New orphaned threads system

[Mike]

As promised in the Young Composers Roadmap, an orphaned threads system has now been introduced to the forum setup.

The system aims to help members track down music submitted to the upload forums that does not appear to have received due attention or comment. In particular, it will probably be of use to the official reviewers, who can access the page and tell at a glance what works might be worth their time reviewing.

As of now, three types of statistics are provided: threads least viewed in relation to start date, threads with least reply material and threads with most contributing authors. More details on these three flavours are available on the orphaned threads page itself.

The processing required of the server to generate the statistics is too intensive to set in motion every single time the orphaned threads page is accessed, so all data is updated automatically every 15 minutes with a backend script. When you access the page, you will merely see the last collection of stats generated by the automated script: the rankings will not be absolutely up-to-date (though the thread information such as views, last post etc. will be). The system also limits its scope to prevent it showing threads that (according to the "new threads per day" statistics) seem likely to have been started less than one day prior to the current day. Right now, it probably returns a little too many results, but this will probably be changed in future.

Some extra functionality will be added in future (most likely improvements to the methods used to generate the stats), but the system is sufficiently complete as to be of worthwhile use. And by the way, it's supposed to update every 15 minutes, but if it doesn't, it's probably just a little hiccup on the server, so don't worry.

Comments are welcome.

[Will Kirk]

Good idea IMO

[bob stole my cookie]

*bows to the master*

[CaltechViolist]

Somehow, I think there will be threads that are missed by all three. I'm too busy right now to give a full analysis, but off the top of my head I would suggest that a thread getting many short replies isn't being reviewed adequately - as is a thread that is getting many views but minimal replies relative to the number of views.

I was more than slightly annoyed when something I spent 3 years writing took fully 36 days to get its first substantial review, and yet it got a ton of very short replies before that, mostly quibbling about rules and the like, so it never would have appeared on any of the Orphaned Threads lists.

[Mike]

Somehow, I think there will be threads that are missed by all three. I'm too busy right now to give a full analysis, but off the top of my head I would suggest that a thread getting many short replies isn't being reviewed adequately - as is a thread that is getting many views but minimal replies relative to the number of views.

That's why it measures the total number of characters of all the replies in a thread (minus those of posts made by the thread starter), not just the number of replies. One thing I am looking to implement, however, is having the script examine individual posts in a little more detail, as opposed to simply lumping them all together with a character sum or whatever. Trouble with this is, it's both more demanding from a performance point of view and will take more effort to code (if it's to be done efficiently enough as to not render it unusable, that is). I'm kind of burnt out from my initial effort right now, but at some point I will sit down and try to continue developing the script.

I was more than slightly annoyed when something I spent 3 years writing took fully 36 days to get its first substantial review, and yet it got a ton of very short replies before that, mostly quibbling about rules and the like, so it never would have appeared on any of the Orphaned Threads lists.

Well, first of all, considering the system limits its scope from "yesterday" (what seems likely to be yesterday; for this, it uses the average number of posts per day over the last three days) to three days before "yesterday", a thread from 36 days ago would be nowhere to be seen. I was thinking of perhaps adding a "long range" statistic that looked at threads maybe 15-30 days old though.

Second, you may well have appeared somewhere high up on the "Threads with most contributing authors" list, even if your thread had been viewed a lot and it had plenty of reply material.

In the end, however, the best judge of whether a piece has been sufficiently removed is a human, not a PHP script that uses an mathematical formula to produce a numerical list. There is no way to measure with code the quality of a review. That said, one-click access to threads which seem likely to have not received much feedback certainly can't hurt.

And yes, advice from anyone with a mathematical mind, in particular statisticians, would help tremendously.

[Hands]

This is not really on topic, but I don't think the issue deserves its own thread. On the very last page of the Free-for-All, there are a bunch of locked, empty, authorless topics. I figure they must be bugs.

The Mike, he has been informed.

[CaltechViolist]

Probably a deleted member. In fact, I'm pretty sure I know exactly which deleted member.

Nonne was stalking me for months, after all... sending me e-mails at all of my e-mail addresses (I didn't even know some of them were public?!), commenting profusely on my blog, etc, etc.

[montpellier]

I suppose how important it is, depends on how orphan it happens.....

it might help us cotton on to threads that weave missed, however...

[Mike]

This is not really on topic, but I don't think the issue deserves its own thread. On the very last page of the Free-for-All, there are a bunch of locked, empty, authorless topics. I figure they must be bugs.

The Mike, he has been informed.

My guess is that they're threads that were moved from the FFA with a redirect, and that the IPB to vB converter mangled them up slightly. It may be safe just to delete them. I'll look into it.

[montpellier]

(from the second section on the orphans list)"Thus, posters towards the top of the list have likely contributed significantly to the community, which might mean that other members should repay them in turn by commenting on their work."

I'm not sure what this means. Are we talking about the "thread/thread starter" or latest poster or...? Only curious - does it mean: whichever poster is referred to has contributed the most reviews or typed the most while reviewing?

I wasn't sure what this was for. I remember just after joining, discussion was invited about asking newcomers to review a few items before being allowed to post music but this is clearly different. Did anything come of that discussion?

cheers

Montpellier.

[Mike]

It refers to the thread author, the user who created the thread, and simply measures the total number of characters they have posted in reply to threads in the upload forums. At present, it can't distinguish between "Very good, I enjoyed it" and "Your link is broken", however we are looking into making it a little smarter in this regard.

No new policites or anything like that came out of the discussion regarding new members, as it happens.