Customer Discussions > top reviewers discussion forum

Effect of negative votes on ranking

Sort: Oldest first | Newest first
Showing 101-125 of 391 posts in this discussion
Posted on 28 Sep 2012 10:08:24 BDT
[Deleted by Amazon on 19 Oct 2012 03:06:41 BDT]

In reply to an earlier post on 28 Sep 2012 10:16:19 BDT
Mr. J. Ryden says:
Exactly, which kind of proves my point about abolishing votes altogether.

It's no point reviewers whining about (allegedly malicious) negative votes when 'washing' and collusion is evidently fair game. Both are pathetic practices but there you go, human nature and all that.

Posted on 28 Sep 2012 10:17:49 BDT
[Deleted by the author on 28 Sep 2012 10:17:58 BDT]

In reply to an earlier post on 28 Sep 2012 11:49:35 BDT
[Customers don't think this post adds to the discussion. Show post anyway. Show all unhelpful posts.]

Posted on 28 Sep 2012 14:32:02 BDT
Thanks for the advice, Peter (if you're talking to me, which I assume you are).
Have followed it!

Posted on 28 Sep 2012 20:38:08 BDT
Last edited by the author on 28 Sep 2012 20:39:28 BDT
Quiverbow says:
"I think that the higher your overall % is is the key to getting to the top"

It isn't. There are people way down the rankings with high percentages, and also people high up the rankings with lower percentages.

"and that is done by deleting the minuses and getting a mate to vote for all your reviews once."

You can't give loads of positive votes to the same reviewer. Yes, you could vote on all their reviews but they won't count.

Posted on 28 Sep 2012 21:42:56 BDT
I bet the Amazon rankings formula makes the Olympics diving one look simple!

In reply to an earlier post on 29 Sep 2012 10:03:06 BDT
I don't know anything about the Olympics diving formula, but in conjunction with other reviewers here and across the pond, I have some idea of how Amazon's current (third) ranking system works. With apologies to those who have seen it before .....

The basics of Amazon's reviewer ranking system

The current system was installed in America in October 2008 and in Europe in March 2010. Nobody outside Amazon knows exactly how it works, but we have a fair idea. The system is complicated and deliberately so, but here are the main points as best we know them. This guide is periodically revised as necessary.

1 Rankings are calculated using an algorithm that takes into account both helpful votes and unhelpful votes, but also factors in recent activity.

2 Unhelpful votes weigh more heavily than helpful votes, and the weight ratio increases with the number of unhelpful votes there are on an individual review. As such, a review with 7 out of 10 helpful votes is not good, but one with 70 out of 100 is bad and 700 out of 1,000 is dreadful, if you are concerned about your ranking.

3 Although they don't weigh as much as unhelpful votes, it is advantageous to have one extremely popular review with hundreds of helpful votes providing it doesn't attract too many unhelpful votes. The best chance of writing such a review is to focus on gadgets. Whenever reviews of books, music, movies and games attract hundreds of votes, they invariably attract a significant number of unhelpful votes, however well they are done.

4 The overall percentage of helpful votes is shown for each reviewer, and this also appears to have some bearing on ranking, but I'm not sure if it really works that way. As noted above, a lot depends on how those votes are distributed. 100 unhelpful votes on one review will do more damage than 100 unhelpful votes spread over 100 reviews. Likewise, 100 helpful votes on one review will do more good than 100 helpful votes spread over 100 reviews.

5 Recent activity is another mystery, especially when it comes to any time intervals that Amazon use, but some things are clear.

5a Amazon's own blurb only mentions review dates, but it became clear soon after the system was installed that voting dates are also a factor. (My original American ranking on the current system was based largely on reviews that even then were between three and seven years old.)

5b If you don't post reviews for a while, your ranking will slide downwards. Remember that number 1 is top while the bottom can be measured in millions, so down means your ranking number gets bigger. Some people get this the wrong way round.

5c If somebody's ranking is primarily dependent on one extremely popular review, his or her ranking will fall noticeably once that review is no longer regarded as recent, even if he or she remains active while it continues to gather helpful votes. New votes will usually be at a much slower pace than when the review was first posted and this, as well as or instead of the age of the review, might be a factor. Thus, people who climb the rankings rapidly because of one extremely successful review usually end up falling rapidly.

6 As the rankings merely list everybody in the order of who has most points, including tie-breakers for equal points, our rankings are affected not only by what we do ourselves, but by what everybody else does. Thus, your ranking may go down because others overtook you by gaining helpful votes, posting new reviews or deleting unpopular reviews, or because your reviews acquired unhelpful votes.

