|
You've lost me completely here.
Chris Maunder wrote: a download is just a very easy click that doesn't take much thinking about
and bookmarking takes a huge effort and a long time? 
Chris Maunder wrote: If possible I would prefer to award 5 points for downloading, opening and then viewing the source code.
do you expect people to start a download, wait, spend the bandwidth and monthly ISP volume, and then typically not open and view what they got?
According to the current FAQ I can bookmark one of your articles and download its code, giving me 2+2 Org points and you 5+5 Author points.
Which action is the bigger and more significant one, downloading or bookmarking? For me the answer clearly is: downloading.
Also, in one minute I could browse to 10 articles (not read them!), just bookmark them; I wouldn't be able to download their ZIP files in that short a period.
IMO downloading should be worth a multiple of bookmarking, say three times more, and that to both parties involved.
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
The download links are front and centre and are clicked far more often, and I would also say clicked just to see what it is. The bookmark link is not as obvious as the links glaring at you from the top of the article, and I would argue are only clicked when someone feels the article is worth more than just a casual "what's this". I never said "huge" effort. I do maintain that bookmarking an article is a more conscious effort and less of a "why not - I'm here, I may as well" effort.
There are 5.2 times as many downloads as bookmarks in case you're wondering about the relative frequency.
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
Your logic clearly isn't mine.
So you want to compensate the imbalance in the observed frequency of bookmarking versus downloading (which may be due to an imbalance of the visual attraction of the related tools, or to the fact that bookmarking not really yields much to the bookmarker), by introducing an imbalance in rep point rewards, which goes completely against the idea of "proportional to effort and added value".
Two things to ponder:
1. If you had to scrap one feature about articles, which would it be: downloading, voting, or bookmarking?
2. which of these four is least relevant to you:
2a. someone gets Auth points for his article being bookmarked
2b. someone gets Org points for bookmarking an article
2c. someone gets Auth points for his code being downloaded
2d. someone gets Org points for downloading some code
My view is clear:
1. bookmarking is the lesser value as downloading yields usable code, and voting yields an aggregate score; bookmarking in my view yields almost nothing (it may mean: I want to be able to refer to this later as it is great; or I may want to maybe refer to it as I did not read it and maybe I should; or I can get easy points here). Me bookmarking something is for my organization (my bookmarks aren't public, hence they are of no value to the community), me voting something up is telling the community I liked it, that adds value for the community.
2: d is completely superfluous, and c is worth much more than a. I suggest b<a (and b=0 would be OK too, but I don't mind too much as long as it is less than a), c=3*a, and d=0.
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
I think you once mentioned that rep points should not only be for the users, but should also be used to ensure users behave in a way that improves the site. This is exactly what bookmarking does, because from my experience an article bookmarked typically means an article valued, and this in turn gives us another method to sort the wheat from the chaff. As you say, bookmarking provides no actual (immediate) value to a member, but it does provide an indication that the member will return to the article, that the member valued it enough to not just blow it away. Downloading, on the other hand, is an action easily done in the moment (hence the 5:1 ratio). You ask a real estate agent or a car salesman what's more valuable: someone coming into the house or taking a car for a drive, or someone coming back a second time.
An alternative is to give 1 points for a bookmark, 1 point for a download, and then award, say, 10 points if the bookmark occurs after the download. We can reasonably infer that there's a decent change the reader had read the article more fully than just glancing at it. However, this idea totally fails for articles with no downloads.
Or we could award extra points for each visit, with a sliding or log scale based on the number of visits within a certain number of weeks, ensuring that after a set number of months the points reset to their base value if the member starts visiting the article again after a hiatus. But that would just be getting ridiculous.
Or, we could have a point decay system whereby anything over 5 years gets dropped.
Or we could embed a ReadMe.html file in each download with an image bug. When the reader opens this readme, and the image is requested from the server, then we know they've downloaded, opened and read at least one file. I'm sure no one would mind the invasion of privacy.
My overarching desire is to keep points reasonable (and simple). If we have 5 points per download then I crack the 1 million point mark which, frankly, seems excessive.
Regardless, I will now have this same conversation all over again with John Simmons, but in the opposite direction (0.1 or 0.01 points per download). As a strong author himself I have to take his views seriously too.
I'm not going to make anyone happy here except possibly the authors who are now, at least, getting recognition not just for posting, not just for getting votes and bookmarks, but also for downloads.
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
Just to add on to the discussion, most of the times, downloading is something one just do it for the sake of seeing if it is going to help (maximum downloads would be following this thrend) - download 5 similar controls and chose one. But bookmarking is making sure that I do re-visit the article and want to keep an eye on the same as it looks good.
Further, I (and I believe lot others) use bookmarking as a feature to track the article and use it as and when required.
