• To log on and post you will need to create an account in the forums even if you already registered on the main Hotness Rater site. These registrations are separate.

Votes but no rating

HotnessRater

HotnessRater

Administrator
Staff member
Morgan Lalique said:
So, that is 10 total votes spread over 10 images, with combined views totalling 7878.

Wouldn't it be more efficient to assign matchups where both pics could use more votes? And shouldn't pics with a lot of views get some priority?
Click to expand...
I limit the number of times the same girl is put into the queue. So if another picture of her is put in the queue, other pictures are less likely to get votes. I like to spread them evenly across girls so that when I get someone uploading 10,000 pictures of the same girl, it doesn't take over the queue.

So no, number of views is not a factor I use for putting pictures in the queue.

In this case, she had a picture that generated a lot of traffic from reddit. Many people hit the next button several times and generated a ton of views of a lot of her pictures.

That doesn't mean I am going to flood the queue with multiple pictures of her
 
HotnessRater

HotnessRater

Administrator
Staff member
Morgan Lalique said:
Well, this makes previous weird listings look normal:

https://hotnessrater.com/full-sized-picture/2998502/gabriella-wilde

92 votes, no rating.

https://hotnessrater.com/full-sized-picture/2998505/gabriella-wilde

119 votes, no rating. Oh, and 24 lifetime views.

Your code seems to have some non-deterministic qualities. I'd hate to have to write up a finite state diagram for it.
Click to expand...
That's because the voting is all over the place. Winning against high 8s or even 9s and then losing against 7s. When the data is all over the board like that, it needs more votes to decide which ones are bogus and which ones are legitimate.
 
HotnessRater

HotnessRater

Administrator
Staff member
Morgan Lalique said:
Well, I never suggested you do that. But surely there is some sensible middle ground, where high traffic can have some influence. I often come across pics with hundreds of rating votes but less than 50 lifetime views. That makes no sense when pics with thousands of views not only don't have a rating, they don't even have any votes, and aren't going to get any, either.

Rating the stuff people want to see seems more important than rating low visibility, low interest stuff. You don't have to abandon spreading rating votes around in order to provide a little priority for pics that suddenly become busy. I wouldn't shift priorities for one-night-wonders, but when pics lead the view list for weeks on end I think it makes sense to shift some extra resources their way.

I guess I'd do something like 20% to the top 100 pics (for those pics with low vote numbers, like < 100), and 30% to the top view getters (unless they are already well-rated), and 50% as you are doing now.
Click to expand...
Through out the years I have added different categories of pictures to the queue. Pictures who have went the longest without a vote, top 100 rated pictures, new pictures limited by user... all sorts of categories. I could see adding another batch for unrated, highly viewed pictures. That makes sense
 
HotnessRater

HotnessRater

Administrator
Staff member
Morgan Lalique said:
Well, yeah, amend that. There is no wayto tell on any particular vote, I just meant that a wide difference of opinion was likely because of the nature of the photos. They're not like bikini shots,or headshots, where there is more general agreement of opinion. I'd bought about 500 pics, my brain was a bit fuzzy in what I was trying to say.
Click to expand...
I agree that these types of pictures with get a wide difference of opinion... which is why it ends up taking more votes to figure it out
 
HotnessRater

HotnessRater

Administrator
Staff member
Morgan Lalique said:
If that rating is accurate, then in a Battle of the Day it should score near 50% but be a clear loser to this pic:
Click to expand...
Why would it be a clear loser if their ratings are the same? That makes no sense

Morgan Lalique said:
which your system also scores as a 7.52, with 584 votes, making it presumably a very reliable rating, to which it has already "lost".
Click to expand...
I'm not sure why you think it has already "lost".

584 votes doesn't necessarily mean it is a very reliable rating. A lot of those competitors pictures were too far out of her range to be used in calculating rating. Really only about 30 of them were used in the final calculation.

Besides, her picture has stayed in the queue and since your last post, Gabriella has went up to a 7.67

Also, this isn't a 100% exact science. It is based on personal preference and the whims of the voters. There is no guarantee a 7.66 is hotter than a 7.65. That type of granularity is simply not going to happen. I wouldn't even guarantee a 7.6 is hotter than a 7.3.... Probably most of the time they will be but I am sure you could find exceptions if you looked hard enough. The whole premise is a rough estimate really.

I know you think they system should divide pictures into 3 decimal places with 100% accuracy but that just isn't going to happen. Again, if you think you can do better, show me the algorithm, complete with all the details I would need to code it and I'll run the two systems in parallel.
 
You must log in or register to reply here.
Top