7 Please note that all of the above is based on pooling of sometimes apparently contradictory information from and observations by a diverse range of reviewers with differing experiences. Depending on our reviewing habits and styles, we will notice different patterns on our own reviews. You may see claims that amount to "my ranking doesn't move that way". Where people have different experiences, there is usually (but not always) a ready-made reason resulting from different reviewing habits. Where no such reason is apparent, this presents an opportunity for further research.

8 Customers can vote however they wish, in the same way that voters at political elections do, so even the most brilliant reviews can be voted unhelpful and even the worst reviews can be voted helpful. In any case, customers look at reviews differently. Some prefer short, snappy reviews while others prefer more detail. Don't worry about it. You can't please everybody, so just please yourself.

In reply to an earlier post on 29 Sep 2012 10:06:39 BDT
[Customers don't think this post adds to the discussion. Show post anyway. Show all unhelpful posts.]

In reply to an earlier post on 29 Sep 2012 13:35:24 BDT
Fantastic, Peter, and thank you so much for taking the time. Much appreciated.

In response:-

1. Fair enough.
2. Why?
3. Hmm. Not much chance of getting hundreds, or indeed tens, of votes unless you're a very early reviewer.
4. Odd that.
5b. Not sure I quite understand.
6. Inevitable in a table of comparison.
7. Would this account for the seemingly inexplicable figures alongside the rankings of "neighbouring" reviews on the page one finds oneself on? Sometimes, it just not seem to make any sense!
8. Indeed. And that's what I endeavour do and what I've always tried to do right from the beginning.

Here's a further question for you, if you don't mind? When looking through the reviews from one particular reviewer, why can't one votes helpful or not there? It's a bit of a nuisance when you've read a cracking review or a particularly helpful one to then have to go and find it amongst the reviews on the actual product page.

In reply to an earlier post on 29 Sep 2012 15:44:09 BDT
Last edited by the author on 29 Sep 2012 15:44:55 BDT
Sophia says:
[Customers don't think this post adds to the discussion. Show post anyway. Show all unhelpful posts.]

Posted on 29 Sep 2012 16:50:57 BDT
Thanks Nemesis - v. useful.

5b - dur. I obviously didn't have my brain on this morning when I was reading that point! I didn't know that it went down to the million+ though - makes me feel a little better about a mid-10,000-ish ranking!

In reply to an earlier post on 29 Sep 2012 22:11:53 BDT
Last edited by the author on 29 Sep 2012 22:12:46 BDT
Quiverbow says:
The lowest ranked reviewer I know of is currently 2,297,593, so your ranking of 3,790 is pretty good.

In reply to an earlier post on 29 Sep 2012 23:01:10 BDT
I imagine it gets tougher the higher up - or rather, as you guys put it - the lower in the rankings you go. Or perhaps I should say, progress gets slower.

In reply to an earlier post on 1 Oct 2012 09:55:51 BDT
2 - most likely because Amazon USA didn't want Harriet Klausner to be #1, which she would have remained for eternity had they kept the previous ranking system.

3 - or if the only existing reviews don't say much.

4 - Books, music, movies and games are about opinions. Other stuff is about whether it does what it's supposed to do.

5b -Amazon prefer active reviewers.

6 - But you'd be surprised at how many people forget. (e.g. "I jumped 2 places but didn't get any votes and didn't post any reviews." - It might just mean that others fell.)

7 - The previous ranking system more closely reflected who had most reviews and most votes and could be said to reward hard work over a long period of time. Unfortunately, hard work doesn't always mean useful work. Hence the present system, which goes towards the other extreme in its determination to avoid the same problems. In doing so, it creates different problems.

Nobody should get obsessed with climbing the rankings. Amazon might realize that this ranking system, like its predecessor, is a pile of mucky brown stuff and bring out another algorithm.

In reply to an earlier post on 1 Oct 2012 09:58:35 BDT
1 million is nearer top than bottom. Amazon UK acquired its 2 millionth reviewing account some time in April. Across the pond, they seem on course to hit 12 million in December.

In reply to an earlier post on 1 Oct 2012 10:00:08 BDT
Higher up is the correvt term, but some people insist on using lower because that's what happens to the number.

Posted on 1 Oct 2012 10:22:18 BDT
Here's a new question: why do some reviews appear twice?

