The Inner Workings of Your Favorite Sites

August 13, 2007 3:04 PM

Ever wonder what's powering your favorite websites? Builtwith promises to give you a clear picture of what software and technologies are behind almost any site. Unfortunately, it's not always very accurate.
posted by IronLizard (20 comments total)

How you confuse anything CF with Drupal is beyond me.
posted by IronLizard at 3:05 PM on August 13, 2007

<snark>How you saw Drupal on that page is beyond me.</snark>

Its detection seems pretty hit and miss, but otherwise it's a neat tool.
posted by chrominance at 3:18 PM on August 13, 2007

So in other words, chrominance, it would be good if it worked. Which it doesn't.
posted by tommasz at 3:27 PM on August 13, 2007 [1 favorite]

Its detection seems pretty hit and miss, but otherwise it's a neat tool.

So its core functionality doesn't work, but otherwise it's great?
posted by blue_beetle at 3:28 PM on August 13, 2007

How you saw Drupal on that page is beyond me.

Well, I'll be damned. It was there earlier. Looks about right now.
posted by IronLizard at 3:30 PM on August 13, 2007

I didn't see Drupal anywhere on the Metafilter page, but it got my own site horribly wrong. I don't recommend it.
posted by katillathehun at 3:31 PM on August 13, 2007

Other sites do list PHP under frameworks, so maybe it was just over-zealous detection.
posted by smackfu at 3:31 PM on August 13, 2007

It's odd, the first time I tried it on MeFi, it couldn't connect. The second time it came up with a very long page of items that were mostly wrong. Now it's showing something entirely different. This thing, it learns?
posted by IronLizard at 3:34 PM on August 13, 2007

Well, when I said "hit and miss" I meant more that it doesn't give as much info as I'd like—for instance, I've got two sites on Apache/mod_python running the Django framework, and it actually manages to figure out Python's running but not Django. I haven't really seen it get anything wrong yet, though there are a lot of non-sequitur elements floating about—for instance, "powered by Subversion"? Um, yeah, okay, I guess.

So the core functionality could use a tuneup for sure, but given my limited understanding of HTTP, I'm not surprised it can't seem to glean more accurate or detailed info than it already does.
posted by chrominance at 3:39 PM on August 13, 2007

(or in other words, nice concept—lacking in execution.)
posted by chrominance at 3:40 PM on August 13, 2007

This is silly. There is a certain amount of information you can glean by looking at HTTP headers, the style and form of URLs, nmap-style host fingerprinting, and so on, but it is entirely advisory and subject to interpretation. The operator of the site is the only person that can authoritatively answer these questions, and if desired can hide those tell-tale signs to make it impossible to determine. Besides, most large sites have back-ends that are not visible to the user in any direct way, and that's where the interesting infrastructure lives.

So this idea of a site that automatically tries to sniff these things is just dumb. And come on, who needs a web site to say "this page uses CSS!" No, what would be actually useful would be a repository of these kind of tell-tale fingerprints, e.g. "wordpress blogs typically have their atom feed URLs like so and their category page URLs like so". Such a repository of web software DNA would actually be useful. This is not.
posted by Rhomboid at 4:38 PM on August 13, 2007 [1 favorite]

It doesn't identify Drupal on any of my Drupal sites. In fact it doesn't even identify Drupal at drupal.org. It does identify Wordpress, though.
posted by mike3k at 6:03 PM on August 13, 2007

It misses that one small site known to me was created with ColdFusion even though the site has "created with ColdFusion" in little teeny type at the bottom of the page. It did, however, reveal that what I thought was an absurdly large and horribly offensive animated GIF was actually an even worse Flash animation.
posted by localroger at 6:23 PM on August 13, 2007

One of the top 5 search strings is "neti pot"
Oh, god, please, no.
posted by unrepentanthippie at 7:03 PM on August 13, 2007

The whole list:
1. metafilter
2. "clip tart" zine
3. neti pot
4. pet food recall
5. birthday present for 9 year old boy

What, no pancakes, vibrating things or politics? These guys have no idea what 's going on here.
posted by IronLizard at 8:30 PM on August 13, 2007

It completely missed that my site is powered by Hamsters 2.0.
posted by davejay at 9:23 PM on August 13, 2007

I'm glad to see that MeFi uses HTML, CSS and Javascript. I was worried for a moment.

'course, back in my day, you were lucky if the intertubes carried your 0's and your 1's through the netterwebs, to the terminalwhatsit, to be gloriously displayed as green on black.

Disclaimer. May not have actually lived "back in my day".
posted by djgh at 9:36 PM on August 13, 2007

Well, now...
This site is either highly customised and does not use program defaults or does not use any technology that is listed at BuiltWith.com.

Yeah, I don't think this thing is terribly useful.
posted by cmonkey at 9:43 PM on August 13, 2007

I checked a site that used zope, and even though the site stated in the response header that zope was the server, builtwith did not report it. So, while I think the idea is interesting, the software is pretty worthless in its current state.
posted by tallpaul at 10:23 PM on August 13, 2007

"BuiltWith.com cannot access sites where a trust relationship is required. Sorry."

Oh, come on. It's just SSL, not a government security clearance.
posted by SteelyDuran at 5:25 AM on August 14, 2007