| |
Fungus
Registered: Sep 2002
Posts: 680 |
Votes
Why do some of you hide your votes? HMM? this makes no sense. And stop downvoting people just because you do not like them.
This database is supposed to be accurate, not a place for you to continue with your stupid grudges and scene wars.
a note to perff: since Secret Man, WDR etc have been deleted from this, I think you should also remove all their votes.
and get rid of this hidden voting, if people cannot publicy stand by their vote choices, then they shouldn't be voting at all.
|
|
... 114 posts hidden. Click here to view all posts....
|
| |
CreaMD
Registered: Dec 2001
Posts: 3052 |
Quote: i still cant see the problem. a few individuals "downvote" people they dont like? so what? if the votes of these few individuals have enough impact on the result to significantly change it in a "wrong" way it only means one thing: the result is calculated on too few votes. given enough votes, up- and downvoting is irrelevant. given too few votes, the calculated result is irrelevant - in any case, with or without up or downvoting.
Yes there is system of weighted voting based on user-trust implemented in one of the websites I maintain. It would require to have evaluation of CSDB users between themselves. E.g. everyone could give -5 to +5 points to everyone else in the system. From the whole data set of relationship the +/- "trust" of user's vote could be counted. That way votes from someone who has biggest trust would have the most impact. And votes from someone with least trust would have minimal effect on resulting score.
The calculated "trust" can be used for anything else. e.g. for giving higher level of user rights. E.g. possibility to unlock and maintain locked entries, etc. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11360 |
yeah great idea, lets invent another flawed thing to doctor the symptoms of a flawed system! simply raising the amount of votes needed before a result is shown to a reasonable value (5 ??? LOL! and you wonder the charts are crap?) would be too easy i guess :=) |
| |
Oswald
Registered: Apr 2002
Posts: 5086 |
either stop whining or make them public. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11360 |
Quote:
or make them public
and that would change... what? read it again: given enough votes, up- and down- and public- or nonpublic voting is irrelevant. if you want reasonable charts, you need enough votes. it IS that simple.
|
| |
CreaMD
Registered: Dec 2001
Posts: 3052 |
Quote: yeah great idea, lets invent another flawed thing to doctor the symptoms of a flawed system! simply raising the amount of votes needed before a result is shown to a reasonable value (5 ??? LOL! and you wonder the charts are crap?) would be too easy i guess :=)
It's basically the same thing but from the opposite point of view ;-). While in your solution (that actually alreay exists here) the number of votes is hard to achieve. In my solution the trust can be obtained relatively easily by voting between the existing users.
Google's pagerank works around similar idea and although it is also open to manipulation (which forced it's developers to count many other factors into the account) the basic idea of "trust" voting works better than anything else invented in the search engine world.
I'm sure this trust-distrust algorithm would work in the closed group of users especially when there is less posibility to have fake user accounts. We are of course discussing this hypothetically coz I don't believe such thing will be ever implemented here (although I could provide the algorithm I use, if Perff wanted it ;-)
|
| |
Burglar
Registered: Dec 2004
Posts: 1089 |
we've had these votediscussions for a very long time now, and nothing ever changed. I dont think anything is gonna change now, so it's another useless thread... |
| |
yago
Registered: May 2002
Posts: 333 |
Strange Discussion... here my 0.02:
Every real scener should vote like he wants... if all he does is downvoting, why not ?
Votes should not be public. If Votes become public, there will be even less Votes, and _that_ is the real Problem on csdb: too few votes
|
| |
chatGPZ
Registered: Dec 2001
Posts: 11360 |
Quote:
It's basically the same thing but from the opposite point of view ;-). While in your solution (that actually alreay exists here) the number of votes is hard to achieve.
calculating a result based on the votes of 5 users out of 3844 (~0,1%) is not a "solution" - its a joke.
looking at the top 10 groups, the amount of people who voted ranges from 11 (~0,3%) to 137 (~3,6%).
wohoho, some seriously meaningful statistics there! not!
Quote:
In my solution the trust can be obtained relatively easily by voting between the existing users.
please explain how voting and charts are related to "trust" and how giving different people different "power" on their votes would not distort the result even further.
|
| |
CreaMD
Registered: Dec 2001
Posts: 3052 |
Quote: Quote:
It's basically the same thing but from the opposite point of view ;-). While in your solution (that actually alreay exists here) the number of votes is hard to achieve.
calculating a result based on the votes of 5 users out of 3844 (~0,1%) is not a "solution" - its a joke.
looking at the top 10 groups, the amount of people who voted ranges from 11 (~0,3%) to 137 (~3,6%).
wohoho, some seriously meaningful statistics there! not!
Quote:
In my solution the trust can be obtained relatively easily by voting between the existing users.
please explain how voting and charts are related to "trust" and how giving different people different "power" on their votes would not distort the result even further.
Ok. So we want to achieve the system that serves the best results for majority of the CSDB users.
We can't force them to vote for every single release. But we can achieve that most of them can vote for anyone else around this system and give them from max to minimum trust points. E.g. as in the mentioned site +5 to -5 .
If you have e.g. 10 people and some of them evaluate some others. Based on how they know them (like them, trust them) whatever. Then from this set of relationships the trust rank is compiled. Not everyone has to vote for everyone else. But if someone doesn't have any relationship his trust isn't moved to any direction and stays neutral.
Usually the one who is most trusted is someone who can behave, have some responsibility and is generally accepted by the majority. On the other hand, his trust to some single users from the group of people also gives them authority. It's an intertwined system and resulting trust rank is obtained after some ammount of iterations. I use 100 iterations.
And now the anwer for your question. Someone with no trust attributed will affect the results only slighlty. It can equal to some neutral trust e.g. 50%. Someone with most distrust will go under 50% and someone with positive trust will have bigger impact on results. The way how it will be implemented into the existing "vote counting" depends on how Perff counts the votes, but generally the most trusted user should have closer to 100% efficiency on his votes. And most distrusted user should have closer to 0% efficiency. That means. The more trusted user's voting will have bigger influence on final result than least trusted. Least trusted users will have no very small inflience on results.
In the end you get get fairer results even for small ammount of votes.
|
| |
chatGPZ
Registered: Dec 2001
Posts: 11360 |
Quote:
In the end you get get fairer results even for small ammount of votes.
no, you will give more impact to votes by people who are most liked, and take power from people who are disliked.
i would not call that "fair". i would call it up- and down voting =)
|
| Previous - 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 - Next |
