Join 3,512 readers in helping fund MetaFilter (Hide)


They said it would happen
August 9, 2009 8:54 PM   Subscribe

URL shortening service trim is closing its doors trim has committed to maintaining URL redirects till the end of 2009 sorta previously
posted by device55 (90 comments total) 3 users marked this as a favorite

 
snip
posted by furtive at 9:01 PM on August 9, 2009


Heh. Guess I should have used a different one. :(
posted by ocherdraco at 9:03 PM on August 9, 2009


Good, I am tired of clicking on tr.im links, only to have my browser sputter and die, freeze up, or say "URL not found."
posted by hellojed at 9:04 PM on August 9, 2009


'Fail Banana'?

I think I'm starting to see what the problem is.
posted by box at 9:06 PM on August 9, 2009


Apparently it's difficult to maintain a widely used internet service without a revenue stream to fund support, expansion, and development.
posted by device55 at 9:07 PM on August 9, 2009 [14 favorites]


From their homepage:

"tr.im is now in the process of discontinuing service, effective immediately.

Statistics can no longer be considered reliable, or reliably available going forward.
However, all tr.im links will continue to redirect, and will do so until at least December 31, 2009.
Your tweets with tr.im URLs in them will not be affected.

We regret that it came to this, but all of our efforts to avoid it failed.
No business we approached wanted to purchase tr.im for even a minor amount.

There is no way for us to monetize URL shortening -- users won't pay for it -- and we just can't
justify further development since Twitter has all but annointed bit.ly the market winner.
There is simply no point for us to continue operating tr.im, and pay for its upkeep.

We apologize for the disruption and inconvenience this may cause you."

Is it just me, or is that kinda refreshingly honest?

(Incidentally, I wonder if you couldn't monetize URL shortening by setting up http://pep.si or something (well, if you were Pepsi, I mean). That'd be a proactive way to leverage viral networks and the power of brands to maximize mindshare. Wait, what did I just say?)
posted by box at 9:11 PM on August 9, 2009 [9 favorites]


"But the biggest burden falls on the clicker, the person who follows the links. The extra layer of indirection slows down browsing with additional DNS lookups and server hits. A new and potentially unreliable middleman now sits between the link and its destination. And the long-term archivability of the hyperlink now depends on the health of a third party. The shortener may decide a link is a Terms Of Service violation and delete it. If the shortener accidentally erases a database, forgets to renew its domain, or just disappears, the link will break. If a top-level domain changes its policy on commercial use, the link will break. If the shortener gets hacked, every link becomes a potential phishing attack." - http://joshua.schachter.org/2009/04/on-url-shorteners.html

Maybe having some higher-profile URL shorteners will convince more services to just shorten URLs themselves. Why can't Twitter just have a URL shortener that's they own and maintain? tw.tr? So as long as Twitter exists, so do those shortened URLs. And so users know for sure that they can rely on the links to be functional and non-abusive.

And why don't more news orgs have short links right on the page? And why doesn't WordPress/Blogger/etc do the same right out of the box?

That would all seem to be the longer-term solution to this sort of problem.
posted by chasing at 9:11 PM on August 9, 2009 [5 favorites]


That should read:

"Maybe having some higher-profile URL shorteners fail will convince more services to just shorten URLs themselves"

It's late...
posted by chasing at 9:13 PM on August 9, 2009


There are some Wordpress plugins which provide "in house" URL shortening, so if metafilter was run on wordpress, you could create a short link that might look like: metafilter.com/u/123 or something, and there is talk of an emerging tag ad hoc standard for specifying a short URL reference to content....so I think individual providers are stepping up to the challenge.

However, why Twitter doesn't just incorporate it's own url trimmer is just baffling.

If you're going to adhere to a primitive SMS-based message length standard on the web, URL trimming and media hosting should be on your feature enhancements to-do list.
posted by device55 at 9:17 PM on August 9, 2009


that should be "<LINK> tag ad hoc standard"

tiping is hrd.
posted by device55 at 9:18 PM on August 9, 2009


Why can't Twitter just have a URL shortener that's they own and maintain? tw.tr?

This is a terrific idea, and you'd think that Twitter would have thought about this by now, but since very little Twitter does as a business seems to make a lot of sense, who knows if they'll ever do it.
posted by blucevalo at 9:20 PM on August 9, 2009 [4 favorites]


If you use URL shorteners you are breaking the web, and you are bad, and you should feel bad.

Unless you use Twitter, because then it's more a case of who gives a flying fuck at a rolling doughnut what you're linking to.
posted by stavrosthewonderchicken at 9:20 PM on August 9, 2009 [27 favorites]


I'm still waiting for ICANN to move forward with the "use any TLD you want", because I'm all ready to register f.u as a url-shortening service. http://f.u/2 is up for bids now.
posted by mark242 at 9:24 PM on August 9, 2009 [1 favorite]


Yeah, I basically only use url shorteners for Twitter (often to point to posts I've made on Mefi). I wouldn't use them at all if it weren't for those damn 140 characters.
posted by ocherdraco at 9:25 PM on August 9, 2009


I think this is interesting, because so many people are about "free." The problem with a "hey, this should be free!" (thanks, NetZero) is that the rest of the world is not free, so the money must come from somewhere. This leads to a separation of where the money comes from and the product itself. That can end up in some very, very strange places, and can have some unfortunate impact on what the product is. Turn on the TV for an example.

