One feature I miss here is a fast* way to get all nodes that link to a particular node. That's something every scientific journal does (for example http://www.sciencemag.org/cgi/content/abstract/291/5507/1304#otherarticles).
Why? Imagine you found a nice old node with Super Search. Then you click on the "Linked by" link (I suggest the same behavior as the readmore tag) and you get a list of newer threads. If you are lucky, then these threads cover newer Perl versions or recent CPAN uploads. In general it is a nice way to find related nodes.
Second, it would be easy to implement a "Most linked nodes" list. This way, we all would have a common place for all the classic nodes (What was the id of the "Homework" node? Or the "How (not) to ask a question?" one) instead of scratchpad lists.
*The only way I see how this is now possible is Super Searching for the node id. This is slow, uncomfortable and gets a few false positives.
|
Re: Proposal: A "Cited/Linked by" list
by jhourcle (Prior) on Oct 10, 2007 at 14:39 UTC
|
Actually, what you mention was part of Ted Nelson's original vision of hypertext, but it never got implemented in those whole 'World Wide Web' thing.
I'm actually surprised that Google doesn't work for this, as much of their search engine is based on the concept of citation counting. (example -- How (not) to ask a question, searching by node id. You get NO results if you also add in 'site:perlmonks.org')
I like the idea ... and in theory, once you've indexed the site, you only need to update the links as nodes are modified. (although, for sake of completeness, you'd need something that could recognize [id://...], [node title], or a direct HTML link (and the direct link might be to perlmonks.(org|net|com). [node title] is the trickiest (or <a href...> using a node's title), as the target of the link when the node was written may no longer match the current destination of the link. Of course, I have no idea how difficult it'd be to actually implement or maintain something like this.
| [reply]
[d/l]
[select] |
Re: Proposal: A "Cited/Linked by" list
by CountZero (Bishop) on Oct 10, 2007 at 16:27 UTC
|
I am very much against such a thing.All this citation counting and such leads to crappy statistics. For instance the Shanghai University Academic Ranking of World Universities is based (for 60%) on just such a thing and --surprise surprise-- as the citations they check are mostly if not all in the English language, European Universities rank consistently low. It also leads to vanity "Look how many people have cited my node" and that is a sin one should avoid in a Monastery. Finally, a well phrased Super Search will unearth the nodes you are looking for anyway as the linking articles are very likely to deal with the same subject (and if not the link is probably not useful for you). Still, I ++ your node because it shows that you have thought about the well-being of our Monastery and that is good.
CountZero A program should be light and agile, its subroutines connected like a string of pearls. The spirit and intent of the program should be retained throughout. There should be neither too little or too much, neither needless loops nor useless variables, neither lack of structure nor overwhelming rigidity." - The Tao of Programming, 4.1 - Geoffrey James
| [reply] |
|
|
| [reply] |
|
|
It's actually pretty easy to abuse (in a general sense -- for bibliographic references, or blog pingback (or whatever the hell they call them)) -- you link to a highly scored article, in order to draw more attention to your own article by someone doing a literary search. Some of the citation tracking systems will rank the quality of the link, to use in weighting reverse links. (eg, if someone cited the article 12 times and only cited 10 articles total, that should carry more weight that someone who cites in once along with 80 other articles).
Typically, where I've seen problem is that because people know about the whole 'reference counting' thing, they intentionally cite ALL of their previous articles, no matter how relevant it really is to the current article. When you only get 1-2 articles per year, and publishing affects your income, some folks will do anything to draw more attention to their research.
For perlmonks, though, I personally think that the issue won't be as bad due to the reputation system (we can always -- nodes if people think someone's abusing the system), and I think that the benefits would outweigh the potential risk.
| [reply] |
|
|
So, you're against this because somewhere someone has crappy data analysis for a puff piece on academic rating? What does that have to do with a reverse index, which is amazingly useful? It's a standard part of library research. No one said anything about ranking nodes or users or making statistical reports. It's just connecting nodes about the same topics to each other to make it better for the user.
Don't give too much weight to an unrelated social consequence. Don't ever make anything useful lest someone might misuse it! Oh, the horror!
| [reply] |
|
|
Nothing wrong with a reverse index, but we have that already in the form of the Super Search function.
CountZero A program should be light and agile, its subroutines connected like a string of pearls. The spirit and intent of the program should be retained throughout. There should be neither too little or too much, neither needless loops nor useless variables, neither lack of structure nor overwhelming rigidity." - The Tao of Programming, 4.1 - Geoffrey James
| [reply] |
|
|
| [reply] |
|
|
Yes, unfortunately in Physics and such, English is the main language, but if you publish e.g. about Belgian law, it is most likely to be in Dutch or French and then that does not get counted. So libraries full of good stuff just sail under the radar of crappy statistics.And yes again, I know that a "cited by / linked by" system within our Monastery will not be the end of the world, but its perceived relevance could easily outgrow its real value. Suppose you have a comment on a node that says: "This node has been cited by / linked by 100 other nodes". Unless you go out and check all these citations, it would tell you nothing if this node is good or bad. Perhaps the links were just of a "this node is really bad" sort. But wouldn't the ++/-- votes tell you that as well and be much easier to check?
CountZero A program should be light and agile, its subroutines connected like a string of pearls. The spirit and intent of the program should be retained throughout. There should be neither too little or too much, neither needless loops nor useless variables, neither lack of structure nor overwhelming rigidity." - The Tao of Programming, 4.1 - Geoffrey James
| [reply] |
Re: Proposal: A "Cited/Linked by" list
by ambrus (Abbot) on Oct 11, 2007 at 06:20 UTC
|
In my practice, Super Searching is fast enough, comfortable, and doesn't get many false positives when you search for a six-digit id.
| [reply] |
Re: Proposal: A "Cited/Linked by" list
by ambrus (Abbot) on Oct 11, 2007 at 06:26 UTC
|
If a node has other related nodes, those should be mentioned from either that node, or from a reply to that node. I sometimes post replies that consist of only internal links.
| [reply] |
Re: Proposal: A "Cited/Linked by" list
by husker (Chaplain) on Oct 11, 2007 at 14:27 UTC
|
I asked for this feature some time ago: Node Trackbacks?
Based on the relative reps of our two nodes, it may be the time has come for this idea. | [reply] |
|
|
