404 vs. Hand-holding

neerav

For a long time now, I have known that many people have made atrocious spelling errors when creating links to my site and sub-pages. I kept getting complaints from surfers who followed these links from other sites (my 404 is a php script and has a link back to my home page and to also one to contact me). The site owners (of the bad links) cannot be contacted because cell phones generally do not send referrer info.

Then I decided to customize the error, based on common spelling errors, stupid entries, outdated urls, etc. and display how it might be corrected and a reminder to update bookmarks and/or contact the webmaster of the offending site. But who reads that stuff anyway!?! 😃

So here is my question. What are the pros and cons of using my "customized errors" verses using apache's mod_rewrite to quietly direct the user? I feel the latter only encourages the bad links to persist and continue to be bookmarked by others, but then again, the surfer wasn't at fault when clicking a link.

I can't decide between 6 of one and half-a-dozen of the other. 😕

bpat1434

I don't think the rewrite encourages the bad links; however, it does not discourage it. One thing I have noticed as I surf the web is that as larger sites redesign or move information, they throw up a small page that says "This article/story/page may have moved." and include a search function to find that page.

The rewrite could be useful, then again, it could not. If cell-phones don't send referrers, I'm sure that you could figure out who is visiting your site (i.e. is it a cell phone or not) and then grab the referrers if they're there. That would be if you don't do the mod_rewrite.

If you do do the mod_rewrite, it doesn't encourage the bad links. But it does keep your site from being "down" to users who are sent there from older links. I know that's one thing that throws me back to google to find a suitable replacement for the knowledge I seek when I get a nice page that says "This information has moved" and then nothing to help you out.

I guess it comes down to what the "customer" or users want. You could always put a mod_rewrite on it, and then point them to a page with a frame (at the top of course) where they can go take a survey as to if they'd prefer the 404 or rewrite page.

But hey, that's just my 2 cents....

~Brett

Weedpacket

bpat1434 wrote:
I don't think the rewrite encourages the bad links; however, it does not discourage it. One thing I have noticed as I surf the web is that as larger sites redesign or move information, they throw up a small page that says "This article/story/page may have moved." and include a search function to find that page.

This is the approach the W3C recommends. They also have a bunch of Quality Assurance tips which are worth reading.

As far as spelling errors and suchlike go, if you can't tell what they're after, a 404 is the appropriate response; give them a visible page saying there's a problem with the URL they supplied and if you can, give them a list of what you think they might be after.

If they're old links that people have bookmarked or have copied out from something they've seen written down or something like that, then (a) redirect them (with an appropriate flag, mod_rewrite will actually send a redirect header to tell the client that "no, you want to ask for this URL"), and (b) you've just learned a lesson about URI design 🙂

neerav

I neglected to mention this problem is on a WML site, so all visitors are using a mobile. 😉 I like the survey idea for some other HTML sites. I'm thinking that the frame will state the page wasn't found but something is displayed below, and ask for feedback if it is close enough or not. If not, then off to a search or home page.

The story is very different on the WML side. Only two URI's have changed and those were forced due to an ISP/carrier stubbornly caching a very old version of the page. A rewrite takes care of that. The major problem was spelling:

dls -> dis, Dis, Dls
index.php -> Index.Php, ind, Index, index
.php -> .ph, .pgp, .pip, .hp

The other problem was linking to pages that shouldn't. Well, the IQ level of a vast majority of WAP/WML site owners rates pretty darn low and for some reason they linked directly to an "offline" notice that was displayed when a service was off for a month. Why? 😕 Not a clue! Someone even managed to link directly to my error pages, which of course redirected to the home page! Thanks to the style guides posted by Weedpacket, among learning other things, I have discovered the joy of 410. "The URI is dead! Long live the URI"

As a related aside, people were hotlinking to my content. Solution: referrer, move content out of webspace, AND copy content to and deliver from a randomized directory (e.g. /download/alsdkn229fgks/picture.jpg), cron job to clean up every 5 minutes. Then I discovered hotlinking to my THUMBNAILS (damn those unethical dirtbags)! Solution: cron job, rename thumb directory every 1-6 hours. Now I can add what I consider as "giving the finger" to those sites: 410.

From your ideas and references, I have gone for the middle path: combination of rewrite, 404 and 410. Thank you Brett and Weedpacket.