wtbell

Question for you Rhone experts: You mentioned the Rasteau area above -- I recently tried a Vin De Pays (Domaine du Trapadis 2002) from that region. It was inexpensive and tasted pretty good so I did some research into the producer and other wines from the area. Domain du Trapadis is a small operation that grows a wide variety of grapes that it blends in its various cuvees. It occasionally, I understand, produces a Vin De Pays -- the most recent was 2000 which was 100% Grenache. Their cuvees are usually a blend of Grenache, Carignan, Syrah, Cirsault, and Mourverde grapes. If I'm not mistaken, 2002 was a rough year for the region. Many lost their entire production. Domain du Trapadis apparently did not release one of the branded cuvees that year -- only this Vin Du Pays --- which, interestingly, this release is a mixture of all the grapes listed above. Would it be safe to assume that this VDP would be a cut above the norm, but probably less than the cuvee? Since I'm only starting on my journey to learn and enjoy wines, I have no way of comparing this Rhone to the ones mentioned above; could this be a great value?

The Tour de France begins this weekend and one of the things that I've looked for on the Internet but have not found is a guide to the wines one might encounter during the various stages. While I could not possible race with Lance Armstrong and others, I might be able to drink what they may be served at the end of a given stage. Has anyone come across information on the wines grown along the way of the various stages of this race?

In my item on judging scales, I mentioned a negative experience that I've had with this one shop employee when tasting the Beaujolais Nouveau. Aren't the regular 2003 Beaujolais due out about this time? I'm wondering how they are going to taste -- given the heat and other stresses on the 2003 wines? BTW, is there anyway to project what the regular Beaujolais are going to taste like from the Beaujolais Nouveau? I'd think that the Nouveau's would be like very early barrel tasting, but there may be other factors I'm not considering.

Well, I took my Detroit International Wine Experience 20-point scale and a little notebook with me to an in-store tasting on Monday. I won't name the store but it is fairly well respected in my community but I've not had good luck talking wines with this one person that hosted the tasting. Maybe it is me? (I went to a previous tasting of Beaujolais Nouveau in this store and was asking a number of questions: specifically how she rated this 2003 Beaujolais Nouveau (because of the heat in France and early harvest) and if we could project how the later released Beaujolais would taste based on the Nouveau. I don't know what it was but she responded with some bs.) In any event, the setup and atmosphere was not condusive to what I wanted to accomplish so I just tasted the reds she had open. Didn't find any that particularly appealed to me but I did buy a pinot (recommended in the recent pinot noir item) and left. I've just got to go to this store when she isn't on duty!

Context. Absolutely! Context is everything. A computer could keep track of as many variables as we would want to collect but the limit, I think is on what we can encourage the public to record. The idea is to remove some elements of mystery to the process without over-promising. If used within a broader scope of teaching and sharing, I think that the service can be of value to the individual who wants to try but isn't ready just to jump in and plunk $20 down for an unknown. We all know that until they try and until they develop their own lists of keepers and avoiders, no progress is made but you have to start somewhere. thanks, again, for your feedback!

Thanks, John: it will take lots of work and a good measure of luck. Still got to work out if the scale proposed give enough discrimination. Will explore the concept a bit this evening. One of the better shops is holding a tasting of its selection of staff recommendations. In my scenario, I would prepare a score sheet for each wine in the tasting and capture the customer's ratings at they taste the selections. It would seem that this is a good opportunity for teaching as well as data collection. At the end of the evening, I would collect the sheets and enter them into the database. What I want to test this evening is staff's reaction to the concept of collecting this data during tastings. It seems to me that valuable information is left on the table when there is no attempt to collect reactions.