This is especially painful in Internet services and digital products. In some respects, the insane VC funding and ad banner services of the late nineties on have contributed to the end user market having some unrealistic assumptions. It may be a very long time before this self-corrects, if ever.

Servers, bandwidth, electricity, co-lo space, and even employees cost money. Until all of those things arrive at no cost, can we just change "free" to something else?

For tr.im it is especially appropriate, but I am thinking "decoupled and redirected cash flow" might work for everyone.
posted by adipocere at 9:31 PM on August 9, 2009 [2 favorites]


Huh, I was pretty sure I typed either a one liner or a whole screed about how these URL redirection services are stupid, stupid shit in that last thread.

Limiting SMS messages to 140 characters was dumb.

Limiting an internet-based messaging to the artificially imposed SMS character limit was even dumber, especially since most phones these days let you write however large a message you want to write then automagically break it into 140 character long messages, which would be insanely easy to parse on a server in to messages of any arbitrary length.

Then using a whole flock of URL-shortening services to get around both of the above was even more ridiculously dumberer.

Can we kill off the rest of the URL shortners, now? I mean, right now. Just fucking nuke them already. They break the web. They're annoying as hell. People use stuff like bit.ly to send links to me through email, without any identifying tag, description or information about what in the hell I'm clicking on. (I'm lookin' at you, Dan.) It's all so fucking stupid and annoying it makes me want to scream.

Granted I don't use twitter or facebook and I wouldn't mind if those (heh) services vanished along with the whole shitty ecosystem of URL shorteners, image hosts and other services that have cropped up to service the lack of services provided by the service.

Hell, they even have Twitzer, which (kind of) lets you send more than 140 characters through a... twit. Yeah, it's actually a twit, not a tweet. They don't call it Tweeter, do they?

BRB, polishing chrome plated dentures on the hide of some damn fool on my lawn.
posted by loquacious at 9:34 PM on August 9, 2009 [8 favorites]


They should have gone with a fremium market. Free for everyone but it costs money for people who send me chain mail. Fuckers. Pay for my internet.
posted by allen.spaulding at 9:41 PM on August 9, 2009


There are some Wordpress plugins which provide "in house" URL shortening, so if metafilter was run on wordpress, you could create a short link that might look like: metafilter.com/u/123

Actually, the way mefi treats urls is such that you can produce a short url for a given thread by just lopping off the text after the final trailing slash, so that

http://www.metafilter.com/84010/They-said-it-would-happen

becomes

http://www.metafilter.com/84010/

We don't encourage it for any kind of archival linking because we'd prefer to have one canonical url out there, but if you're ever needing to kluge together a shorter url because of text constraints, that'll do it for you. With the upside of not relying on any third parties or introducing unnecessary obfuscation into the process.
posted by cortex at 9:41 PM on August 9, 2009 [6 favorites]


adipocere, I think that in this case you have this peculiar phenomenon where you have this huge free service (Twitter) that doesn't have a clear means of making any money. Then because of Twitter's popularity you have all these remora like services, like tr.im and Twitpic, that rely on Twitter to survive and don't have clear paths to monetization either.

You have free on free on free...if something trembles at the bottom of the pyramid all these ancillary services go 'bonk'
posted by device55 at 9:42 PM on August 9, 2009


Actually, the way mefi treats urls is such that you can produce a short url for a given thread by just lopping off the text after the final trailing slash

Cool.
posted by device55 at 9:44 PM on August 9, 2009


Actually, the way mefi treats urls is such that you can produce a short url for a given thread by just lopping off the text after the final trailing slash

It would be kinda neat if you guys enabled that behavior for mefi.us, just maybe add a handler that looks for an all numeric string.
posted by signalnine at 9:49 PM on August 9, 2009 [1 favorite]


Why no mention of tinyurl.com?
posted by uncanny hengeman at 9:51 PM on August 9, 2009


http://www.metafilter.com/84010/This-seems-like-it-has-the-potential-for-a-man-in-the-middle-attack
posted by mark242 at 9:52 PM on August 9, 2009 [2 favorites]


http://www.metafilter.com/84010/Granted-that-we-hope-people-don't-abuse-it
posted by cortex at 9:55 PM on August 9, 2009


But, practically speaking, there's nothing someone can do with an altered mefi link other than cause someone to go to a different thread than they think they're going to, which could be obnoxious fun for a prankster who likes wasting someone's time but doesn't create the same threat of obfuscated nastylinking that a generic shortener/redirect service does.
posted by cortex at 9:57 PM on August 9, 2009


> When tinyurl closes shop, I'll post it here too.
posted by device55 at 10:04 PM on August 9, 2009


I now see it was sorta mentioned in the "sorta previously" link up top. So, when will it fail?

