Beefy Boxes and Bandwidth Generously Provided by pair Networks
Think about Loose Coupling
 
PerlMonks  

Loads of Nodes

by AcidHawk (Vicar)
on Jan 07, 2003 at 08:52 UTC ( #224914=monkdiscuss: print w/replies, xml ) Need Help??

I don't pretend to have any of the answers but a few things have occured to me recently.

There have been far more nodes recently that have been simply a request for some-one else to provide code or answer where no attempt has been made to find the solution to the problem. E.G.:Ping from HTML and Easy way to search files? I believe that these constitute poor nodes and there are increasingly more of them around.

Personally I prefer to see that fellow monks have made an attempt even if it is a small one, ie Reg Expression on file name , This I believe causes far more constructive replies to be posted as can be seen from Re: Reg Expression on file name.

As a suggestion for Perl Monks, could we not do a similar thing as the postgres web site where you are not able to post a question if you have not first searched the site for key words that appear in your title? Done via cookies I think..?

Could we not implement some kind of tracking to test if the monk has indeed used Super Search before posting a question to SOPW that has been previously asked or asking for code that already exists..?

It might even be an idea to associate XP points for the amount of use a user gets out of the Super Search engine. Some kind of reward system to encourage the use of the Search Engine..?

By doing some limiting on the number of superfluous nodes we will get far more reliable search results to our queries.

Only a thought to improve the Monastery.

-----
Of all the things I've lost in my life, its my mind I miss the most.

Replies are listed 'Best First'.
Death of PerlMonks Predicted (Film at 11)Re: Loads of Nodes
by virtualsue (Vicar) on Jan 07, 2003 at 13:17 UTC
    It might even be an idea to associate XP points for the amount of use a user gets out of the Super Search engine. Some kind of reward system to encourage the use of the Search Engine..?

    Your heart is in the right place with this suggestion, but in practical terms I don't think this is such a great idea. What if XP hungry newbies run search after search just to get points? I'd rather they voted their little hearts out instead instead of thrashing the PM servers with pointless database searches. If the tool is good, people will use it without extra encouragement, and I believe that it does in fact get quite a bit of use.

Re: Loads of Nodes
by Basilides (Friar) on Jan 07, 2003 at 10:09 UTC
    Perhaps I'm missing something, but I don't see the problem with a bit of repetition. The monastery is pretty arcane, and I think that's one of the fun things about it. Of course it has the Tutorials section, but other than that it's not like a text book where everything needs to be stated once, precisely and comprehensively.

    If a newbie asks a question that's been asked before, I don't see that it's really a problem: it's not like you have to post an answer.

Re: Loads of Nodes
by adrianh (Chancellor) on Jan 07, 2003 at 10:56 UTC

    Personally, I would be against doing this.

    I've only been using the site for six months or so, but I've not noticed the quality of postings changing significantly.

    I don't think any node is superfluous as long as the person has a genuine question, and that question is being answered.

    For me, one of the primary reasons perlmonks is a nice little site is its focus on communication between people. Adding barriers to posting, especially for novices, is going to cause some people to drop out rather than search or RTFM.

    (wearing my usability hat - I've done user testing on an intranet that added this sort of feature and users did drop out at the stage when they were requested to search before posting).

    While repeated and novice questions may be mildly annoying to some users they're easy to skip over and - more importantly - many people do take the time to answer them in useful ways.

    As for searching, I've not had problems myself. Perhaps add a Super Search option to return results by node reputation?

      Point taken.

      I agree totally with you on the promoting communication bit. I find it really interesting that poeple dropped out when asked to do a bit for themselves.. how hard can simple search be..?

      I find the repitition less irksome than some of the "can you code X for me" ...?

      Just a thought.. ;)

      -----
      Of all the things I've lost in my life, its my mind I miss the most.
        I find it really interesting that poeple dropped out when asked to do a bit for themselves.. how hard can simple search be..?

        It's not totally an issue of difficulty (warning: Adrian is about to wander off-topic and ramble about usability. Anybody interested in perlmonks should leave now :-)

        Some of the issues we found were:

        • Instead of one action (posting a message) you now have two (searching and posting a message). Each one has some finite possibility of a user fouling it up. More actions means that there are more chances for confusion and failure.
        • Since posting questions was more complex people skipped it completely and just used the search engine, or fell back to calling people they thought might be able to answer the question, email, etc. They got an answer (eventually) but, since it occured off-site the information never got onto the intranet.
        • It's hard to force the search and not annoy the user. We looked at two alternatives:
          1. The user composes the whole message and then prompted to search. Some users feels they have invested the time to compose a message - so don't bother checking the search results.
          2. Get them to enter the subject line first, then force the search. Users found entering subject without associated message confusing. People rapidly started writing lists of keywords instead of good subject lines to try and get better search results.
          Think of it this way. The users wants to get an answer to their question with the minimum effort expended. Posting a question will get you an answer. Searching may get you an answer. Writing a question takes a period of time that people can estimate well. Finding information using a search engine takes you an indeterminite amount of time - you decide when you want to give up playing with keyword variations.
        • Users didn't return to ask a question after the search task. They get into looking at their search results and "forget" that they were originally trying to post a question. Even if they fail to find a useful search result they never go back to ask their question!

        I could go on, but I'm sure everbody is bored by now!