Yes, Jim: my proposed system "requires" individuals to record their ratings of the wines they have sampled. Of course, that requirement is sorta voluntary, because one cannot expect everyone to participate fully (it is sort of a community thing). Why the ratings? Your ratings are important to the system so that we can establish your profile. The hypothesis is that if you rate 100 wines (arbitrary number for now; some statistical analysis will be performed over time to determine some optimal number and some minimum), if the rating scale is sensitive enough, we can establish a profile of your tastes over that 100 type sample. We can then take a sample of your ratings and find all the others within our database who have scored those same wines the same way that you did. While rather crude, we can say with some confidence that this group shares the same tastes. Now comes the speculative part: we believe that we can predict, with some high level of confidence, that you will like a wine that your peers in this grouping rated high but you have yet to sample or rate. We cannot guarantee that you will like it, but the probability will be high based on your self-reported results. The 20 point scale referenced in the first posting of this group might give us the discrimination that is needed for this sort of analysis. It is weighted towards Taste and Aroma/Bouquet (0-6 possible points) and Overall Impression (0-4) is relatively highly weighted. What I like about it is that it pushes the reviewer, who may not be so sophisticated and experienced, to consider the various aspects of a wine's characteristic. And, again, it is important for this scoring to have value, to be a judgement of the wine tasted and not a reflection of your preferences or some comparision to some other wine. Boris, I like that empty bottle rating system. I would probably be a poor participant is such a test as I hate to leave partially-filled bottles! I guess I belong to the "clean plate and empty glass" generation!

Recommendations. I think that trust has a great deal of impact on how we receive recommendations. For example, as "newbie" to this forum, I spent some time going over the archives. I see lots of examples where people have come asking for suggestions and names are given. (There was even one case I remember where a member passed on a suggestion that another, apparently trusted member made and I got the impression that the recommender had yet tried the suggested wine.) Trust does not have to carry experience with it, but is probably higher if it does. I trust that if I went through all the recommendations given here that I'd probably like most of the wines and that would be considerably better than making uninformed, blind choice in the wine shop. Again, probably nothing replaces the shop owner or employed that we've developed a relationship with. But what if they are not available? For example, someone here mentioned Binney & Smith; I'm not personnally familiar with them, but I understand that they are a big shop in the Chicago area. I went to their website and search for American Pinot Noirs. The initial search came back with 119 entries! While I could slice and dice that list by price or area, how might I choose wines to try from that listing? I'm thinking that it would be good if I had some semi-objective way to see how folks who apparently have a similar palate to mine feel about some of these wines. If there is a match, I think I would feel more comfortable trying something new to me -- recognizing, of course, that in the end, re-purchasing a given wine depends on my own experience with that recommended wine. One of the many challenges my concept has is getting folks to add their ratings to the database. Until there are sufficient ratings within the system, recommendations derived from the data will not be statistically meaningfull. One of the ways that I'm considering is to make the scoring system available at local wine tastings and then recording the results. Once the system is operational and able to provide reasonable recommedations, I think people may be more willing to record their own ratings. What might be other incentives to participating in such an exercise?

I really appreciate all who have responded to my question; while a relative newcomer in terms of being registered and posting, I've been "lurking" in this forum and others as I have began my work on this recommendation concept. On the web, there are sites where you can register your preferences and receive movie recommendations, based on matching your preferences with the preferences of others. There is no guarantee that you will, indeed, like the movies recommended, but the probability is that you will. On the web, there are sites that share ratings and reviews of music; same thing as above: by sharing your profile and match it with others, a play list that you may appreciate can be generated. Why not for wine? Actually, there are a number of recommendation services that might not exactly position themselves as such, but they do. In the States, check out the weekend sales figures for wines discussed in the Wall Street Journal's WeekEnd wine column by Dorothy J. Gaiter and John Brecher. While they go out of their way to say that these are not recommendations, people accept their judgement and taste and go out an purchase and try the wines listed. On this site, there are tasting notes. Some of us will, based on our view of the writer of the notes, try some of those wines ourselves. Our experience is neither improved or degraded by the note -- however, it might encourage us to try something that we might not otherwise. So, what I'm trying to create (adapt is probably a better term) is a means whereby a diverse group of people with differing levels of knowledge and sophistication, can submit a quasi-objective judgement on wines they have experienced. By aggregating this data, I should be able to provide someone who is looking for a recommendation something that has a high probability of meeting their expectations. If so, great! If not, their review goes into the database and helps to further refine it. This has nothing to do with Parker's ratings or the ratings of any of the other sources. As one person has stated, a wine with an 84 rating could be deemed excellent in some circumstances. Does this make more (or less) sense now?

