Brain Storming: A shaving product recomendation database

[TourettesTuna]

Ok. So, I'm a little new to this forum business (this is my first time participating in a forum), and I've only been wet shaving for about eight months. That said I had an idea during my explorations of various forums and blogs, and I wanted to see if anyone here has the interest and/or tools to do it. The idea is to build a system for rating shaving products and receiving recommendations based on what worked well for you. It would be similar to the anticipated rating system that Netflix uses to recommend videos to you (or any other trargeted marketing program).

I feel this would be helpful to new and experienced shavers alike, because my experience in combing the forums has been that I tend to latch on to certain experienced members over other experienced members because what works for them also works for me. This method of research and experimentation took a great deal of time sifting through the likes and dislikes of particular experts, and left me feeling a bit ill informed until I tried something. Thus the natural desire is to have a program that can do that sifting for me and give me quantitative answers as to the personal efficacy of a product. An added bonus would be recommendations for which shaving experts to follow and learn from.

I guess what I'm really trying to eliminate is some of this "Your Mileage May Vary" (YMMV) business, and get to the point where we can start giving product and procedure recommendations with a little more confidence.

Though I feel I should be clarify that I don't want to replace the wonderful places that the wet shaving forums have become. I merely want to take out some of the guess work in finding good products, and free up some time for talking about the finer points of shaving with you wonderful people.

 

[mblakele]

The idea is intriguing. Do you know much about the technology behind http://en.wikipedia.org/wiki/Recommender_system implementations? It sounds pretty challenging to me - and expensive. Even with a large investment, success is not guaranteed. Netflix has one that seems to be pretty good, while I have given up on the one at Amazon. Solving the problem well would seem to require a specialized engine, incoming streams of training data, and well-designed feedback loops.

It would be easier if shaving products were as cheap and commitment-free as watching a movie or listening to a song. Blades might be the easiest area to tackle: gather feedback from sample packs.

I wonder how well a system could be made to work for new wet-shavers, though. Think about how many new members start out with Proraso cream. If they stick with that one product, they might not need another for 2-3 months. But how good will the recommendations be for "you liked proraso, so you might like..."? A poor system would probably recommend Bigelow: the same stuff with a different name. It might be much easier to discover shaving creams liked by people who did not like Proraso.

This might be part of a larger problem: getting stuck in a local maximum. For example when I began reading B&B there was a lot of talk about the Jagger/Mühle R89 razors, and vintage TTO razors, but little about vintage open combs. It turns out my favorite design is the vintage NEW, but I do not recall finding it through recommendations. I think I happened on one in an ebay lot, and tried it out of curiosity. At first I could not find the right angle, and set it aside. Months later I tried it again for Open Comb March, and never looked back. A perfect recommendation system would have guided me there, but I cannot imagine how. It would be like Pandora noticing that I only listened to classic rock, and then recommending Bob Dorough. Pandora actually does things like that, and gets away with it because songs are short and have a very low marginal costs. Most folks feel a little more commitment to their shaving products, I think.

Working something like this as a for-profit venture is fun to think about, even though I suspect the market would be much too small. But it might work out as something like a "sample pack of the month" club. Say that for $X monthly you get a sample pack of whatever you want to explore next: blades, creams, soaps, or AS. You supply feedback through a web site, which goes into the recommendation engine. Your next pack reflects your preferences so far: products you are very likely to enjoy, plus one or two riskier selections. Extending this to brushes and razors would be more difficult, because the raw materials are more expensive and there are real or perceived hygiene issues to sending the same equipment from person to person. You could also have a starter kit option for new members, so that getting started wet-shaving would be as simple as checking that box when you sign up for the monthly package.

 

[dpm802]

You can never eliminate the YMMV factor. There are too many differences between mens' beards, the water one shaves with, preferences of those around you (who have to smell what you chose to wear that day) and even seasonal differences. What works well for me in Winter doesn't fly in the Sprintime. And even if I were to duplicate the identical set-up from one shave to the next, I'm might not get the same results every time.

Besides, experimentation and guess-work are half the fun. At the beginning, I was buying anything and everything shave-related, but after a few months, I found that I prefer DEs over SEs and Injectors, I like a Straight bar over a Slant over an OC, I don't like adjustables, etc. Now, my buying is much more selective, since I've already figured out what classes of products I actually use, rather than just collect. You can't get this kind of knowledge by reading other people's opinions, you have to try them out for yourself.

That's not to say you don't have a good idea with this proposed system. I'm just wondering how hard it would be to actually implement. It would probably take several years to fill up the database enough for it to be useful. In the meantime, I will continue to read the Reviews forum if I'm contemplating a new purchase. It already has a rating sytem of its own, but I've found that many products get undeservedly high marks, simply because people are more likely to review products that they really love, and ignore the ones that are so-so.

 

[Chadao]

Nice idea, but there is so much variability in human nature. Here, just look at the pro-Arko and anti-Arko contingent. Or the Veg? I love MWF--it is my favorite--but others couldn't care less--disdainful, even. But, who knows, perhaps it IS quantifiable and you can come up with that system. But for me, it would take a lot of the fun out of all this. Plus, I might try one soap, set it aside and then come back to it a year later and decide I love it. Same thing with blades...or brushes...or after shave...Anyway, good luck, but YMMV

 

[Go West Young Man]

Netflix only works because it has millions of constantly updated reviews; we just don't have enough data points here for that kind of system to work.

 

[BrookR1]

The Reviews section works OK, but it requires people to be more active to keep it current.

 

[TourettesTuna]

Do you know much about the technology behind http://en.wikipedia.org/wiki/Recommender_system implementations? It sounds pretty challenging to me - and expensive. Even with a large investment, success is not guaranteed. Netflix has one that seems to be pretty good, while I have given up on the one at Amazon. Solving the problem well would seem to require a specialized engine, incoming streams of training data, and well-designed feedback loops. -mblakele

I'll be honest, I don't know much about implementing big projects like this. I'll make a point of reading up on the link you posted. Really ,the idea here is to see if there is sufficient interest to warrant the project, brain storm about possible implementations/problems, and try to find people with the expertise to start it. I was also initially thinking of this as an open source project, but that may just be naive on my part.

I wonder how well a system could be made to work for new wet-shavers, though. -mblakele

I'm a little worried about that as well. It would be easy enough if they had tried a few things, maybe at least three soaps or some such, but a complete newbie presents a problem that could likely be solve by having sample packs available for consumables at least. As, you said things like razors present a serious problem.

This might be part of a larger problem: getting stuck in a local maximum. -mblakele

I couldn't agree more. I've spent some quality time with lens design optimization software, and that is a considerable problem for any attempt at multivariate optimization. However, I don't consider it as big of a problem since, ideally, people with the largest amount of experience would still be largely looking to their own curiosity to continue improving. After all, competitive research isn't something you can do by just reading a textbook on the topic. In this way the innovations of people like yourself could trickle down to those of us who are less intrepid. Also such excursions into the unknown are the sort of thing that I think these forums are perfect for.

Say that for $X monthly you get a sample pack of whatever you want to explore next: blades, creams, soaps, or AS. You supply feedback through a web site, which goes into the recommendation engine. -mblakele

That sounds like an awesome way to feed the system.

Thanks for the great input.

 

[TourettesTuna]

Besides, experimentation and guess-work are half the fun. -dpm802

I have to agree with you about the experimentation. That really is what makes this fun, but getting consistently good results is also fun. Consistency is more what I'm shooting for here. Think of it as a tool to improve the rate of convergence to a set of wonderful shaves.

 

[Mick]

what someone thinks is the "cat's pajamas" might be "litter box fodder" to others.