How much can tr.im etc. possibly cost? It seems like something that can easily be automated in the room the size of a broom cupboard. Get some old IT geezer to come in once a month, give it a kick, she how she's ticking along.

I use tinyurl to avoid net nanny type filters blocking or altering emails I send, so I still see a handy, legitimate use for them. However, I do like loquacious' points.
posted by uncanny hengeman at 10:07 PM on August 9, 2009


Just as long as my precious Moourl stays intact, all is right with the world. It's interesting though, I'd never seen/heard of tr.im until this post. Maybe that was their problem.
posted by saturnine at 10:17 PM on August 9, 2009


I think there's a legitimate need for trimmed URLs, but like permalinks in days gone by, I think most publishing systems will begin to generate those automatically for sharing purposes.

The use you describe for tinyurl is a form of URL obfuscation - while you might see the use as legitimate - the people running those nanny filters sure don't.

(if I were in charge of some corporate nanny filter, url shorteners would be high on the black list)

How much does it cost? More than free - and if they wanted to grow the site by attracting more users they would have to add features and add bandwidth - the cost goes up.

tr.im didn't appear to be a labor of love, they appeared to be trying to find a business opportunity and discovered there is no money in URL shortening, only cost.
posted by device55 at 10:29 PM on August 9, 2009


I use tinyurl to avoid net nanny type filters blocking or altering emails I send, so I still see a handy, legitimate use for them.I use tinyurl to avoid net nanny type filters blocking or altering emails I send, so I still see a handy, legitimate use for them.

I never thought of this usage.

But it's of limited use if an actual web-based filter triggers on the resolved URL, no?

Anyway, there's probably better ways to solve the filter issue. Seems kind of like using a hammer to drive a screw. (Ok, that's a bad metaphor, certain screws are properly applied using a hammer.)
posted by loquacious at 10:35 PM on August 9, 2009


Limiting SMS messages to 140 characters was dumb.

Maybe you should go back in time to 1985 and warn them and possibly provide them with an alternative method. Remember that you don't have ubiquitous IP networks or even packet switching over GSM to play with. Make sure you don't take up any voice time slots either while you're at it.
posted by Talez at 10:46 PM on August 9, 2009 [15 favorites]


There is no way for us to monetize URL shortening -- users won't pay for it -- and we just can't justify further development since Twitter has all but annointed bit.ly the market winner. -- box

Bit.ly, of course, is funded by some of the same VCs that funded twitter, that is Fried Wilson who blogs at avc.com.

But anyway, URL shortening completely sucks anyway. There's no reason for people to use them other then the fact that Twitter sucks and limits web users to 140 characters. If they had any sense they'd only shorten URLs for users receiving messages on cell phones, and let people link links with normal HTML.

I actually wanted to post something on Twitter a while ago (one of three posts) and it forced me to use a URL shortener (I think it was tinyurl), even though my post plus the URL was less then 140 characters. The URL was only 30 characters long or so, it was to a youtube video.

By the way, all tr.im had to do make some money would be to redirect a portion of it's clicks to pages with ads. It would have been unethical, but they would have made a ton of money. It would have killed their business, but then, so would pulling the plug.

http://www.metafilter.com/84010/This-seems-like-it-has-the-potential-for-a-man-in-the-middle-attack

What are you talking about? What do you think "man in the middle attack" means? It generally has to do with public key cryptography.

I think there's a legitimate need for trimmed URLs, -- device55

No

The use you describe for tinyurl is a form of URL obfuscation - while you might see the use as legitimate - the people running those nanny filters sure don't. -- device55

It doesn't matter, the filtering software will be able to see the real URL before it gets shown to the user. If they're using blocking software to block email based the hyperlinks in it (assuming it doesn't contain links to phishing sites, in which case it's almost certainly spam), they're doing it wrong.
posted by delmoi at 10:58 PM on August 9, 2009 [1 favorite]


160 (not 140) character SMS messages make sense, since they can work over analog networks. Of course, they don't actually get sent on analog networks if they don't need to be. The 140 character limit actually works well for the format, it brings out a lot of creativity and people feel more comfortable posting quick notes if it's not possible to send long ones. But that's no reason not to allow people to include ordinary hyperlinks.
posted by delmoi at 11:03 PM on August 9, 2009 [1 favorite]


Limiting SMS messages to 140 characters was dumb.

They didn't just decide that it should be limited at 140. The restriction originated with the fact that SMS was piggybacking unused bytes in the system. To quote wikipedia:
The key idea for SMS was to use this telephony-optimized system and to transport messages on the signaling paths needed to control the telephony traffic during time periods when no signaling traffic existed. In this way unused resources in the system could be used to transport messages without additional cost. However, it was necessary to limit the length of the messages to 128 bytes (later improved to 140 bytes, or 160 7-bit characters), so that the messages could fit into the existing signaling formats. Therefore the service was named “Short Message Service”.
It's as if someone figured out a way to build and market homes that fit entirely within the grassy 2 foot wide median between the two sides of highways. Years later after the concept took off, when everyone is accustomed to cramming their asses into impossibly narrow homes, someone wonders why they didn't just make these homes wider.

