First let me say that I have no first hand knowledge on how the rating system is set up. However, like you, I assume it relies on Bayesian rating, based on how it seems to perform. I actually think this is a good approach for this site, considering the relatively small number of reviews for many of the beers, the inter-rater variability and the fact that there is no ability to standardized or otherwise assure the validity of the rating process. This method attempts to take into account the "believability" of the ratings when there are only a small number of reviews. The greater the believability (or certainty), based on increased numbers of reviews, the more this Bayesian rating will approximate the unweighted rating. The lower the certainty, the more the rating will tend toward the average rating of all the beers, or maybe all beers of that style - not sure how it's configured. Anyone know the specifics?

It is a Bayesian method, which for the purpose of beer ratings is perfect. It prevents contrarians from taking their favourite beers (which may be low rated)and getting together with a group of friends and given them all 5s to increase the rating, or taking a highly rated beer you hate and having a group of friends all give it 1 to lower the value. In short, it keeps people honest.

One of the most important reasons for using Bayesian weighting is simply to prevent constant turnover in the site's Top 5 list. Each new beer starts out with the average score for its style, and then gradually moves up or down as ratings come in. There is a minimum number of ratings required before a beer's rating counts in the overall rankings. If not for weighting the ratings in this way, it would be possible for a person to add a new beer to the vault, give it a perfect score, and see that new beer go to the top of the Top 5 list--with one vote. I am not suggesting anyone would be so bored with life as to do this for entertainment, I'm just saying, it could happen and each time it did, it would be an unwelcome consequence of a poor rating system. More than merely nice in theory, the Bayesian Method is very well suited to this site and our application. We estimate the beer's true ranking based on what we already know about the style and then all the subsequent information that comes in in the form of rater reviews. The errors cancel out and the collective membership finds the true ranking. It would be interesting to "adjust" ratings for individual preferences. Do hopheads over-rate really hoppy beers? If so, should their ratings be adjusted to correct for the bias? Search through the archives and you'll see where this topic was bantered about before.

Funny you asked this, I was thinking the same thing. I recently picked up the Sierra Nevada Southern Hemisphere Harvest and was checking to see if anyone else had tried it. http://beerpal.com/Sierra-Nevada-Southern-Hemisphere-Harvest-Ale-Beer/30937/ It was pretty evident after one 4.0 review from stoutlover72 and the beer was 3.2. After 2 reviews, its at 3.3. I guess I understand the reasoning, it just made me wonder on 1st glance.

And now it's available again, limited supply. Scored a bottle. Wish list forever. Sorry to see ASSsssspitititties sorry ass in this thread though.

Ratebeer and Beeradvocate both have the 2016 version listed as a separate beer than the original. I'm not privy to the details of why, except that since Alexander was last brewed, Palm bought Rodenbach. This might be like a movie based on a true story. I hope not, though, as I'll have one set aside for B10k in a few months.

geez, had forgotten about this thread.. gotten to try Alexander twice since then. amazing stuff. on the hunt for the new one...

Got a bottle socked away for a tasting at the end of next month. Looking forward to it!