In Reply To #1 Sorry if it's confusing. When you rate a beer, you're supposed to judge the first 4 (Aroma, Appearance, Mouthfeel and Flavor) in comparison to other beers in the same style. This is how I intended it to work, but I guess not everyone does it this way. The last one, Overall, is the one where you say how you actually liked the beer, regardless of style. So, if it was the best IPA in the world, yet you can't stand hoppy beer, you could give it 10, 10, 10, 10, 2. If you look at each person's review you'll see the "Final rating, by style: #". This seems to be the #1 in your breakdown, which you explained correctly. It's the average of all 5 attributes. Then you call out #2 as "Rating by Style". That's the exact same thing as #1, depending on where on the site it's listed. If you're looking at the "Latest Reviews" page, or a user's member page, it gives the #1 result by that user, or the average off all the #1 results if you're looking at the beer's Rating by Style, not just one member's. Overall rating is only based on the last of the 5 attributes, "Overall". Depending on whether you're looking at the beer's overall rating, or a single member's overall rating, it's either the average of all "Overall" numbers for that beer, or the single number of that member's review. It all makes sense in my head (but isn't that how it always goes?), but I suppose it's pretty confusing and not very clear. The purpose of having two different numbers was to make it more fair. Someone may hate wheat beers, which you could see by the overall rating, but it wouldn't detract from the style rating of which they may honestly give it a high rating. In any case, this will all be irrelevant soon. People who have checked out the new site have probably noticed that all (but one, that I just noticed) of the pages has trashed the old system. There is only one rating listed, which is the style rating. I got rid of it for two reasons, with the first being the cause of your inquiry, confusion. The second reason was to cut down on server usage. The databases can lose a couple fields and page processing time will be quicker without having to show two ratings all over.

In Reply To #2 Flash, thanks for the reply. I admit I'm still confused even after your explanation. The article at Beer Info ) Judging ) Basic Guide to Judging Beer doesn't agree with the intended use of the Overall score. It describes the Overall score more like a Drinkability score, unrelated to style. I like that. And the new site looks like it has a tool to show the style averages under the rater's score for the particular beer under evaluation. That's good, too. This means that the style definitions will need to be pretty clear for users, and expanded from current menu options. My suggestion is pretty simple. Just normalize the value of a rater's score by considering established preferences. To illustrate: Since I like dark ales, my average score for dark ales is higher than my average score for light lagers. Normalization is a way to take this bias into consideration. To me, it makes more sense than the current method. It's not realistic to expect raters like me who put up a bunch of 2s for a light lager to then award a 9 for drinkability because it was a damn fine example of a style I do not care for. Likewise, it’s not realistic to expect raters to give out high marks for technical merit (Aroma, Appearance, Mouthfeel and Flavor) and then low marks for Overall. If you love the Aroma, Appearance, Mouthfeel and Flavor, aren’t you going to want to drink that beer, regardless of style? If a person’s rating profile demonstrates a bias, as mine does, then the ratings cannot be taken at face value. Given a preference for dark ales, and a mean light lager score of, say 3, my 3.5 for a light lager might be even more impressive than a light lager lover’s rating of 4.0 for the same beer. Anyone interested in Bayesian estimating techniques will recognize what I’m driving at. If you are interested in considering this, I'll email you some observations and suggestions, rather than posting them here. Bottom line, it would not require any rater to change his/her existing ratings--the ratings that are already in the vault would require no modification. Some folks might want to edit ratings if they rated the way you described in this thread, but I think that’s a small number. So this could be implemented without upsetting the apple cart. However, I predict you’d see some changes in the top 25 rankings if we consider individual rater's profiles in how the rankings are calculated. Your thoughts?

In Reply To #3 Just a quick reply to this... My overall rating sometimes is nowhere close to the other 4 attributes, since the overall should be the only biased observation. I hate altbiers. I could give one 8's and 9's and then a 3 for overall. I don't think it even matters how you rate them because no matter how specific BP is, I can guarantee you that half the people will either not figure it out or ignore the instructions. I have intended on using the Bayesian method eventually. There will be other kinds of datashaping as well. Being an engineer, I have a couple books on statistics from college that I plan on dusting off. But this will have to wait until the skeleton of the site is finished. I was also going to put stuff on the site explaining that all these ratings are purely for fun.

Converting to the new layout is taking so much time it's not even funny. It's really upsetting me because I'd rather be adjusting the brains of the site, like stats. And I'm sure that's what a lot of you guys want to see improved and have been making many suggestions about, yet I continue to have this attitude of "i'll do it later" or "don't worry about that now". I've spent over 200 hours on this damn new layout and sometimes I want to give up. I have to keep reminding myself... it will be over! At least I've reached the half way point. I've been working on the beer pages in the past week on another site so that I don't screw up the database. The reviewing page has a few new chanegs.

In Reply To #5 Be encouraged, Flash. It's great work you're doing. My suggestions are not helping at this point, so I will take a hint and keep them to myself. I don't want the good idea fairy to create so many distractors that the real job gets set aside. I can only imagine how much time and effort you are pouring into this awesome site. Purely a labor of love, and I for one am grateful. I see where Clash is an admin now, so hopefully you are getting some additional competent help. If I can help, let me know. And if sitting on my hands is the best help I can offer, I'll do that, too! Less comments from the gallery! I see where Pub Crawlers sells team T-Shirts that help raise money for their site, and of course BA flat out asks for donations at theirs. I'm saying, I would buy and wear a BeerPal T-Shirt, if that would somehow help. Shots ring out! And the sound you hear is the Good Idea Fairy, twitching in the dust.

In Reply To #6 Thanks for the encouragement, Foamdome. Clash has always been an admin, but recently changed his name. A while ago we considered making BP T-Shirts. It even became a poll. It's still a good idea, but the challenge is creating a good design or logo for the shirts! On top of that, I'll have to put up hundreds of dollars to make an order. I did some research and it looked like it would be about $6-10 per shirt to print up, with huge differences depending on colors and quantity. There are places that will print to order but they can go to hell because the site would make next to no profit. I also said I'd give away shirts in contests if profits were made. Donations will be accepted once the new layout is completed. I'd be embarrassed to ask for money with the current state of this site :)

This message was edited by Flashpro at 8/25/2004 10:31:45 AM.