(And this also hints as to the criminally obscene fact that they charge you 20 cents to receive a message over this bandwidth that is not otherwise being used.)
posted by Rhomboid at 11:05 PM on August 9, 2009 [7 favorites]


The use you describe for tinyurl is a form of URL obfuscation - while you might see the use as legitimate - the people running those nanny filters sure don't.

Exactly! That's what makes it so laughable. I can employ a "trick" like that and beat 99.9% of companies that try to filter such links. I must be a |33+ h4Xx0rz and not even realise it.

[we are talking Luddite Australia here]
posted by uncanny hengeman at 11:23 PM on August 9, 2009


"I actually wanted to post something on Twitter a while ago (one of three posts) and it forced me to use a URL shortener (I think it was tinyurl), even though my post plus the URL was less then 140 characters."

Yeah, this drives me absolutely crazy, I can't think of any good reason to shorten URLs that fit and it'd surely be incredibly simple to fix.
posted by malevolent at 11:28 PM on August 9, 2009


I think there's a legitimate need for trimmed URLs, -- device55

No
To be clear, I meant short URLs generated by the content author or provider, not by third parties. For example, I could have a "share this story" link on my blog which provides a short canonical link, while the web site itself used full hierarchical URLs. The same short URL could be used for those ubiquitous "share on foo" badges everyone seems to like.

I think third party URL shortening services are, at their very best, a pain in the ass
posted by device55 at 11:31 PM on August 9, 2009


[we are talking Luddite Australia here]

For example, the other day our CFO sent out an email whining about 1MB attachments, Maude Flanders style. "Won't somebody think of the bandwidth!"
posted by uncanny hengeman at 11:31 PM on August 9, 2009


For example, the other day our CFO sent out an email whining about 1MB attachments, Maude Flanders style. "Won't somebody think of the bandwidth!"

To be fair that was Helen Lovejoy not Maude Flanders.
posted by Talez at 11:52 PM on August 9, 2009 [1 favorite]


Well, I imagine the guys can at least get some money for the tr.im URL, from either the weight -loss industry or, ahem, the porn world.

Now there's a brilliant idea...name your service something you can sell to a porn site in case your business fails, as a sort of last-ditch insurance policy. Of course, explaining it to a VC might be difficult:

"I'm intrigued by your venture, but after reading your business plan I have a question..."
"Our business model is simple: we leverage the memetic synergy inherent in social media applica-"
"No, no...I understand that. My question is: why do you call your service Nasty Ass Virgins?"
posted by Ian A.T. at 12:15 AM on August 10, 2009 [5 favorites]


They should have gone with a fremium market.

I don't think desert-dwelling nomads can support much of a market, unless it is for water and wormteeth.
posted by Dr Dracator at 12:20 AM on August 10, 2009 [3 favorites]


Why doesn't twitter simply shorten it internally? If they displayed the url as a hyperlink then the address shouldn't count towards the character limit. The source address then is a property of the hyperlink; it's metadata.
posted by cotterpin at 2:27 AM on August 10, 2009


I thought the canonical urls looked like
http://www.metafilter.com/mefi/84010

I want to see an archive.org or Coral Cache pop up for the shorteners. I can't think of a decent name/domain for it, though.
posted by Pronoiac at 2:28 AM on August 10, 2009


I'm always amazed at people who complain about being continuously ripped off. Like SMS users. YOU ARE ALL IDIOTS.

(Insta-reply: I'm always amazed at people who complain at complainers. YOUR EFFORTS ARE FRUITLESS.)
posted by Civil_Disobedient at 2:59 AM on August 10, 2009


A legitimate use for URL shorteners: twelve foot long URLs generated by Google Maps (or eBay or car/real estate sites). These can still break in transit via email and whatnot.

Of course those sites should provide sane links but as long as they don't a URL shortener can come in useful.

Guess I should consider rejuvenating my own tinyurl-clone. At least it's my decision when my links fail.
posted by _Lasar at 3:47 AM on August 10, 2009 [1 favorite]


Luckily, this is still available.
posted by Obscure Reference at 5:32 AM on August 10, 2009 [1 favorite]


It's always amusing to see the "use cases I don't understand or encounter are stupid" set appear in a thread. Blogs? Evil. Twitter? Evil. IRC? URL shorteners? Vegetarian hot dogs? The HTML tags I don't use? Your favorite band? All evil.
posted by verb at 5:44 AM on August 10, 2009 [6 favorites]


(And this also hints as to the criminally obscene fact that they charge you 20 cents to receive a message over this bandwidth that is not otherwise being used.)

Well, the infrastructure used to route all those messages to their appropriate destinations isn't just some random hardware that would otherwise not be used...
posted by effbot at 5:50 AM on August 10, 2009


I've just used tinyurl, for years. Perhaps it will suffer the same fate.
I only use it for ridiculously long URLs though and it's demise will not
affect me much.
posted by rmmcclay at 5:57 AM on August 10, 2009


it forced me to use a URL shortener (I think it was tinyurl), even though my post plus the URL was less then 140 characters.

