Idea for a "Recommendations" resource

[thecrowdog]

I’ve been thinking about all the advice that has been doled out bit by bit whenever someone asks for the “best” tobacco blend of any particular style. Despite the large number of blends available, and the many members on the board, it seems like there are usually 10 or 20 blends that are recommended more than others.

I’ve been wondering if it would be interesting and helpful to create some sort of resource that compiles all those piece-meal recommendations into one place… like a simple chart of the top 10 “most recommended” blends for each particular style. It could also serve as sort of “if you like ‘Blend X’ then you will probably also like ‘Blend Y’.” I could keep it updated as more people find the thread and add to it, and if it is clear that there are more than 10 strong contenders I can expand the list to 20. The point isn't to list the greatest blend that has ever been created (and is probably no longer available)... the idea is to list the "most often recommended" by our members who are very familiar with a particular style. I’ve looked around and can’t find anything like that. If you know of one, let me know and it would save me a lot of work.

If that sounds like a good project, I’d be happy to tackle it, with a little help from the board. Firstly, I would need some help listing the main categories of non-aromatic tobaccos. Something like this… (please make suggestions and correct my mistakes)

Virginias
VaPers
Burley blends (burley as the main flavor profile)
Cavendish blends (cavendish as the main flavor profile)
English (would we need subcategories for highlighting latakia, oriental, perique?)
Balkan (I’m not 100% clear on what makes a Balkan blend a Balkan blend)
Anything else?

I thought about the fact that there can be some crossover between some of those categories, but if the idea is to name your top blends that really exemplify a style, then the crossovers will probably not be the most recommended anyway. If there are gray areas that deserve to be their own categories, let me know and discuss it in the thread, we’ll get a consensus.

Once we have the categories figured out, I can start new threads to get your top recommendations (or I could cull that from past posts if I’m beating a dead horse).

I also thought it would be helpful for your recommendations to be blends that are readily available instead of something that hasn’t been made for 10 years. And I wondered if we should have separate fields for tinned and bulk recommendations (for those trying to cellar-up). Any other suggestions, I’m all ears.

If you don’t know of a link where this has already been done, I think it’d be a pretty good resource and I’d be happy to tackle it.

 

[LL]

Virtually everyone who has taken up a pipe in the computer age has had dreams of a tobacco "recommendations encyclopedia."

That there aren't any should tell you all you need to know about the implementation of such a thing. :twisted:

 

[thecrowdog]

well that sounds ominous. what happened to no such thing as bad weather?

I don't feel like it'd be trying to do the impossible. It would simply be agreeing on 6 or 8 categories, and then simply count the recommendations that come in for each category. Each time a blend gets recommended, I give it a +1 in the chart, and the most recommended blends rise to the top.

Like I said, I'm not trying to name the "best" tobacco ever and get everyone to agree on it... I'd just be adding up the blends that do get recommended for each category. Technically I guess I could do it from past posts without any new input, but that's not what I imagined being part of a bulletin board community was like.

 

[LL]

well that sounds ominous. what happened to no such thing as bad weather?

Take it from a 20 year IT guy... you don't want to go there.

As for bad weather, that's just practical advice here on the Montana/North Dakota border, not a newage-y feelgood mantra. I'm too old for that shit. :lol:

 

[thecrowdog]

ok, so I've got you down for "not interested." that's cool. but as far as IT, I don't imagine needing anything more than reading the posts and adding +1 on my fingers. Let's not let technical concerns derail what you would agree everyone has dreamed of.

 

[LL]

ok, so I've got you down for "not interested." that's cool. but as far as IT, I don't imagine needing anything more than reading the posts and adding +1 on my fingers. Let's not let technical concerns derail what you say everyone has dreamed of.

You consider BoB to be a statistically valid sample size for a simple tally to be meaningful?

On the other hand, the only one that is---the national tonnage sold---says Lane OTC's like 1-Q win by 50-100X over everything else made.

That sort of problem can only be addressed with weighting algorithms. Followed by the experience level of the recommender, and on and on.