Lets say. Your Luc's CP Vanity article, I have bookmarked it and kept. Since it is bookmarked, it has helped me quickly to move on to it. I have downloaded it some 15+ times till now. Whenever I need, I download it, use it. (It's hard to maintain the download binaries always and on all the pc's I work on!)
|
|
|
|
|
Chris Maunder wrote: I think you once mentioned that rep points should not only be for the users, but should also be used to ensure users behave in a way that improves the site
I said something like that, in different words, much simpler IMO: you will get what you reward )whether you like it or not, so think carefully).
The reason is:
- either people don't care, and then there is nothing you can do; these people should not dominate your decisions!
- or they do care and will adapt their behavior to your reward system.
So you decide what you want, then make sure your reward system is in line with it and does not stimulate behavior you do not want.
---------------------------------
For rep points I suggest one always awards something proportional to "effort and added value", which means whatever is harder to do (effort) or adds more value for the community (such as voting, as it builds scores), should be rewarded more, hence proportional to effort+added value.
And nothing should be rewarded too high or too low. If you give 1000 points for bookmarking, I will stop posting messages, I'll just click three bookmark icons a day and be done. If in my view you give insufficient points for posting an article, I will no longer post articles. That is what a reward system is supposed to do, and believe me, it works, except for those people who don't care.
---------------------------------
[EDIT]
Sorry, what followed has been completely wrong; I now have replaced it.
There was a bug in the votes, downloads and bookmarks as handled by CP Vanity (all version up to and including V2.2 failed to properly handle numbers with thousand separator, I'm puzzled nobody ever noticed). Here is the adjusted table and my new comment.
FYI (assuming the Who'sWho numbers are reliable, article counts include T&T) here are some VIP figures:
member total total total articles
votes downloads bookmarks count
Sacha 9,023 37K 17.0K 175
Marc 4,144 64K 11.2K 137
Nish 6,472 156K 7.8K 133
JSOP 1,944 10K 2.5K 84
#1 5,540 210K 7.0K 104
.dan.g 2,666 168K 4.0K 16
CG 1,632 24K 2.9K 37
From these numbers, I see total download counts are huge, and I now understand you have a problem with the rep points.
Using the same letters as before:
(a) Getting bookmarked is not a problem; the only comment that I would give is: if it resembles anything it would be: getting views, which aren't rewarded. I would give no more than 1 point for getting bookmarked.
(b) Earning points by actively bookmarking is not documented by the table; it is probably very distributed over the community, I don't see much value for the community as the meaning of a bookmark is uncertain, and I would lower its reward, to one or zero.
(c) Getting downloaded is like getting up-voted. I'll present some ideas below.
(d) Actively downloading is not documented by the table; it is probably not a problem as I guess it is distributed very much over the community; however IMO it does not deserve any reward. There is no value in it, all the community sees is the download count and the author's rep going up.
Author points for getting your code downloaded, four ideas:
1. CAPPED: give 5 points each, and simply cap it to the first N downloads per article; N could be 1000, or it could equal the number of votes for that article.
2. COLORED: give 5 points for each download by a colored author (colored as in bronze/silver/gold/platinum), i.e. ignore blank author downloads.
3. PSEUDO-LOGARITHMIC: works in theory, it would not look nice in the rep history table, as individual downloads would fall below 1 point each at some point, what do you show in the history table then?
4. HYBRID: give 5 points for each download by a colored author, i.e. ignore blank author downloads; add 1 point for each download by a non-colored author, but this limited to N (e.g. N=1000).
My preference would be #4, then #2.
[/EDIT]
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
signal: I heavily modified my earlier reply. Please check.
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
Luc Pattyn wrote: my bookmarks aren't public, hence they are of no value to the community
You can make it public. It's on you. Settings are provided for the same.
Luc Pattyn wrote: bookmarking is the lesser value as downloading yields usable code
Already shared my thoughts here.[^] If bookmark feature is least important than it would be real difficult to keep track of articles and I for one cannot just download the binaries of each and every article and organize it on my system properly such that I can refer it as and when needed. Further, doing this would not give me mobility and I cannot refer them if I am not on my personal system.
Luc Pattyn wrote: If you had to scrap one feature about articles, which would it be: downloading,
voting, or bookmarking?
Though all are important, still, if you ask me than I will opt for hiding download option. A good article would exlain the important code snippets in the code itself and that should be enough as an idea to move on. (Leaving those who wants exact thing without doing a single line of code by themselves.). Now, I can further add that, lets show the full code (keep browse code feature) but hide download. It would still do. But removing a bookmark feature is like multiplying the effort of getting back to that article number of times, just like finding a needle in the haystack....(or lets say not that much, but still a good amount of effort! )
|
|
|
|
|
I don't want to make my bookmarks public, they are mine, for my organization, and they do not express my appreciation on anything.
The bookmarking system is weak (it does not provide searching and sorting, it only handles CP material, etc), and CP has pretty good search facilities nowadays. I use it all the time, I seldom use bookmarks to return to an article I had marked.
I strongly recommend you investigate CP search, it is good!