The fact that Twitter even *contemplates* doing this is just so fucked up. I came in under your stupid limit. Now you're going to force me to rely on a URL shortener I didn't need? Twitter is run by extremely stupid people.
posted by mediareport at 6:13 AM on August 10, 2009 [4 favorites]


Last month I emailed the guy behind is.gd about his provisions for future archiving (which I started using because it gave me an extra character let me copy-c right after generating the url, which I like better than bit.ly, which makes you use the mouse to click "copy"). He was very nice and responded right away, but his answer was as exactly reassuring as the reassurances on the FAQ pages of the bigger companies - that is, not reassuring at all:

I certainly intend to keep the site running for the foreseeable future but if for any reason it did get shut down completely, I'd be likely to make a version of the database (minus any identifying details such as IP addresses of link creators) available if a 3rd party was interested in running it. Whether anyone would actually be interested in doing this is another matter!

In other words: "YOU ARE STUPID FOR RELYING ON THESE SERVICES FOR ANYTHING LONGER THAN A FEW WEEKS!"
posted by mediareport at 6:29 AM on August 10, 2009


it gave me an extra character and let me copy-c right after generating the url
posted by mediareport at 6:30 AM on August 10, 2009


The reason we need link shorteners is because so many sites use such bad URLs.

Unclutterer just launched this service to let users create intuitive, humane, and functional links to Amazon on the fly without stopping to find the correct link.
posted by pjdoland at 6:32 AM on August 10, 2009


I use tinyurl a lot, thanks to a Firefox plugin that lets me create them in one click (straight to clipboard). I should stop relying on it and back everything up.
posted by WPW at 6:40 AM on August 10, 2009


Pretend I am dense: could someone explain to me why --other than Twitter with its character limit-- the length of urls matters when linking to something?
posted by Orb at 6:56 AM on August 10, 2009


I thought the canonical urls looked like
http://www.metafilter.com/mefi/84010


No, that's the old url structure; it was deprecated a while back. We still support that format for legacy reasons, but the site never produces a url in that format on purpose.
posted by cortex at 7:14 AM on August 10, 2009


Pretend I am dense: could someone explain to me why --other than Twitter with its character limit-- the length of urls matters when linking to something?

Long urls can be a pain in any context where either the total text volume is constrained (Twitter/SMS being the standard example at this point) or where long urls may "break" for formatting reasons (email clients that fail to autolink a url if its wrapped, or which don't allow correct copy-and-paste of wrapped url text; IRC clients that don't handle wrapped urls correctly).

As a response to that, url shorteners are understandable, though (and I'd be curious to see counter evidence to this) my impression is that nothing before twitter had created anything like the explosion of apparent demand (and evident supply) of shortners we have now.
posted by cortex at 7:20 AM on August 10, 2009


Pretend I am dense: could someone explain to me why --other than Twitter with its character limit-- the length of urls matters when linking to something?

Long URLs used to break in text-only email clients if they stretched onto a second line. The carriage-return/line-feed would cut off the end of the URL, creating a link that wouldn't work if you tried to follow it. You would have to copy-and-paste the separate parts into your browser's location bar by hand. This still happens sometimes in a reply or forward that breaks lines and prepends a > character, for instance.
posted by stopgap at 7:22 AM on August 10, 2009 [1 favorite]


Pretend I am dense: could someone explain to me why --other than Twitter with its character limit-- the length of urls matters when linking to something?

That's the only reason it matters. The people who came up with HTML cleverly decided that long URLs should be able to be shortened. So, if I wanted to show you a google map I wouldn't need to actually display the full URL on your HTML-capable device:

http://maps.google.com/maps?f=q&source=s_q&hl=en&geocode=&q=new+york,+ny&sll=37.0625,-95.677068&sspn=49.444078,79.013672&ie=UTF8&z=11&iwloc=A

...since I can conveniently use the anchor tag and embed the URL behind a few words. I can also emphasize words without saying "okay guys, the next two words should be emphasized as you read them." I can let your HTML display device take care of the metadata that I included to actually apply visual emphasis!

It's like they were geniuses, or something.

Only, they didn't take into account the endless bastardization of the internet that would occur only too soon afterwards, or the use of out-of-band data channels to transmit text messages, and the further attempts to cram HTML into such messages.

So, yes, the only reason you'd need to shorten an actual URL is to place said URL into channels that have arbitrary or technical limits.
posted by odinsdream at 7:48 AM on August 10, 2009


I feel bad for the tr.im guys; their shutdown blog post conveys a lot of frustration. I'm not wild about URL shorteners either, but they provided a decent service for free. And they're shutting it down responsibly; redirecting URLs for months is cool. The only thing better is if they'd give a dump of all the redirects to someone to host in a permanent archive.

And while I dislike redirectors for all the reasons it breaks the web, I have to say bit.ly is pretty great. The stats are great, it's clever how they create an "account" for you before you ever bother with a username/password, and the redirection has been very reliable. I wish there were some way to do clickthrough tracking across the Web that didn't break the link destination.
posted by Nelson at 8:21 AM on August 10, 2009


Maybe you should go back in time to 1985...

Why does time travel always seem to involve the 1980s?