Enjoy. :lol:

 

[thecrowdog]

I'm sorry, I think you don't understand what I'm interested in doing, which is fine, but I also don't want to have some running disagreement with you on why you think I shouldn't bother with this. I think that BoB is a completely statistically valid sample size when what I am interested in is the opinion of BoB members. I think Budweiser sells the most beer, but I'm certain it will not be the most recommended beer on a board of serious beer tasters. Every time someone posts something like "what's your favorite VaPer?" do you get on there and tell them not to ask? All I wanted to do was compile the answers and see what is recommended the most by BoB members. As simple as that. No fancy IT problems. No tonnage stats. No attempt to be scientific. I didn't expect anyone to actively oppose it.

 

[LL]

I'm sorry, I think you don't understand what I'm interested in doing, which is fine, but I also don't want to have some running disagreement with you on why you think I shouldn't bother with this. I think that BoB is a completely statistically valid sample size when what I am interested in is the opinion of BoB members. I think Budweiser sells the most beer, but I'm certain it will not be the most recommended beer on a board of serious beer tasters. Every time someone posts something like "what's your favorite VaPer?" do you get on there and tell them not to ask? All I wanted to do was compile the answers and see what is recommended the most by BoB members. As simple as that. No fancy IT problems. No tonnage stats. No attempt to be scientific. I didn't expect anyone to actively oppose it.

I'm not opposing it. Just trying to explain why the first thing I said---that everyone wants a recommendation engine when they start smoking a pipe, but such a thing never gets made---is the bottom line.

Knock yourself out.

 

[DoverPipes]

Not to over simplify things, but wouldn't be easier to go to tobaccoreviews.com and see the ratings there? :scratch:

 

[LL]

Not to over simplify things, but wouldn't be easier to go to tobaccoreviews.com and see the ratings there? :scratch:

Garbage in, garbage out.

This comment from another (non BoB) forum touches on some of the reasons tr is largely considered useless:

I have been reading tobacco reviews nearly three years now. It's taken me that long to realize it's really a mediocre site at best. There are so many things wrong with it, I hardly know where to begin. The owner of the site should really seek out pipe smokers and ask them for input that could help create relevant information for that page.

If I had to pick the one issue with a site, it would be the reviewer who gives a one star vote because they didn't like the tobacco. Seriously, how relevant is that? Of course, the reviewer is left no choice because of the recommend options. They think 'Well, I didn't like it, so I can't recommend it,' that's only natural. How about an option to indicate how much you like the blend, and an option to indicate if you can recommend it? For example, I don't like Key Largo because of the cigar leaf, but I recommended it to a friend who likes cigar leaf and he loves it. Makes sense, no?

Overall, it's just a poorly conceived site and I caution anyone using it to consider these things. And it's not the users fault; they are only working with what they are given.

The Big Picture problem is the same one that wine, cigars, beer, etc. recommendation & rating systems have: BOTH ends of the transaction are 100% subjective. The taster/recommender and the looker-upper are both people, and sensory input does not quantify well. Yet crunching numbers is all a computer program can do. (Well, there might be some Japanese scientists working on a taste AI of some sort, but that hardly counts.)

 

[shootist51]

Have you ever taken a look at www.tobaccoreviews.com? It may be as close an example of what you are looking for, though it is not limited to BoB members.

I've seen alot of pros and cons on the sites effectiveness, but I like it as just a general reference.

The thing you run into is that our activity is sooo subjective. One man's trash is another man's treasure. This tends to produce some pretty dramatic differences of opinion concerning the different blends.

I always read the reviews posted on BoB, but many times when trying a recommended blend find myself wondering why I don't have the same experience with blend in question.

I would never try to dissuade you from trying this as a project, but at the same time I can see it in the same light as bailing the ocean with a teaspoon. It could be a collasal effort with little return.

