|
The reputation system was mostly designed to award members for participation, and not so much as a driver of any particular behaviour, though this is certainly baked into it as well.
A couple of examples might make things clearer.
An author posts an article. That's awesome and they get 100 times the number of points that someone posting a message in the lounge would get (the base currency). The value of their article, however, isn't 100 points. The value completely depends on how the community accepts the article. The 100 points is merely a down-payment, so to speak, and as members vote for the article, download the zips, and bookmark it, more and more points accumulate providing the author with the true reward for an article. The weighting system kicks in heavily here and is designed specifically to counter sock puppets while also recognising the value of an experienced vote.
A different example is that of posting a witty message in the lounge. A message is posted, it gets its single point, and then members love it, hate it, or ignore it, and it achieves votes and bookmarks that provide an indication of the communities reaction. Members can accrue a large number of points for being active, interesting, or even just entertaining in the community, and can almost as quickly lose those points for being anti-social.
In both examples the system is designed to continue rewarding members for activity long after they have carried out an action. This isn't a system designed to guide you through a maze like a rat. This is a system designed to reward the members because we, as site organisers, want to recognise the contributions, in all ways, that members make.
For an author posting an article they will achieve Author points as their contributions are awarded. A small number initially, but then over time they gain the true value with points awarded from all members who are helped by the author. Similarly for those answering questions: they get Authority points. Those posting in the Lounge get Debator points which recognise their contribution to the community but do not, in practice, provide them much in the way of access to special functionality or rights.
The Debator points were a fundamental design of the system because we have many, many members who rarely answer a question, or have never posted an article or a tip, but who nevertheless may be in the core of the community. They provide great conversation, insights, help to others in general ways, police the forums and report malicious activity or just nudge members back on track. Even if points gained don't provide as many practical benefits as, say, an author, they are important and should be recognised.
There are definitely a few apparant inconsistencies within the system that could be debated forever (and probably will) but obviously no system is perfect. We only weight voting (opinion) actions and not usage actions like downloads or bookmarks. We value a download as 1 point and a bookmark as 5 points, partly because our feeling, and member behaviour, seemed to suggest a download was done "just to see" while a bookmark implied a member valued the item enough to revisit. We also don't reward views - partly because we don't have fractions of a point, but mostly because identifying a real view from a spider or bot is near impossible to do with accuracy.
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
Thanks. Good Reply.
-Richard
Hit any user to continue.
|
|
|
|
|
Although that explains a lot, it doesn't justify why a message vote is 24 but a download is 1. At the very least, a download should be 2 or 5 (same as a bookmark). Also, shouldn't a vote in a programming forum be weighted more (say 2) than in the Lounge? It would allow those who post technical suggestions in those forums to fare better than those who just use the Lounge for idle chatter.
|
|
|
|
|
The vote is weighted by the reputation of the voter, not by the location. That's why somebody like JSOP has more influence in his votin than somebody who joined yesterday.
|
|
|
|
|
I understand that, but a Lounge vote from JSOP (24) shouldn't weigh that much more than a bookmark or download vote from a newbie. I'm not for increasing those votes per se, but maybe the non-technical votes (Lounge, Soapbox, etc.) need to be taken down a bit.
|
|
|
|
|
Part of it's a matter of numbers. Downloads are set to only 1 point to keep them from totally overwhelming everything else. Look at Chris's[^] article list. On average, only a few dozen votes/article and thousands of downloads. Bookmarks might deserve a higher weight since they only appear to occur about as often as votes (probably because most people are bookmarking in their browser not CP; and aren't counted as a result).
3x12=36
2x12=24
1x12=12
0x12=18
|
|
|
|
|
Hmmm, the numbers explanation makes sense. I guess we need to teach people how to BM within CP. Maybe we need to weight everything (views + downloads + bookmarks). A lot of articles may not have downloads and are succint enough to help without having to be bookmarked. Food for thought, I guess.
|
|
|
|
|
Bassam Abdul-Baki wrote: Although that explains a lot, it doesn't justify why a message vote is 24 but a
download is 1. At the very least, a download should be 2 or 5 (same as a
bookmark).
It was that way initially, but the resulting top-scorer list was not considered ideal and it seemingly gave an advantage to authors with very popular articles. So the download points were tweaked until the resulting rep scores list were in sync with the popular expectation of what the top ranked guys should look like. And in my opinion, it was not a bad idea.
|
|
|
|
|
Then set a limit on the number of points each article can garner, or make it logarithmic (ideally). After all, an article that keeps on helping should keep on amassing and the top CP authors have earned it. Not sure I care if you, CG, or any of the others are ten times ahead of me or a thousand.
|
|
|
|
|
Bassam Abdul-Baki wrote: Not sure I care if you, CG, or any of the others are ten times ahead of me or a
thousand.
You probably don't. But a new author who considers himself to be way smarter than CG or myself, may believe that he's contributing far more to the site that either of us but that we are way too ahead of him to even give him a remote chance of catching up.