"Okay, we've invented a time machine, but for some reason it will only take us within a ten year period in the late 20th century."

"Destroy it and then torch the building."
posted by ODiV at 8:31 AM on August 10, 2009 [1 favorite]


I've thought about bit.ly and Twitter, and the only reason I can figure for why they make Twitter behave like it does, is because they think the bit.ly data has market value. That would explain why they force so many links to go through it, even when there's no reason to do so (even with the stupid 140-character limit).

It wouldn't be hard — a few lines of code, really — to only shorten URLs when they break the 140-character boundary. And it wouldn't be that much harder, if they had thought about it when they were designing the system, to never shorten URLs except when they're going out via the SMS gateway to devices where the 140-char limit actually matters. In other words, you could always see full URLs on your computer, but get bit.ly or tinyurls in SMS messages because of the pre-existing hard limit. You could even (since a bit.ly link is always a known length) ensure that a message never exceeded the 140-char limit without actually shortening it, by recognizing a URL (which the system does anyway) and counting its "shortenable" length for SMS purposes.

But they don't do any of that; they seem to prefer everything gets shortened instead, despite the obvious downsides of shortening. Given that it's bad practice, they must have a compelling reason to do this, and the only one I can imagine is that they are datamining the resulting statistics, and think there's some financial value in it.

Unfortunately if I'm right and that's the case, the practice is going to continue until it becomes a serious problem that overwhelms the profit motive.
posted by Kadin2048 at 8:55 AM on August 10, 2009


Pretend I am dense: could someone explain to me why --other than Twitter with its character limit-- the length of urls matters when linking to something?

I first heard about them about five years ago, in the tech section of a print newspaper. In that context, space and ease of transcribing would obviously be more critical.
posted by rollick at 8:57 AM on August 10, 2009 [1 favorite]


Thanks for all the explanations. I hadn't personally run into any instance where a long url caused me a problem of any sort, so was a little clueless.

Still don't like these services though, because if I can't see where something is going to take me, I am not likely to click on the link.
posted by Orb at 9:53 AM on August 10, 2009


People use URL shorteners as permanent links? Until now I actually thought that the URLs were recycled after a few weeks/months.

Anyway, LongURL Mobile Expander (greasemonkey) makes browsing twitter (and any other page that uses URL shorteners) a lot less annoying.
posted by Memo at 9:59 AM on August 10, 2009


cortex: > I thought the canonical urls looked like http://www.metafilter.com/mefi/84010 ?

No, that's the old url structure; it was deprecated a while back. We still support that format for legacy reasons, but the site never produces a url in that format on purpose.


Eh, mathowie preferred them, & they still show up in the contact activity on the front page. I could see Mobile Safari's losing #anchors on redirects, breaking, say, this link to to the same comment as being problematic, though.
posted by Pronoiac at 10:44 AM on August 10, 2009


box: "That'd be a proactive way to leverage viral networks and the power of brands to maximize mindshare. Wait, what did I just say?)"

I don't know what you said but I know where you've been.
posted by KevinSkomsvold at 11:01 AM on August 10, 2009


The people who came up with HTML cleverly decided that long URLs should be able to be shortened. [explanation of the anchor tag]

Right. Except that the most common form of person-to-person communication on the internet--email--is a text-based, not HTML, protocol. Trying to cram HTML into email creates all sorts of additional problems.
posted by mr_roboto at 11:10 AM on August 10, 2009


every time I try to think about twitter, my head hurts. mind you, i'm not a twitter hater. i have my own account and everything, and some of my favorite mefites routinely fill my phone with neat little gems of cool stuff every day. some of them do so many times a day because they are obsessive, though still cool as hell. so i like twitter.

but i can't think of any way for it to survive. not. a. one. I mean, here's one direction my thinking sometimes goes. We're looking at what is arguably the smallest medium for mass communication. It's designed for easy readability via mobile platforms, using incredibly condensed no frills delivery limitations such that no code is possible to transmit, so links are actually strings of text. It has a hard character limit to facilitate this low bandwidth. So everything about it says "low scale low scale low scale." on the plus side, this has created a remarkably mobile communications medium using virtually no resources for the user. yay!

except it doesn't work. we're operating with a web that seems to be moving increasingly toward natural language links. as a point of reference, we can see how our own metafilter thread links have added the title text to the url in the interest of giving the user some small idea of what they're about to click on. in other words, we're expanding our urls for readability at the expense of url size because, outside of twitter, the economy of links is such that, as odinsdream pointed out, url length is essentially negligible. the data required to say to somebody "check this web site out, it is located here" is so small by internet standards that we can arbitrarily increase it for the sake of adding more info because we feel like it, and it won't negatively impact anybody.

