extended rant about liberal URL interpretation and how there isn’t
enough of it, mostly pertaining to one or two examples at the BBC
If you’re a British person and you’re reading this then the chances are you’ve heard of iCan,
the latest Internet super-project to come out of the BBC. (In case you
haven’t heard: It’s an ambitious project all about helping UK citizens
to find the local services which they need, or helping them start
campaigns to change the services which aren’t working the way that they
should. Information, communication, organisation, digital democracy –
all the kinds of things we wanted the Internet to be good for. And it’s
So Auntie put a load of work and money towards building iCan, and
now it’s put a load more money and work into promoting it. Hoardings,
trailers, even a regular TV show. This is understandable, as the service becomes dramatically more useful as more people use it.
There’s only one little problem. A little big problem.
Here’s how you hear BBC presenters on radio and television announce the URL:
Here’s how the name of the service is written on the website and all the promotional material:
Here’s how the name of the service is written on the site logo:
Bearing all of the above in mind, three questions for you, and more importantly, for the BBC website administrators:
(1) How likely is it that a user will enter this URL: http://www.bbc.co.uk/iCan/ (note the capital C)
(2) What do you think a user will see when s/he goes there?
(3) If, like I do at time of writing, you see a big fat “Page Not Found”, don’t you think this is rather silly?
Thing is, if this was the kind of thing that was easily fixed by a
single email to one of my chums at the Beeb (if you’ve worked in the UK
new media industry at all, you have chums at the Beeb) then I wouldn’t
be posting this. Unfortunately, it’s both way too easy and big and ugly
and hard. It’s easy because you can fix this kind of thing in a few
minutes on Apache with mod_speling.
It’s big and ugly and hard because I’m sure that this argument has
already been had several times in planning meetings with the boffins at
BBC Technology, famous for having almost limitless technical skill,
infamous for having similarly huge egos and stubbornness. Back when
they were situated in Kingswood Warren, there was horror story going around about how a remarkably similar situation had sprung up with the Radio 1 website. A producer at BBC New Media asked if, since the URL was being read out on the radio, whether they could please have /Radio1 in addition to the existing /radio1. KW’s response: “I don’t see why we should have extra work to deal with the stupidity of your users”.
It may be, however, that if enough of a noise is made
externally, then it’s possible to escalate this kind of thing high
enough so that something can be done about it. If, like me, you think
that this kind of idiocy has gone far enough, then please link to this
blog entry from your own, so that with a bit of luck, a good project
with a lot of potential to help people isn’t turning those same people
away purely because they paid attention to the project’s branding.
A couple of extra bits: Firstly, someone suggested that I put up a
wiki page so that people could contribute their other little niggles
about BBCi – that wiki page is here.
Please feel free to attack it, though my preference would be to focus
on the silly little things that would take a couple of hours to fix,
rather than a year.
Having said that, what would be preferable would be if the BBC made
such a discussion possible on its own site. At present, all I can see
is the traditional feedback form black hole of the kind that makes you think nobody’s actually going to read what you write. There are also the BBCi Messageboards, which have eight forums for Dr Who but apparently only one for the whole of the Corporation’s digital output and nobody from the staff appears to be there.
Leaving the BBC alone for a minute, I was wondering who else has
this problem. Who else, for example, starts a product name with a
lower-case i? Oh yes…
http://www.apple.com/iPod/ – Oh dear.
http://www.apple.com/iLife/ – Oops.
http://www.apple.com/iMac/ – This works, though. Presumably the person at the company who cared about this kind of thing was still with them, back then. (Update: Apparently all of the above work – half the time. Try hitting Refresh and see what happens. Oi, Apple! Keep your bloody web servers in sync!) (I forgot to credit Shim for this. It was Shim‘s keen eye that spotted it. Say yes to Shim.)
I’ve always admired Microsoft for being better at this kind of
thing, though they do have an advantage in their OS being
case-insensitive anyway (which in itself leads to problems elsewhere).
My admiration lies in users almost always being able to suffix http://www.microsoft.com/ with a product name and it just working. Try it now. The only ones I had trouble with were:
http://www.microsoft.com/foxpro – because I forgot the inital V
http://www.microsoft.com/vb – but even then, note how the error comes up with suggestions, including the correct one.
I’m very keen on that last one because it’s such an obvious
usability benefit. It reminds me of the tricks we came up during the
design of h2g2. We wanted to keep the URLs short as possible, so we made a rewrite rule that would deal with URLs of the form http://www.h2g2.com/A1234,
with the A indicating an article number. The number included a checksum
digit so that if a user typed it in wrong, the system would do a search
and return the possible articles which the user might have wanted.
Looking back, it was probably misplaced engineering enthusiasm, since a
numeric URL with more than three digits is complex enough to rarely be
typed in from memory if ever, and far more often to be clicked or
copied-and-pasted. (And probably would have better to have put more
work into searching on title strings instead of numbers, too) Of
course, these nice simple URLs got broken when the BBC took it over –
any /Axxx URLs will now dump you on the front page, for want of a simple rewrite rule.
Another update: In the comments below, Tom links to a fab discussion on one of the worst problems with the BBC website, namely that if you leave off the “www” from the front of the URL, then all that you can get is the front page.
I find the discussion fabulous because it accurately displays the kind
of stubbornness and technical handwaving that KW were famous for.
Apparently this problem has existed forever, and (according to a BBC
chum) is only now being fixed because the entire might of the marketing
and branding departments descended. To quote a different chum: “It
defies imagination that we sit here and try really hard to reach the
most people, with the lowest computer or any other literacy and these
fucking technology snobs ruin it. i hate them.”
Also, to David Cantrell: That’s a good point about hard log data. I
don’t currently have access to any, though I’ve just heard from within
the Beeb that an initial grep of the logs seem to show no iCan
accesses at all for the first couple of months of iCan’s existence.
Hmmm. I think I’ll hold off for the full results before desperately
trying to save face, if I may… though, in advance, you have to admit
that basing your whole branding on a particular capitalisation and not
having it work in a URL at launch (before stats come in) is silly, no?
No? As for the points about mod_speling: Yes, performance could be a
problem, and mod_speling is probably too much of a musket for this
particular butterfly anyway. A module or rewrite rule that simply
checked for the existence of a lower-cased version of a URL before
returning a 404 would be a minor hit if at all, and surely worth it for
the user benefit. (But yes, mod_speling is available for Apache 1.3 as