Thoughts on the latest updates? Top 250, etc.

HF was added in 2010. Trillium in 2013. 2012 or 2013 is around the time they started allowing quick ratings to determine the overall rating. That was right around the time Trillium came around. HF had about 3 years when they were generating reviews before things changed.

 

If you really understand statistics then you will understand that no one is upset that their one vote out of hundreds or thousands is not counted. I don't think anyone here is complaining about that. The point is every vote should count, resulting in more data and a more valuable statistical result.

I am not complaining about my vote not counting, I am complaining about the useless waste of the majority of the ratings data simply because an arrogant belief that those who take the time to write a review are are superior than everyone else.

Why does this noise upset you so much? When there is enough statistics then the noise it just gets washed out in the background and becomes irrelevant. But you need more statistical data to accomplish this, not less.

It was already hard enough to defend BA to my friends when they are using RB or Untapped, but now when I go to pull up a new beer on BA that has only been out for a few months, from a local nano brewery, I will have no useful ratings average or score from my BA page because it might have had only a few dozen ratings and 1-2 reviews, while they will have real valuable data on their phones from RB and Untapped. I suppose, as they are laughing at me and my choice of BA, I should just explain to them that after a few months or years, the score will slowly fill back in because the site is now forcing people to write 150+ characters to get their voice heard, we just have to be patient and wait. That doesn't help me in that moment, does it?

 

No I have not had any of their beers. One is on the way in the mail, and I am looking forward to reviewing it. That said, since we are calling people out - how exactly should I interpret your opinion of a beer? Particularly given that this is your last review:


Not sure I want your considered opinions included in the ratings to be honest.

 

Even if that review doesn't have much, when it is one of a hundred, collectively they are useful. That's what this change lost, the trend of a lot of small pieces of data.

 

That review was required because his rating was considered an outlier. Forcing him to do a review doesn't make his rating any less valuable.

 

Yeah... And I basically told everyone to ignore my rating. So now you are saying I can't rate a beer unless it is a style I truly enjoy? How about we have a 50 page questionnaire when you join the site to make sure you are only reviewing beer styles that you like? I started out drinking sours mainly, then moved into IPAs and now I am trying to get into porters/stouts. Some of them just suck to me still... It is what it is. Would you rather I just lie about it and say this is the best beer I have ever had because BA told me it should be? It is my rating and how I felt about the beer.

But back to you... I hope you really enjoy the Trillium man. You will see why some of us are upset soon.

 

Disagree - I have (for fun and geeky-ness) spent hours looking into the data yesterday and this morning, and I don't think we 'lost data' we excluded biased ratings. Can you honestly tell me that a rating with all 2.5 across the beer and 'not my thing' as the review is worth anything as a data element?

 

Of course its us biased because make the best beer. Merica fuck yeah!!!!

 

When thousands of ratings are thrown out, obviously some are going to be biased but many more were legit ratings. Also, just because a review is required doesn't mean the review won't be biased. If people want to continue to pump up the value of the beers they are trading, they can still just enter a review and tell everyone it is the greatest beer ever made.

 

But that;s not what the data is. It will have some unexpected 2.5, it will have some unexpected 5.0 and a bunch in the middle. With most beers hovering between 3.8 and 4.3, all that data will help let us know whether it is higher or lower. That's the beauty of stats, they won't all be 2.5 without justification, they will be all across the board. It's much easier and better to include them all rather then hunt out the few you don't like and exclude them and much better then removing >90% of the data because people didn't justify their rating. When you are going for mass data, the justification is not necessary. The justification comes from when you want personal experience. Two very different things that are both very important.

 

I understand that the review is required because of the outlier. I believe the intent is to get people to think about their rating. First and foremost one would hope you are rating to style, not personal taste, and that you have some considered thoughts around why it 'isn't your thing'.

I don't like sours - but I still rate them to style on the rare occasion I have them, and don't take a dump on them (or hype other beers) based on my personal visceral 2 second reaction.

The intent of the change is clearly to have ratings based on more considered and thoughtful reviews and not ticks. I personally like that, and you guys clearly don't, but it doesn't change the fact that a quick 2.5 and a 'not my thing' isn't adding any useful info for other users

 

http://www.conceptstew.co.uk/PAGES/nsamplesize.html

http://www.select-statistics.co.uk/article/blog-post/the-importance-and-effect-of-sample-size

 

You guys keep vaguely quoting 'stats' and 'data' - well the data is up there, and you can pull it down just like I have. Feel free to demonstrate that the data isn't biased under the old system. So far I am not having much trouble showing that it was, but presumably you can disprove my hypothesis pretty darn quickly if I am so wrong in my thinking.

 

I get into this argument with a friend of mine all of the time. He rates beer as a representation of the style. I rate beers on how they taste to me. Maybe people as a whole dont like barleywines. But you are saying the best one you have ever had should be a 5? That is also wrong to me. It should be what it is to you. It is your review.

 

You have repeatedly expressed an opinion and quite well expressed. What you haven't done though is backup that option with facts/data.

I personally am not losing any data because I don't have unqualified belief in the value of numbers alone, and having been through the change in which ratings only were allowed I learned their negative impact on the quality of the information provided by this site as more and more people abused their intent.

 

Then you are using the wrong website, as that isn't the expectation of the community here. No-one is asking you to rate beers to show standards, but you are expected to keep style in mind:
http://www.beeradvocate.com/community/threads/how-to-review-a-beer.241156/

 

Now this is actually one of the concerns I can relate to. I wonder if there is a way to make those individual brewery pages more useful while still separating out reviews from ratings.

If this is really what is hurting Trillium, then this should balance out with some time invested into this new system, as people start to review more to get their scores counted. Unless @pagriley is correct and they are being dinged because a higher-than-average number of the scores are coming from people ticking their beers off at festivals, bottles shares, tastings, etc.

 

I prefer more information from a broader spectrum of beer drinkers. There is nothing inherently preferable about a reviewed ranking vs a number ranking. In both instances I could check an individual score and compare it to others scores by the same person (of beers I've tried) to see if the poster is giving me useful information. After all, I'm looking for beer that I enjoy drinking. Beer that people think tastes good. The tickers do a fine job helping me with that information. There are many variables and biases at work regardless of the system. I'm not sure lowering/removing the scores for many of the beers people are really excited to try does anything positive for the vast majority of BAs. Why someone reviews a beer at 4.5 (or 2.5) means very little to me by itself, much like the simple rating. On the other hand, a simple rating (or review) backed up with a list of other ratings (from the same user) gets much closer to telling me something valuable. I prefer more data so I can tailor it to myself.

 

When did you prove that all the hads were biased? I must have missed that. You looked at a ratio of hads to reviews and missed many other variables in your data. You included a couple data points and claimed your opinions as facts.