except on twitter. now twitter's low scale approach has become mired by the web's natural progression. so url shorteners come to the rescue. I feel like url shortening services have existed before twitter, yes? I always considered them to be highly suspect because, anywhere but twitter, they only served to mask whatever you were clicking on. either way, now twitter is their business. they cosmetically undo the progress of natural language in our web links, obscuring urls so much that it is impossible to know what they link to from looking at them. ironically, these services that are supposed to facilitate using the semantic web on twitter obliterate the semantic parts of the web. further irony: these services that are supposed to make twitter's character limit more manageable for linking in a tweet now necessitate adding characters to a tweet to explain what's behind the shortened url, thus making the character limit once again a problem in some tweets. but this requires bandwidth. a tiny amount, sure, but nevertheless. and there's no revenue to be made by running this service. why hasn't twitter developed one of their own? because they already have one popular service suffering from no feasible revenue model, thankyouverymuch. so now twitter is not only suffering from bandwidth problems, and service outages, and a colossal drainage of funds with no reasonable way out, but it's spawning satellite businesses that area also hemorrhaging bandwidth and funds and going out of business. It's like a firecracker, consuming itself and sending out other little exploding bits of self-consuming business as it goes. and why? because now instead of just being packets of 140 characters, most tweets are now redirects, consuming bandwidth at twice the pace from two different sources.

which wouldn't be a big deal if either source could figure out a way to make money. but how can they? the system is already set up to be scaled down burst of tiny info to 3rd party applications on other platforms. twitter is paying the considerable bill for whatever small amounts of money apps like tweetdeck and twittelator pro are able to make by being the mobile client app for parsing twitter data.

no matter what I think of, I can't see any way for this system to stay alive.
posted by shmegegge at 11:24 AM on August 10, 2009 [3 favorites]


Eh, mathowie preferred them, & they still show up in the contact activity on the front page.

Granted on the preference (I'd forgotten his argument there on the redirect-to-canonical issue, which is a good point), but my understanding is that any part of the site that actually generates that style of url is doing so only as forgotten cruft at this point.

I'm not sure where you're seeing them in the Contact Activity sidebar—I'm seeing only new-style links—but if you can let me know in more detail I'll check it out with pb.
posted by cortex at 12:08 PM on August 10, 2009


They're on the front page sidebar, for new posts & questions from contacts; not on the Recent Activity from Contacts page.
posted by Pronoiac at 2:37 PM on August 10, 2009


Huh, i had never used trim. I opt for cli.gs.
posted by TonyDanza at 2:45 PM on August 10, 2009


Where? I see posts with the metafilter.com/xxxxx/ format url but none with the old metafilter.com/mefi/xxxxx style.
posted by cortex at 2:49 PM on August 10, 2009


I was not a fan of tr.im. In the last few months, most of its URLs that I tried failed to load; not sure why, but it was a real pain. In fact, just yesterday, spookily enough, I discovered a Post-it on my desk from July 2. I think I'd meant to write a blog post about it. The message was short and to the point:

"tr.im never works"

As cool as the tr.im guys seem to be, their product just flat-out didn't work half the time. Forget monetization; their issue was basic functionality.
posted by limeonaire at 4:12 PM on August 10, 2009


*points at monitor*

Oh, hang on - *adds contact* - okay. I now have "ocherdraco posted Barbara Ehrenreich on Poverty... to MetaFilter. an hour ago" near the top of my sidebar.

posted by Pronoiac at 4:18 PM on August 10, 2009


Er. I mean, you should have that listed in your sidebar, too.
posted by Pronoiac at 4:21 PM on August 10, 2009


Of course, if you are going to use a URL shorterner, Make 'em really short. I mean, what's cooler then a URL like http://➡.ws/㨥

(your browser will automatically un-unicode those when you click on them, but they only take up one actual Unicode character)
posted by delmoi at 4:21 PM on August 10, 2009


Oh ho! Yeah, that looks like it needs fixin'. I'll tell The Bausch.
posted by cortex at 4:26 PM on August 10, 2009


I could go for some trim right about now. MeFiMail me, ladies!
posted by Eideteker at 5:44 PM on August 10, 2009


twitter is paying the considerable bill for whatever small amounts of money apps like tweetdeck and twittelator pro are able to make by being the mobile client app for parsing twitter data.

Going a bit off topic, but I find it hard to fathom (like most thing about Twitter's business model) why Twitter didn't start making and selling their own mobile client apps first. Make them really good and competitive, and cheap, but not free. Looking at my iPhone, I've got a Skype app (made by Skype), a Shoutcast app (made by Shoutcast), a Facebook app (made by Facebook), an Orb app (made by Orb), an OurStage app (made by OurStage), a FourSquare app (made by FourSquare, although completely useless to me since they've decided to make their service geographically exclusive), a SomaFM app (made by SomaFM) as well as apps made by half-a-dozen other obscure web sites, some of them free (with ads), some of them paid for.

And then I've got Twitterific - not made by Twitter.
posted by Jimbob at 9:13 PM on August 10, 2009 [2 favorites]


Going a bit off topic, but I find it hard to fathom (like most thing about Twitter's business model) why Twitter didn't start making and selling their own mobile client apps first.

1) They're lazy

2) They want to get a lot of third party apps created and create an ecosystem of developers who are invested in the platform.