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
Luc Pattyn wrote: I don't want to make my bookmarks public, they are mine, for my organization,
and they do not express my appreciation on anything.
So do I. I agree to it. Since you said they are not public, I thought of sharing it with you that they can be made public if someone wants to.
Luc Pattyn wrote: The bookmarking system is weak (it does not provide searching and sorting, it
only handles CP material, etc
Partially agree. Current/new search give you this feature to search in your bookmarks. To an extent some sorting too.
Luc Pattyn wrote: I strongly recommend you investigate CP search, it is good!
No doubt, I have started using it. Using 'Just my stuff' helps me a lot.
|
|
|
|
|
Saying that bookmarking counts for 5 but downloading counts for less than that is ridiculous. Download points should be at least as much as bookmark points; and since anyone can slap together some BS with no download, downloads should count for at least twice as much as bookmarks.
You're concerned rep points are getting too high for some people? So what!?! You have said in the past that authors should be given more recognition, and now you are tweaking the rep points once again to ensure that they don't. The worst thing about this whole situation is that you have completely abandoned any pretense at what is fair and equitable for a specific action (like downloading), and instead are using the totally lame generic excuse "But people will accumulate too many points".
Next to actually writing an article, downloads are a primary indicator of an author's true reputation; people wouldn't waste time downloading anything from a crappy article. And now, downloads are to be treated as drive-by bookmarks?
This is not fair, not equitable, and sends totally the wrong signal to anyone thinking of writing an article.
|
|
|
|
|
"sends totally the wrong signal to anyone thinking of writing an article"
Steady on there Hans. We're not slapping them in the face with a wet fish. We're awarding points. The discussion is on how to balance the points.
Do you really feel that a bronze member voting 5 for an article is only worth twice what a bronze member downloading a zip is worth?
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
Chris Maunder wrote: The discussion is on how to balance the points.
What's about this (rule per user/file):
- first download: 1 or 2 points
- any following download: 5 points
And maybe there is a maximum of total points per download file - for example 1000.
|
|
|
|
|
For once in a lifetime, Hans and I seem to agree, and yet you don't yield? 
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
For now, I am basking in the illusion that Chris secretly agrees with us, but wants us both to suffer, for all the crap we've given him.
|
|
|
|
|
That had occurred to me too. All he will achieve is the prolongation of his suffering though...
I'm off to bookmark a couple of things now.
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
Hans Dietrich wrote: people wouldn't waste time downloading anything from a crappy article
Well, I have seen it and they do. Very few(generally regular users) download only good ones.
People search, they find some 10 related articles to the topic, create a login, download all 10. Gone.
How many regulars do this? I hope you agree that it would be very few comapred to such users.
|
|
|
|
|
S Mewara wrote: People search, they find some 10 related articles to the topic, create a login, download all 10. Gone.
How many regulars do this? I hope you agree that it would be very few comapred to such users.
I'll do this on occasion although I don't think it's ever been more than 4 or 5 articles that triggered downloads. Some of them almost always turn out not to be what I need, some turn out to be seriously buggy, hopefully at least one is what I need. The ones that turn out to be not what I need rarely get a vote. The ones that are garbage will get low votes, and I don't like the idea that having to look at the code to determine the article is crap will counter a large part of my vote.
3x12=36
2x12=24
1x12=12
0x12=18
|
|
|
|
|
same topic, similar view, some suggestions here[^].
Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum
Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.
|
|
|
|
|
Luc Pattyn wrote: it is bookmarking that is way too high
Luc, I agree to it that bookmarking is also high but 5 for downloading too is.
|
|
|
|
|
Even 2 is probably too high. I'd say 1 point per DL unless major amounts of rep are also added in other areas and threshold levels updated as well; less if the infrastructure can support it.
Maunder has about 200k downloads, .dan.g. (todo list) has ~165k. None of the other prolific article authors I looked at were close. IIRC one of the prolific authors I looked at had ~100k. Chris didn't like the idea of being catapulted to the top of the rep chart. More generally this would favor articles with downloads much more than other types. Writing something popular as massmarket freeware and having CP host it would skew it even more (intentional or not .dan.g.'s article is the 8th hit on google for: todo list), and arguably is giving rep for the wrong sort of thing entirely.
3x12=36
2x12=24
1x12=12
0x12=18
|
|
|
|
|
Ok! I did not went into that much detail.
Based on what you say, even 1 point per download would be high and not do justice with all. 
This needs to be reduced on to some 1 point per 100 download or so. 
|
|
|
|
|
That's ok, apparently neither did anyone else working at CP before setting it to 5 points (for new downloads, they haven't parsed the logs for historical data yet). 
3x12=36
2x12=24
1x12=12
0x12=18
|
|
|
|
|
I think downloads is actually a very good measure of the article's popularity. 5 is too much indeed, but 2 or 3 would be perfect.
Nick Polyak
|
|
|
|
|
Dan Neely wrote: apparently neither did anyone else working at CP before setting it to 5 points
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
