Morgan Lalique
Member
Z
Last edited:
These should be combined now
AgreedThe first instance of the photo does not have a celebrity link, even though I think it should
No way, the pink dress picture is meh... the other one is way hotterNo, I just think the photo in question is way overrated by the system, it should be 9.4-5 at best. Personally, I think the photo of her in the pink dress is hotter.
Agreed and fixedAnd, it isn't particularly salacious, it could have a celeb link no prob imo.
No because we can always use more votes. An idea that gives more votes does not mean we can use a system that ranks with more votes. It just means our current system would benefit from more votesWith this change, your objection to my system of not having enough votes would be solved.
I'm not sure if you think it should be higher or lower. I mean she is pretty hot but that stupid look on her face brings it down. Seems pretty accurate to me actually.8.37???? Seriously? From 71 "votes", no less. Your system fails to weed out the juvenile h8ter votes. That is what your total lack of validation and statistical significance permits. It is impossible to put any reliance on any rating your system currently produces.
right here: https://hotnessrater.com/picture/4398346/jana-jordanBTW: where is the listing of her "wins" and "losses"?
What is this skew and invalid results you speak of?I think it would still suffer from the same skew and invalid results you get now with your 1 vs 1 system
I get lots of validation. 1 vote doesn't make or break a rating. In fact, it is only even counted if the pictures are close in ratings, and if they are close in ratings, who is to say the vote was invalid?because you would still never get any validation of voting results before you build your ratings on them
Yeah well I think the same thing about you.but you do have some peculiar ideas when it comes to ratings and priorities
I don't know about that. The desire to not throw out years of work and start from scratch is a driving factor though. Truth is, you just haven't offered anything I haven't tried in one form or another and haven't convinced me that it would be worth the effort... From my past experience, it would just burn through too many votes. What's better, 10 data points that are 80-90% reliable or 100 that are 70-85% reliable? I don't believe the best of 5 votes or 10 votes will get you a proper winner in each case. I also believe you can obtain a lot of information by beating a 9.0 75% of the time over 25 votes when each of those 9.0s beat a 8.75 75% of the time over 25 votes.I think your emotions are involved, and they are driving you to defend your decisions.
Pictures would have more than a 5% chance of error just by looking at them several times. If you gave me 2 close match ups several times, over and over on different days, I would probably pick a different winner. This effect is amplified greatly when you open up voting to all different kinds of people all over the world. Men and women vote. People from different countries vote. Some people love Kim Kardashian. I'd give her a 7. With that type of disagreement, you aren't going to differentiate between 9.5 and 9.3 Even with 1000 votes those 2 pictures could go either way depending on who voted.If all pics on the system are rated, and all the ratings have less than a 5% chance of error
And I've seen lots of battles of the day that after 100 votes, someone won 80-20 and I completely disagreed with the result... so batch size does not bend the results to your will. There is no accounting for personal taste.I see a LOT of pics whose ratings give me a headache.
The thing is there isn't necessarily a better picture when given 2 pics. A lot of it is up to personal interpretation. I believe my system captures thatNo, my system would leave such close votes as undetermined because they don't meet the minimum required confidence level (statistical significance). It would assgn more voting tasks for that match (not infinitely, at some point it would be deemed a tie, or whatever conclusion could be made with at least a 95% confidence rating. I mean, if the vote is like 55-45, I'd have to get my stastics book out to calculate if that is a large enough sample size to achieve an error rate of less than 5%.
hahaha... that in itself is ridiculousIf all pics on the system are rated
Wow I did not get that from that pictureWell, clearly they caught her thinking about licking someone's pussy
BabesRater doesn't show wins/loses yet. Its on the listNot on that pic, on the one rated 8.37.
I mean it is ridiculous because I get pictures in faster than we get votes to rate them.Hmmm, yeah, you are sort of right. There could be new unrated pics, but they would have matches created for placement at the time they enter the system. Beyond that, though, the system would create new voting tasks as I described. The purpose is always to increase accuracy. Your system seems to create new votes for pics just because they won a lot. That spends votes without gaining any accuracy.