I think #2 is their idea, but it does make them seem pretty lazy. I don't think their business model is that bad. The bandwidth used by the system must be minuscule. I mean, 1 minute of video on youtube is equal to 16 thousand tweets (assuming 300kbps vs. 140 bytes/tweet)
posted by delmoi at 6:08 AM on August 11, 2009


delmoi: "The bandwidth used by the system must be minuscule. I mean, 1 minute of video on youtube is equal to 16 thousand tweets (assuming 300kbps vs. 140 bytes/tweet)"

well, here's a quote from wikipedia:
Although estimates of the number of daily users vary because the company does not release the number of active accounts, a February 2009 Compete.com blog entry ranked Twitter as the third most used social network[5] based on their count of 6 million unique monthly visitors and 55 million monthly visits.
going strictly from that and the experience of constant recurring downtime from the twitter server, I'd say the bandwidth bill is more than miniscule. I believe it's miniscule per tweet, obviously, but without a revenue model I suspect the cost of hundreds of millions (if not larger) tweets a day is taking its toll. assuming Ashton Kutcher sends 3 tweets a day (I have no idea how many he sends) he is, by himself, creating 3 million+ tweets worth of data for twitter to send out daily.
posted by shmegegge at 7:41 AM on August 11, 2009


There's a lot of overhead involved in a system like Twitter. It's not like they are just blasting 140-byte messages back and forth. To send a tweet, you need to establish an HTTP connection and send a POST request, and what you get back is typically JSON or XML. I haven't actually measured it myself, but I suspect that there's at least several thousand bytes of overhead for each 140b message.

However, I don't think that's where Twitter's scalability problems come in. They probably do have a surprising amount of traffic for a service that handles such tiny messages, but it's still small compared to Facebook or WoW or probably even a lot of high-volume dynamic web sites. The solution to bandwidth problems is easy: you buy more. I don't think that's their issue (except when being DDoSed).

My (admittedly limited) understanding is that their problems were and are less related to actual client/server bandwidth, than in the backend architecture. The whole service was built in Ruby on Rails, which while a great rapid application development/prototyping platform, is not exactly a great choice for high-volume OLTP. They have issues, like handling lots of incoming submits/writes and maintaining concurrency, that are different from the typical LAMP-backed dynamic website, and seemingly more akin to issues you'd run into with financial systems.

Of course that's all clear in retrospect; I don't think Jack Dorsey (who by all accounts that I've ever read, is an incredibly bright guy and has been working on presence-notification stuff since 2000 or so) or anyone else can really be blamed for using the tools and design paradigm that they were familiar with when they initially threw Twitter together. It's just an interesting example of how what appears to be a straightforward choice early on in a product's lifecycle can be very constraining later on.
posted by Kadin2048 at 9:12 AM on August 11, 2009


I've thought about bit.ly and Twitter, and the only reason I can figure for why they make Twitter behave like it does, is because they think the bit.ly data has market value.

Interesting observation. The end-to-end principle makes it pretty hard to track user behavior on the web. Alexa relies on people to install a plugin to track their behavior; Google uses an inhouse redirection service for their search results. Some people say that Google's large scale gives them advantages in all sorts of related fields, like local search.

Somehow, Twitter and Bit.ly have come along with both a problem (message size constraints), and a solution that has the potential to track a portion of activity on the web in a central location. If you think about it for too long, you might imagine that if Twitter isn't a covert domestic spying tool, it's a tempting target with already cozy relations to the state department.
posted by pwnguin at 12:54 PM on August 11, 2009


if Twitter isn't a covert domestic spying tool, it's a tempting target with already cozy relations to the state department.

Maybe; although it seems like it's more of a marketer's wet dream than an intelligence service's. Being able to track the success of your latest attempt at social media astroturfing or viral advertising in real time would be worth a lot of money.

Google seems more like the one that the three-letter agencies would love to get in bed with. I've often wondered if the threats of anti-trust action (which have always struck me as out of place; Apple would be a better target if monopoly-busting was really the goal) are really the government's way of making bedroom eyes. In reality it's probably not the case — a scheme like that would require far more coordination and foresight than I've ever seen demonstrated by the government — but I still wonder if an anti-trust problem down the road could be made to 'disappear' with the right sort of cooperation.
posted by Kadin2048 at 1:11 PM on August 11, 2009


Never mind!
posted by Jaltcoh at 5:38 PM on August 11, 2009


That contact activity sidebar non-canonical url thing is fixed, btw. Go pb!
posted by cortex at 5:41 PM on August 11, 2009


Who is going to trust the service now? (Though, I suppose the real moral should be 'who is going to trust any of these services.)

TinyURL existed before Twitter came along. It's not like these services didn't have a purpose before Twitter. Linking to things outside of the context of an HTML page is kind of a suck. (Try sending a Google Maps link in an email. Awesome!)
posted by chunking express at 7:05 PM on August 11, 2009


I miss the non-canonical addresses already.
posted by Pronoiac at 7:51 PM on August 11, 2009


And now it's going open-source, and is going to be community owned and run. Also, that dude really hates bit.ly, though I can see why.
posted by chunking express at 11:43 AM on August 17, 2009


« Older Back of the Mike | Around the Corner | Behind the ...  |  In 1999 and 2000, and again fr... Newer »


This thread has been archived and is closed to new comments