Thoughts on the latest updates? Top 250, etc.

Well, there is also something called a ban hammer... I imagine once that drops a few times people will stop pasting in crap reviews.

 

Stop by Hicksville , drink some whales and pass some of that brainpower over to a dumb mechanical engineer.

 

Forcing people to write reviews is going to create a lot of bad reviews. At least now I know there a few reviewers I really trust and know what they are talking about. Now I am going to have to sort through pages of BS to find the good reviewers.

 

True if their is only one or a couple scores. But normally that one review is sually joined with 40 or 80 scores that should be a fair representation of how good that beer is. You won't know why, but when you are starting with no info to begin with, those 80 or so averaged scores come in really handy.

 

I guess my point is who determines a crap review versus a crap reviewer? And does anyone really have the time to police/monitor/babysit reviewers all day long?

 

They don't need all day. They get alerted. They take a quick look. They do a quick evaluation of the text. They take action or not. No witch hunt, just simple recognition of someone gaming the system. Just like reporting a post.

 

This might very well happen, and you will also have quality quick ratings from people who are thoughtful but just don't have the time to write a full review. Maybe they are at a bottle share or beer festival tasting dozens of beers in a single day. And you also have the casual person quickly rating a beer 1 or 5 or somewhere random in between, who cares about them, statistically they will be insignificant in the long run. I still do not see the harm in inflating all reviews by 0.13, or whatever the number might be. All this does in raise the floor for everyone.

 

On the original site where all you could do was review and not tick, I don't ever remember there being a major problem like this.

 

Except that for years, you could only write a review on this site. There was also a minimum character requirement, which the only other rating site did not have. I much preferred all the reviews here over that. With "hads" you are sorting through BS to begin with.

 

I expect reviews to be similar to how you leave feedback on a trade. It's always some generic ebay review like , great beer, fast shipping, great trader or something (that I'm guilty of as well). Reviews will be similar, especially if you have to go back the next day to do the review because you are busy enjoying the company you are with and not writing the review while ignoring everyone around you.

 

It wasn't until that change did I start participating. I now use BA to keep track of every beer I had so I know whether I like it or not to get it again. That has led me to branch out into the forums and other things. Before the change, I used a giant spreadsheet because that was more convenient to me then this site. I fear their will be a mass exodus to untappd or somewhere else because the site is making it harder for the user to stay an active user.

 

I don't know what those 40 to 80 scores mean at all if there is no review, it is incredibly vague and general. 150 characters is not a lot, and people certainly don't have to fill a book. On top of which, not everything you drink needs to be scored in some way.

As far as I know they have not disallowed hads, so if all you want is a record of what you like or not, that function is still there.

 

I wasn't here back then, unfortunately, but I understand a ton of frustration from people I greatly respect. Because someone chooses or doesn't choose to type out a review doesn't make the score they give a beer any less or more powerful than someone that does. There is a reason BA changed from what they were doing. Also, things have changed dramatically since the original site.

 

As I said, I never really liked thermodynamics and statistical mechanics. I would rather drink a good new beer that I found rated on BA than crunch numbers like that. I just know that BA scores would be better with a higher number of ratings than a lower number of "reviews". As you should already know, the standard deviation goes as 1/(n-1) which is effectively 1/n (physicists like back of the hand, order of magnitude calculations), so the higher n (sample size) is, the lower the standard deviation is, and therefore the lower the statistical error is, which means a more accurate measurement of the particle mass, oops, I mean ratings average.

 

I am strongly against the new rating system. I thought there was nothing wrong with the way it was. So there will be a few extra 5.0's with the old system, so what? The old system was pretty dang accurate to the quality of the beer. Now, you will have a rating based on a small handful of reviewers that "took the time" to write a review, rather than dozens/hundreds of reviewers, that is statistically the wrong thing to do.

I base a lot of my trades on BA average scores to understand the quality of the beer I'm ISO. If this is the way it will stay, I'll be changing over to, *cringe*, ratebeer.

 

One thing about this change that sucks is it is going to difficult to get feedback about new beers. For example, I was at the store yesterday contemplating the purchase of a new IPA from a local brewery that had been introduced a month or two ago. This was before the change and there were 10 ratings and the score was enough to give me a decent idea of what people thought of it. However, now there is one review/rating of it, so its current score is pretty much worthless.

I wonder if a happy median would have been to give reviews more weight than ratings, but still let ratings count for something.

 

How about have only reviews show by default, and toggle to show reviews or reviews & had ratings?

 

Technically yes its more accurate. But the fundamental question is whether that additional accuracy is worth worrying about at all. For example, Nielson has been able to predict, within reasonable margin of error, the TV viewing habits of the entire US based on a sample size of 1000 when dealing with a population of approximately 320 million. (Reasonable in that lots of folks pay them for their numbers and their results... .)

Also, it is in fact possible to have too much data in the sense that I've seen a data set using 50,000 observations where a correlatiion between two variables of 0.0002 was statistically significant with a probability of less than 0.00001, which provides an effectively meaningless relationship between the two variables. Hence the argument that bigger n is not necessarily better. Further more if the additional n brings with it unknown/unknowable biases the numbers alone do not help us and can in fact be very misleading, especially in the absence of qualitative information. (In recent years I've seen relatively new beers with no reviews and just numbers without any way for me to tell why those numbers were given.)

So my bottom line is that combining numbers with qualitative information from reviews is more beneficial in making sense of the numbers than simply increasing the numbers. And if I can get both an increase in numbers and an increase in qualitative inforamation we have a win-win.

(BTW I can understand why you might not like thermodynamics and/or statistics mechanics. There's a very good reason I never went there.... .

 

One good thing is it prevents the dude who gets a 2 ounce of pour of a beer late in the day at a big tasting (whose palate is probably shot) from quickly giving a rating and getting the same influence and somehow who actually had the time and place to do a review.

 

I see a lot of talk about statistics etc. Does anyone actually think the new top 250 list is more accurate than the previous one? Someone mentioned Trillium and Treehouse got hit pretty hard. They happen to be two of my favorite breweries. Would anyone here really rather have Stone IPA or Union Jack than beers like Mosaic Fort Point or Haze that are no longer on there? Really? If you have had them both it is an easy answer.