        In my experience people tend to over-estimate the utility of the information built up using sites like perlmonks, and under-estimate the utility of easing communication. Weird as it may seem, people actually like to help :-)

      I tend to agree - as the author/maintainer of a few fairly popular modules (mainly because they allow you to connect to MS-SQL server from non-MS platforms via a non-standard/Open Source library package) I get quite a few requests for information. Usually it's about some incompatibility between my code and the library (FreeTDS), and in almost all cases the problem has been solved, and the answer has been given multiple times on various mailing lists or newsgroups.

      I try to always answer the question politely, and also point the user at the tools that would have answered the question directly (www.google.com and groups.google.com, usually).

      That way the user will get his/her answer, and may try a search next time s/he has a problem...

      Michael

Re: Loads of Nodes
by Tanalis (Curate) on Jan 07, 2003 at 09:59 UTC
    I think that I generally agree with your thoughts, although I'm not sure that forcing a search before allowing a post would solve the problem.

    Implementing something like that would potentially have a knock-on effect into other sections of the site: forcing a search before allowing an SoPW post is fair enough, and probably a good idea, but would this not simply push the "Code this for me" problem into other areas of the site? Having to search before posting to, say, Meditations, or Snippets, would be fairly pointless, in my opinion - even if they cover old ground, it's often a good thing (to a point) to revisit ideas and gain new ideas and suggestions.

    Something I would be very interested to see is some sort of "Related Nodes" link that can perform a quick search based on the content of the node being viewed - that could help refine searches immensely (the "That's close, but not exactly what I'm looking for" scenario). I think that something like that may help to cut down the number of duplicate questions that seem to come up time and time again.

    I guess the more extreme solution is to point-blank delete nodes that are nothing more than a code request where no attempt has been made to solve the problem. Again, though, I'm not sure that's the best of ideas either.

    Something tells me that things like this are something we just have to live with (and ignore?), at least for the time being.

    Just some quick thoughts.
    -- Foxcub

      Instead of forcing a search, how about making a strong suggestion to do so for the first five questions that a user submits? This can be as simple as offering a link to SuperSearch for the words in the title of the question, or as prominent as a "Before you submit a question" page with links to the SuperSearch, CPAN, etc. that is displayed before a user can submit a question. Any comments?

        Sounds sensible :-)

Re: Loads of Nodes
by FamousLongAgo (Friar) on Jan 07, 2003 at 16:30 UTC
    Since we're a programming site, it seems good to solve the problem with a hack: why not have the site run a search automatically, based on the content of the post? That way, it could show a list of potentially relevant nodes in the preview page, before the user submits the question, giving him/her a chance to find the answer immediately.

    If you build a term-vector model search engine, running natural language queries like these is easy and relatively fast. I'm working on a HOWTO for doing it in Perl, but I'm sure other monks (especially those who built this site) have at least as much experience.

    I know one of the main obstacles to me using the search is that the site is slow - pageloads take a long time, no matter how fast my connection. Does anyone know where the bottlenecks are, and what we could do to improve the system?

    Anyhow, requiring users to search before posting seems a little too Nazi-ish for this site; it doesn't fit in with Perl culture. Whenever many new people come to a site like this, it is a struggle to acculturate them to the local values (like - don't post a question without showing at least some token effort towards solving it yourself). But that doesn't mean we don't want them coming to the site - it's just a burden of success. Requiring a (slow) search will surely not help there.

    I've been using the site for several months, and I'm only starting to really get the feel of the place. So I try to be forebearing with people who don't know better, especially when they later turn out to be kids, or non-native speakers of English. I still find the number of really good questions (and answers!) makes up for the minor irritation of all the less interesting, write-me-a-program posts.

      Does anyone know where the bottlenecks are, and what we could do to improve the system?
      A lot of the bottlenecks are known, from what I gather - however, there's only so many people with indepth knowledge of the site's innards. Despite the pmdev group's size it seems to be mostly tye, ar0n, vroom (who rarely has any time anymore) and a select few others who have the necessary expertise. And all of them are rather busy running their lives as well. The situation is slowly improving though ("slowly" being the keyword).

      Makeshifts last the longest.

Re: Loads of Nodes
by theorbtwo (Prior) on Jan 07, 2003 at 15:19 UTC

    While I don't think doing a search should be required before every SoPW post, it might be a good idea to require reading of certian nodes, such as how not to ask questions, before one's first SoPW. In any case, I think the best system to get good questions out of people is a combination of the two existing ones. First, downvote bad questions, and upvote good ones. Second, when somebody posts a "code this for me" node, tell them what they've done wrong, in a helpful and non-abusive way. "You might want to look at Some::Module; if you show us some code we'll give you specifc help". If there's some search they should have done that's fairly obvious, link to the search, and not to the results, or at least explain how you got the results.

    "I searched on search.cpan.org for 'ping', and found Net::Ping" is much better then "look at Net::Ping.


    Warning: Unless otherwise stated, code is untested. Do not use without understanding. Code is posted in the hopes it is useful, but without warranty. All copyrights are relinquished into the public domain unless otherwise stated. I am not an angel. I am capable of error, and err on a fairly regular basis. If I made a mistake, please let me know (such as by replying to this node).

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: monkdiscuss [id://224914]
Approved by blokhead
Front-paged by Tanalis
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others exploiting the Monastery: (6)
As of 2020-06-06 15:20 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?
    Do you really want to know if there is extraterrestrial life?



    Results (41 votes). Check out past polls.

    Notices?