Just a reminder: my interest in this subject is based on a concept I'm developing for collecting ratings and combining them using some advanced statistical techniques in order to arrive at a system that provides recommendations. To achieve my goal, I need to come up with some measure that is useful in differentiating wines and is easy to collect. This Detroit 20-Point scale seems to satisfy those needs; may need some incentives to encourage folks to use it.

The Detroit International Wine Experience has published a 20 point scale for judging wines: DIWE 20-Scale for Judging Wine. In summary, it breaks the judging process down into 5 broad areas: Appearance (0-1) Aroma and Bouquet (0-6) Taste and Texture (0-6) Aftertaste (0-3) Overall Impression (0-4) Among their caveats and instructions: "Remember to judge the wine, and not your preference for wines. Rate each wine individually and on it's own merits. Try not to compare one wine to another during judging. While subjective, would using this sort of judging system allow you to rank wines within certain subgroups (type and price range, for example)? Should the categories be weighed differently?

Jim, Thanks for your feedback! Designed for commercial purposes? Not exactly. I'm looking for ways that folks can share recommendations with other folks. But, facilitating that process has costs and those costs have to be covered. Perhaps through sponsorships; perhaps with an advertising model (similar to the one that offsets the costs of this forum?). I've not worked out all the details on that end. What I'm more interested in at this time is how those recommendations are made. Certainly, the most common way is through the shop owner or staff. If we have established a relationship with them, we can ask them for recommendations. Problem is: they are not alway available and we sometimes wonder if they are just steering us to the more expensive bottle (most don't). We can see from the responses here that many have problems with Parker's system or any of the major magazines (not to mention the fact that many of the wines discussed are not available in many of the markets). So the scenario that I'm trying to address is that overwhealming feeling that some feel when that are told that they should pick up a pinot noir and are faced with 20 different labels at different price points and no clue as to which one to try. A specific recommendation from an objective source would go a long way, I think. In the coming days, I think I will start a new topic to look at different rating and ranking approaches to see which ones might give users meaningful information.

Introduction: a brand-new participant that is very, very interested in this specific topic and wine in general. My specific interest lies in a venture that I'm planning: a way to somehow rank or rate wines with the goal of using those ranking and ratings to generate recommendations. Allow me to say that I fully understand the problems associated with my goal and that ranking and rating wines is extremely subjective -- but the same can be said about other areas where recommendations are offered: books, movies, etc. What I hope to do is to use the computer to do some interesting statistical analysis that can result in a meaningful recommendation. The premise goes something like this: A group of individuals have tasted a variety of different wines and have recorded their ratings (at this point, the rating system doesn't matter but it may when exploring this concept further). The computer can easily take these ratings and break the large group into smaller groups -- composed of individuals who share the same subjective opinions about some subset of the wines. For analysis purposes, they do not have to have exactly the same ratings, but the computer attempts to locate the largest group of individuals who have very similar ratings of some sub-set of wines. (This task, while straight-forward, is computationally complex and requires some sophisticated programming to achieve the goals. It is commonly known as "collaborative filtering" and has been applied to a number of different areas including movies and music.) Let's assume that you are looking for a recommendation for a pinot noir. The system I envision would allow you to "ask" your subgroup for their recommendations. (All of this would be automated and based on the data within the recommendation system.) In other words, the system would determine a group that best fit/match the preferences and ratings you've entered into the system and would look at that group's ratings of pinots and give you an automated recommendation. The premise here is that if you share the same/similar tastes in the wines you've rated in common, there is a high probability that you will be satisfied with the wine that is known to the group but new to you. (The emphasis is on "probability," as there are, as we know, no guarantees.) This introductory post is getting too long and I hope it is not too far from the starting subject about measuring quality. My goal is not to "measure" quality but to come up with ways to quantify subjective experiences and share them in meaningful ways with others. Your thoughts?