low level skin tannins [Aside: which is why I have floated the trial balloon idea of fermenting them using carbonic maceration — just like Beaujolais — to reduce the extraction of skin tannins. An experiment that I have persuaded some winemakers up in Oregon to conduct with their Pinot Gris.] Other white wines have low level tannins from contact with new oak barrels.)

  21. Bob Henry says:

    For the wine wonks who want to do a deeper dive into this subject, use this link to a chart comparing the various extant scales:

  22. I don’t know how to make this any clearer, Charlie. I said that if you give the same wine to the same critic multiple times, blind, he or she will rate it differently. And the more often you repeat the experiment, the more widely the scores (or puffs, or whatever) will be over time. That is not the same as saying that experience doesn’t matter. It’s just allowing for a standard deviation.

  23. Bob Henry says:

    Steve and Charlie,

    Why not invite Tom Hill — a research scientist and frequent commenter on this blog — to draft a quick “tutorial” on scientific sampling and randomness?

    Or invite Caltech lecturer Leonard Mlodinow (who has written on both wines and randomness) to likewise elucidate the subject?

    ~~ Bob

  24. Steve,

    “the more often you repeat the experiment, the more widely the scores (or puffs, or whatever) will be over time”

    You need to review some basic statistics.

    This is the exact opposite of what should happen, if you assume that all scores are normally distributed around an average score, which would be a given critic’s “true” score for that wine. (True in the sense that this is the critic’s rating of the wine, not true in any absolute sense that would apply to anyone else).

    If you do a bit of reading about statistics, you will realize that the reliability of the score will get better with more scores. (sidenote – This is the cause of “regression to the mean,” and why we would expect an extremely low or high score to be moderated with further ratings, on average.)

    The very first score a critic gives to a wine will likely be close to the average (“true”), but it could be very far. With more and more repetitions the scores will begin to fill in the bell curve and the “true” score will emerge.

    So it is exactly wrong to say that the scores will become more variable with more repetition, although of course the chances of getting one outlier will increase (but with more scores, the effect of that outlier on the average will be minimal).

    In fact, if critics submitted to lots and lots of retesting, we could measure the standard deviation of their scores, and actually have a good measure of a critic’s reliability. Then scores would start to look like polls: “This is a 92pt wine +/- 2 pts, 95% confidence). Which would actually be a much more honest way of reporting the score.

    Unless you are saying that scores are truly RANDOM, as in thrown at a dartboard, in which case… why would anyone ever care about a score at all?

  25. Steve–

    What Dan said.

  26. Bob- industry professional for 17 years. Commented previously, but not obsessively. Thanks for the welcome-what’s your story? Do you consult? Why the Rube Goldberg ratings system?

    All due respect to yours and some other ratings systems- any system that tries to organize the disparate qualities of all wines into a unified quality theory is effectively ruined by its attempt to find the lowest common denominator. This is not to say ratings aren’t useful, see below. Further, in many wines, color is completely irrelevant (current releases mostly). Do you account for winemaker tricks like mega purple, acidulating, watering back? Do you look at tech sheets to analyze total acidity and the breakdown of acids? It seems like you’re caught between worlds. Steve, and others critics like Parker, Laube, etc, may have a bias, but it is a predictable bias, like all critics of subjective things (art, food, music). These critics are therefore, knowable and any experienced individual should be able to quickly calibrate against them. Trying to bring science (or pseudo scientific methodology) into the process loses the point of what make wine great. The best we can do is try to bring consistency to the process and some general order.

  27. All – commenter Dan has it right. Anyone who has studied statistics and reviewed any scientific studies in this space could only logically conclude that wine reviews do not follow a random distribution (wine competition medals from some competitions, such as the CA State Fair, do follow a pattern we’d expect from random data, however). What’s more, the comparison to the lagging performance of actively managed mutual funds is really not appropriate here. Those funds lag the s&p 500 because of obscenely high costs dragging down any returns above those of the market, which is why anyone with half a brain would instead invest in index funds (but that’s another topic altogether 🙂

    There probably isn’t a performance drag equivalent in wine reviews, apart from temporary circumstantial stuff like palate fatigue, etc.

  28. Bob Henry says:


    “. . . what’s your story? Do you consult? Why the Rube Goldberg ratings system?”

    I consult for wine stores and restaurants in Los Angeles. A “Have wine glass/Will travel Paladin.” I served previously as a judge for Bon Appetit magazine’s monthly “Tasting Panel” hosted by Anthony Dias Blue. And served previously as a country fair wine judge.

    I have organized and hosted over 100 sit-down “single blind” winetasting luncheons here in Los Angeles, attended by the wine trade, the wine press and wine collectors. A partial list of the “themes” and “Top Three Preference Voting” results:

    My “Rube Goldberg ratings system” is taken directly from U.C. Davis — “tweaked” to reflect contemporary winemaking practices: more points for hedonic components; less or no points for components that modern-day winemaking has solved (e.g., Volatile Acidity).

    Color in a young wine is less of a concern these days than in the past. But an off-color can be an early sign (before engaging one’s olfactory sense) that the wine has been “cooked” or is “long in the tooth.”