Posted on 1 Oct 2012 10:41:12 BDT
Three main possibilities

a) in the early days, especially before there was any option to edit reviews, posting of reviews was not only allowed but sometimes necessary, Two of the three reviews here are from the same anonymous reviewer, one being an update of the other

Connie Smith/Miss Smith Goes to Nashville

b) Amazon's software has a history of bugs, so accidental duplicates can occur. These normally have identical dates, but in the very early dates when getting reviews to appear at all,

c) More recently, Amazon UK publish American reviews where they have fewer than 3 UK reviews, but they don't check if the review is a duplicate. As I post in both countries and often review stuff that few others review, I have lots of duplicates, one of them being clearly labelled in green "Published in U.S."

Posted on 1 Oct 2012 11:01:39 BDT
I just noticed that Donald Mitchell Jesus Loves You gets a shedload of +votes for the first Montalbano book and then, a bit further down, gets another 30 or so for a duplicate review with the same date.

Posted on 1 Oct 2012 11:01:48 BDT
Bacchus says:
I started this discussion and have enjoyed the responses. However, I still don't think anyone has really answered the question; why do negative votes have a bigger effect on ranking than positive ones?

I have just suffered from a big drop in ranking when I got 5 negatives in a row, which was very different from the previous pattern of voting. The total drop was about 80 places. Had these been positive votes, the rise might have been about 20 places. The experience can be very frustrating.

I am not intending to resort to deleting and reposting reviews which have garnered negative votes in order to boast my score. It really should not matter that much. I would only do this if after re-reading a review, I find that it there is a lot wrong with it.

My theory is that the numbers of negative votes in total on Amazon is far smaller than the total number of positive ones. If they had equal power, the numbers of negative vote would have little impact of peoples overall ranking. I suspect that the ranking I have now would be the same as it is now if the rules were different.

Posted on 1 Oct 2012 11:07:39 BDT
As they said in Shakespeare in Love, "It's a mystery"!

Have you asked Amazon, Bacchus?

In reply to an earlier post on 1 Oct 2012 13:54:28 BDT
Bacchus says:
Given that if the rules were different my ranking would most likely remain unchanged, I have just learned to accept that a negative vote will have a disproportionate effect of my ranking. So I am not going to take it further.

However, I do not write reviews principally to get votes or to rise up some league table, which I have learned is a very greasy pole. I write reviews for the fun of expressing my opinion and sharing ideas with others. I would much rather someone writes a comment on one of my reviews or writes their own review giving a different opinion of the same product.

Still, it is nice to get a positive (or indeed a negative) vote because it shows that someone is actually reading what you have written and have felt strongly about it to exercise their opinion.

My relatively recent experience is that your recently reviews get more attention and are more likely to attract votes than ones you wrote a year or more ago. Therefore, I would argue that it is essential to keep writing reviews on a regular basis if you want to rise up the rankings.

In reply to an earlier post on 1 Oct 2012 14:53:48 BDT
Certainly, everyone seems to agree that you must keep writing to keep rising!

I am relatively new to all this but I am certainly forming the opinion that the league table is something of a greasy pole. Shame. I only started because I am an avid reader and, I suppose, something of a frustrated book reviewer. I know the press would never pay me to review so I thought it might be quite rewarding to do it for nothing on Amazon. And it is! However, I can see that once the rankings issue starts to bite, it can really take hold and you can find yourself (by which I mean I can find myself!)becoming a bit OCD about it. Worrying, eh!

To return to your original question, Bacchus, I think you perhaps answered it yourself earlier. If, overall, there are so many fewer negative votes than positive votes, perhaps any small increase makes a greater impact, algorithm-wise?

In reply to an earlier post on 1 Oct 2012 14:59:39 BDT
Mr. J. Ryden says:

d) ensure that one's review gains maximum exposure for the gargantuan effort of copy-and-paste.

Review a book? Stick two reviews up one for hardback, one for paperback
Review a film? Stick one up for VHS, and another for DVD; and another for BluRay for good measure.
Review an album? CD & Vinyl, bobs yer uncle, two for the price of one.

Ad nauseum
[Add comment]
Add your own message to the discussion
To insert a product link use the format: [[ASIN:ASIN product-title]] (What's this?)
Prompts for sign-in


This discussion

Participants:  67
Total posts:  391
Initial post:  12 May 2011
Latest post:  20 Sep 2014

New! Receive e-mail when new posts are made.
Tracked by 12 customers

Search Customer Discussions