Having said that, one of the things I enjoy as part of the smoking experience is keeping a log or diary, if you will, noting MY experiences with tobaccos I try. This probably would have little or no value to anyone else, but for me it helps keep straight what I like and more importantly, blends that just didn't make it for me. It also is a help for me in tracking my own tastes as they mature and become more sophisticated.

I am often suprised at the entries I make on a fresh tin, in comparison to the same tin revisited under different circumstances or as opposed to the same blend with a bit of age.

Any way, good luck and good smoking :lol!:

 

[Aaron]

There was a "touchstones" webpage a while back that was sort of what you are talking about crow... I don't know what happened to it, I can't seem to find it anywhere.

 

[thecrowdog]

thank you for the suggestion and I have gone to tobacco reviews dot com for quite a while. before I was aware of this board, it was my main source. but as most people have noted, one man's trash is another's treasure, and I had to wade through reviews of something I already liked hoping SOMEONE would say "if you like this, try this...." but that was only one person's opinion... if I was really lucky, more than one person would have the same recommendation.

And I can't understand all the people on TR that will say "I hate all English blends, so I tried this English blend, and I give it 1 star. I'd give it zero if I could." Well, I want to see the recommendations of people who LOVE a particular style, not those who hate it.

The other source of frustration with TR is that you can't look something up by "style" or main component... you need to know a brand or blend name (or just start surfing the highest rated). What I ended up doing was finding the products with the most ratings (quantity of votes) that also had the highest ratings (subjective quality)... a poor man's way to find something that at least a large number of people on that particular site agree is good for what "it" is. But that is not an ideal way to go about it.

When I joined this board I realized people are doing this all the time. Every time a person says "I like Virgina Flakes, which ones should I try?" There might be 50 answers but 10 of them will be the same. And someone else will ask the same question a year later, and there might be 40 more answers, but the top 5 are the same as the top 5 from a year ago. A pattern emerges, even if it's not scientific. I thought if I just put all these recommendations in one place it would help people out.

I think maybe if I had just gone ahead and made a start of the project, it might have helped people envision it. It's nothing complicated. I can create it from previous posts, but I thought I'd make it a communal effort from the start.

 

[Harlock999]

I like the idea, if it's do-able, however, I doubt I could contribute anything more than generic, stock reviews along the lines of "I like it", or "I didn't really like it". I just can't get the taste buds to sense much beyond the basics.
It will take diligence on the part of the compiler to stay on top of all the reviews and recommendations.

 

[Justpipes]

Crowdog, you have a good idea and there is nothing stopping you from carrying it out. There will always be some folks that never see the glass half full and will come up with 57 reasons why something won't work before they ever consider the positive reasons why something is a good endeavor. LL happens to be one of those folks.

 

[thecrowdog]

Yeah, I think what I'll do is compile what I can from past posts and then see if that sparks interest. If not, no biggie.

 

[Harlock999]

I'll definitely be checking it out!

 

[Bub]

Check this out
http://www.smokingpipes.com/tobacco/tinned/bestsellers.cfm

 

[ZeroContent]

You'd have to be a little more selective of reviews and lay out a template of points to touch on in every review. You'd also what to pick people who can be constructive in their reviews and leave opinions of "i liked it" or "i didn't like it" out and describe the experience. Also you can avoid the problems that tobaccoreviews have where people you hate English in general throw the useless review out and skew the overall positive experience of people you like English and bring a 4-star rating down to a 3. Maybe you could create a flavor matrix that you could search on of different type of flavors that smokers perceived and weight those on how many people have perceived the same flavor. It could get complicated to come up with a standardized system.

 

[LL]

Crowdog, you have a good idea and there is nothing stopping you from carrying it out. There will always be some folks that never see the glass half full and will come up with 57 reasons why something won't work before they ever consider the positive reasons why something is a good endeavor. LL happens to be one of those folks.

You forgot the based on experience part, JP, which in this instance was the sole reason for my comments. There is plenty stopping people from carrying out such a project, I'm afraid. Positivity (or a lack of it) is irrelevant.

Computers don't respond to HappyThink, only people do, and the implementation of a recommendation database requires dealing with computers by definition.