    (Aside: I dropped a private e-mail to Adam attaching my 100 point scale notetaking sheet for red wines. There’s also one for white wines and sparkling wines. All three drafted in Excel to use the arithmetic function to sum up component scores. As a consequence, it reproduces poorly as a Word program-like text.)

    So my scale is less Rube Goldberg and more an homage to U.C. Davis research pioneers Maynard A. Amerine and his cohorts.

    Company I’d be honored to keep.

    ~~ Bob

  29. Bob Henry says:

    Let me throw a little more “publicity” behind a book I touted above:

    The late Canadian wine writer Andrew Sharp wrote a wonderful guide on this subject titled “Winetaster’s Secrets.”


    Robert Parker, Jr. review:

    “An extremely well written book with the most informative and perceptive chapters on wine tasting I have read. This is the finest book for both beginners and serious wine collectors about the actual tasting process — lively, definitive and candid.”


    ~~ Bob

  30. TomHill says:

    Bob sez:”Why not invite Tom Hill — a research scientist and frequent commenter on this blog — to draft a quick “tutorial” on scientific sampling and randomness?”

    Sorry, Bob…that’s something waaay out of my area of expertise. Not something I’m really competent to speak to. Amerine & Roessler’s book is half devoted to statistics & wine sampling. It’s a very good read for Amerine’s material…but a bit tedious on Roessler’s statistics stuff.

  31. Bob Henry says:


    As a research scientist, you are immersed in a world where mathematics is considered a “second language,” with such terms as normal distributions, means/medians/modes, standard deviations, confidence levels and confidence intervals invoke as “tools of the trade” . . . hence my thinking of you for a “brush up” without having to write a Wikipedia entry.

    So thanks again to Dan for the comment.

    Yes, Amerine & Roessler’s book can be a bit of a hard slough on the “dry” stats material. For the interested readeer, having a nodding acquaintance with the nethodology and vocabulary imparted by the tome is worthwhile.

    Likewise Peynaud’s tome.

    But Sharp’s book gives you insights found nowhere else. (Example: how does the color of the paint on the walls of a tasting room affect your perception of a wine? Sharp will explain . . .)

    ~~ Bob

  32. Bob Henry says:


    Make that “a hard slog.”

    (This seems to be my week for fessing up to errors.)

  33. Dan Fishman says:

    Hi Bob,

    While I am definitely not a mathematician I did get a fair amount of basic statistical training in graduate school before leaving for the wine world (that’s statistics for research psychology, so certainly more practical than theoretical and not very complex). I am actually working on a blog post about how I think simple statistics could be applied to wine ratings for the new Stemmler blog that is going to launch in July. I hope you will check it out when I post and share your thoughts!

  34. Bob Henry says:

    Steve, Dan, Tom et. al.:

    It’s been quite a while since I last closed the cover of my college stats textbook . . . so I feel ill-equipped to compose a tutorial.

    What I do recall is that for sampling to be “scientific,” you have to have at least six data points. The more a wine critic submits to repeated tastings of the “same” wine and the subsequent submission of scores/forced rankings, the more accurate will be the “taking the measure” of the wine. So for wine critics, “rinse and repeat” is the exhortation.

    (As for my current knowledge of statistics, the phrase “use it or lose it” became a self-fulfilling prophesy when I entered the business world. Fortunately, I had the benefit of Caltech-trained engineers as my work colleagues to do that heavy lifting.)

    Coincidental to this discussion is a book review from today’s newspaper . . .

    Excerpts from The Wall Street Journal “Review” Section
    (June 14-15, 2014, Page C6):

    “Math for the Nonplussed”

    Book review by Mario Livio
    [an astrophysicist at the Space Telescope Science Institute and the author of “Brilliant Blunders.”]

    “How Not to Be Wrong:
    The Power of Mathematical Thinking”
    By Jordan Ellenberg
    (Penguin Press, 468 pages, $27.95)


    . . .

    Mr. Ellenberg’s key point: Mathematics is not some strange language used by a few single-minded experts. Rather, it is a powerful extension of our common sense, one that anyone can employ to tackle real-life problems.

    . . .

    Mr. Ellenberg truly shines in his extensive discussion of probability. One of the most important take-home messages is that, given enough trials, improbable things happen a lot. In the words of the statistician Sir Ronald Fisher: “The ‘one chance in a million’ will undoubtedly occur, with no less and no more than its appropriate frequency, however surprised we may be that it should occur to us.” Along the way, Mr. Ellenberg shows us how mathematics provides us with powerful means by which we can discuss the uncertain, taming it to the point where we can firmly assert: “I’m not sure, this is why I’m not sure, and this is roughly how not-sure I am.”

    . . .

    Will this book teach you how to never be wrong? Not really. In fact, it is fine to err if your mistakes are the consequence of taking calculated risks or of creative thinking. This book will help you to avoid the pitfalls that result from not having the right tools. . . .

    [And for those who wish to know more about mathematician-turned-professor/author Jordan Ellenberg, he “penned” an “Op-Ed” piece for The Journal recently:

Leave a Reply


Recent Comments

Recent Posts