To work around that, I'd think it's a good idea to have an all-time score as well as a last 12 months score, with the default being the last 12 months score.
|
|
|
|
|
That idea would work nicely too to see who's improving and who's falling.
If someone thinks he's better than the top ten, we need to draw them out and challenge them to a duel, or put their names in a wall where they can be publically mocked (or rocked if they win).
|
|
|
|
|
Could be useful with the rise of G+ to get a new profile field for G+ accounts.
|
|
|
|
|
I'd hold off until it's known if G+ is going to survive any longer than any one of the other 300 betas that Google killed off over the last year.
|
|
|
|
|
+1
Henry Minute
Do not read medical books! You could die of a misprint. - Mark Twain
Girl: (staring) "Why do you need an icy cucumber?"
“I want to report a fraud. The government is lying to us all.”
I wouldn't let CG touch my Abacus!
When you're wrestling a gorilla, you don't stop when you're tired, you stop when the gorilla is.
|
|
|
|
|
Henry Minute wrote: +1
Univoter, eh?
"I love deadlines. I like the whooshing sound they make as they fly by." (DNA)
|
|
|
|
|
Just had a thought about reputation wearing off ( with a similar system to tennis ATP ranking, for instance) with time. This would:
- keep the reputation more "up-to-date", e.g. it is not because I have contributed ten articles ten years ago in VB6 that I am still "reputable" enough to answer ASP.NET questions. I take my own case as an example: I have not touched a line of code since 2007 and still profit from the reputation I gained answering questions in the C++ forum more than 8 years ago.
- give newcomers with good competences a hope to reach the "gods circle" of top posters/members, since the top scores would not be so high as today. It is nowadays impossible to outscore them, or even to become slightly visible...
|
|
|
|
|
Rage wrote: "gods circle" of top posters/members, since the top scores would not be so high as today
I doubt that.
The top gods (CG , Nish , Pete , ...) are still active so there reputation would remain pretty much the same.
And who says your not qualified to answer questions.
I'm sure if you wander into the c++ forum you could still answer some of the questions there and you'll probably find questions the same or similar to the one's you answered 3 years ago.
|
|
|
|
|
Tom Deketelaere wrote: would remain pretty much the same.
Exactly my point: keep the same, not climb to the sky. And by keeping the same, other could come in their reach.
Tom Deketelaere wrote: And who says your not qualified to answer questions.
Trust me on this one A lot of APIs have changed since MFC...
|
|
|
|
|
Rage wrote: A lot of APIs have changed since MFC...
Yeah but the questions haven't, that's the problem
|
|
|
|
|
So you're saying only show reputation based on x number of years, and every day, the points for the day beyond that time period is no longer counted? I think that merely calls for a different graph.
This graph would should the points gained for the last 365 days, or since Jan 01 of the current year, and would merely be an indication of daily accumulations of each category of points. I would certainly bring the less easily obtained points higher up on the graph. I envision trend lines and other features. I think Pete was supposed to be working on a Silverlight module to allow more flexible graphing of data, but I don't now what the status of it is.
".45 ACP - because shooting twice is just silly" - JSOP, 2010 ----- You can never have too much ammo - unless you're swimming, or on fire. - JSOP, 2010 ----- "Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass." - Dale Earnhardt, 1997
|
|
|
|
|
John Simmons / outlaw programmer wrote: I don't now what the status of it is.
Memory leaks in the Silverlight RIA services that I'm still trying to work around. Show too much reputation and *boom* the graph dies in a pathetic heap.
|
|
|
|
|
Pete O'Hanlon wrote: Show too much reputation
...and on my end I need to provide weekly or monthly summary data for the graph. Daily is too much.
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
If you could do that with the current ODATA service, this would ease the pain in the RIA.
|
|
|
|
|
I'm working on a desktop app (article) right now that only shows the most recent year's worth of rep data 8 series * 365 days = 2920 data points. Yesterday, I finally got moving on it again, and got the daily scrape/database stuff working. I now have a whopping 1 day (8 data points) worth of data. Of course, it only scrapes once per day and it's gonna take a while - well, a year - to get enough data gathered to see it's true performance characteristics, but I don't anticipate many/any issues.
The user will be able to display the data as daily, weekly, or monthly data, including trend lines, and end of week/month/year projection of points earned.
".45 ACP - because shooting twice is just silly" - JSOP, 2010 ----- You can never have too much ammo - unless you're swimming, or on fire. - JSOP, 2010 ----- "Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass." - Dale Earnhardt, 1997
|
|
|
|
|
Why are you using RIA? We started to and ran into the same problems, and switched over to a WCF web service with a single method in it (accepts stored proc name and parameters in the form of xml data, converts the xml params to sqlparameter objects, calls the stored proc, and the stored proc returns data as xml).
It works great, and is very performant.
".45 ACP - because shooting twice is just silly" - JSOP, 2010 ----- You can never have too much ammo - unless you're swimming, or on fire. - JSOP, 2010 ----- "Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass." - Dale Earnhardt, 1997
|
|
|
|
|