I must have been an elephant in another life, given how much time I seem to spend these days shaking my head over “avoidable stupidity”. Or maybe I’m just becoming a grumpy old man. That might be it – I’m losing the ability of youth to look at a slice of cold pizza and go “yummmm”. These days, I look at it and think “The cheese is hard, the cat sniffed it, I can’t even remember when I ordered it” and chuck it out. Ah but I digress.
This week’s headshaker is the way we seem to be loosing control to the developers, control over things that should not be in the remit of developers. Things like letting some algorithm “identify” the language of web content and adjusting my search results based on that. Who dreamt that up? No idea but I bet he was white, monolingual and only had the faintest notion that apart from English, there’s that thing the people making tacos speak and then maybe the thing the Chinese takeaway people use. Choice of three – easy, if L does not equal English, check for non-Latin. If it’s non-Latin in must be Chinese, if it is, it’s Spanish. At least that’s the way it comes across.
The problem is, dear developer, that there’s a great many languages out there and there’s quite a few which are fairly close to each other. Like Irish and Scottish Gaelic for example. So if you’re decide to automatically identify content by language and modify my search results based on that, then bloody well make sure you get it right! Anything else is just seriously annoying unless you give me the option of manually tweaking it.
Given that it’s not like it’s impossible to teach a computer to figure out the difference (for one, Irish uses acutes, Gaelic graves… the one goes up, the other one down, see?) it also raises the question of exactly whom they’re getting to program this stuff? High school students?
Probably not actually, I suspect they’re all really good at code. But listening to my other half, a business consultant with his very own set of why-oh-why’s, I suspect the problem actually is NOT the ability to do code. It’s lack of guidance at all levels. The way big companies hire folk these days goes something like this:
- Company A identifies an apparent problem. Without making sure they identify the root cause, they call for a Fixer-of-Problem-A. First mistake. You’re granny breaking her ankle may be the apparent problem but without checking, you don’t know if the problem is actually osteoporosis.
- So, having rightly or wrongly identified the problem, these days, a job spec gets sent to an agency. Second mistake. They usually get the wrong person to write the job spec, which means the agency is already at the receiving end of a potential mis-diagnosis and a badly written job spec. I’ve seen some of these… the really bad ones are the equivalent of needing a plumber and calling for someone with a proven track record in “the physical aspects of interior decoration as relates to waste disposal”. Yes. THAT bad.
- So we move onto mistake four. The agency usually adds its own flavour of inane, if not misleading, waffle. Using the plumber again, they add something about needing an end-to-end CV showing more than 20 years of experience in toilet seat lifting in blue-chip companies.
- Because it’s an IT related job everyone on this daisy chain assumes that the fixer and/or overseer of the fixing have to be IT people. Wrong. Fifth mistake. Of course you need IT folk to do the black magic but the overseer of the circus does not have to be one. In fact, I’d go as far as saying that they shouldn’t be one. Developers, when left to their own devices, tend to lose themselves in coding “fun” stuff. A failing I guess we all suffer from in our respective domains but for some reason, we let developers get away with policing themselves. In other words, a herd of sheep needs a sheepdog and a shepherd for guidance and direction, not another sheep. The sheepdog and shepherd should have a track record of having dealt with sheep but they don’t have to be sheep themselves.
I reckon it’s this nauseating daisy-chain of mistakes which blesses us with nonsense like the above. We need the coder to do the fancy stuff which, for example, helps identify the content of web pages. Jolly good, I can see the use of that if well done. But it should not be left to the developers to decide that is what’s needed right now, what it will do, how it gets tested, how it gets implemented and how to make sure the user has the necessary control over it if they need it. For that, we need a shepherd who’s not a sheep. If we manage that, I suspect we’d see fewer Siris, fewer counter-intuitive user interfaces, better language in the interfaces and a way of stopping Google from asking me every two seconds if I want to translate this damn page. No, I’m multilingual, and besides, running Irish machine translation over Gaelic won’t work anyway, dammit!
I’ve just been through a head-scratching exercise and before you suggest anti-dandruff shampoos, it was about access keys. Yes… or was it shortcut keys? Or a hotkey? Or a quick key? Which sums up part of the problem – there’s too damn many of them. Now the basic idea is solid – access keys are keyboard combinations which allow you to instruct your computer to carry out frequently used tasks without having to click through the menu. So far, so good. For examply, on Windows CTRL c has long been the command for copy and CTRL v for paste. Then there’s CTRL z for undo and… errr… yes, to be honest, that’s all I ever use and I use PCs a lot, more than I care to think.
I don’t know who invented the first access key but our friends the consumers-of-too-many-pizzas must have thought this was brilliant. If copy and paste access keys are good, surely there must be other useful ones… like for open, save, close, tools, help, save as, pluck a chicken, pick your nose… and soon the whole program was peppered with the damn things. Not only that one program of course … wherever it has started, it soon spread to the rest and like the thing about electric plugs, everyone used a different name and a different key combination without ever giving a thought to the end user. Was it CTRL j, ALT j, ALTGR j or ALT CTRL SHIFT j? Or ALT OPTION or hang on, that was my DoodleBug program on Windows, I’m now on a Mac in VLC. Should I use the Apple button or Fn?
I bet if you did some research, you’d find that a lot of people only ever use a minute fraction of the available access/shortcut/whatever keys. Anecdotal evidence would suggest that the smart everyday user knows less than half a dozen and that most know none at all. And certainly no one uses them to navigate 5 levels down to the Proxy Settings of their browser. Yes, there have been attempts to streamline them but that again ignored the basic question of “do we need them” or “how many do we need”? In any cases, the attempts have been as successful as moves to standardise electric plugs or convince China that the rule of law is a good thing.
So what does this have to do with localization and my headscratching? Well, unfortunately no one bother either to automate the process. Which means that when you localize software, you have to manually add them. Now if the localization process in general was smarter, then that might sort of work but remember that when localizing software, from Microsoft to LibreOffice, what you essentially get is a table with one language in one column and your translation in another. Certainly no visual context. And usually no info which tells you anything about the scope (as in, which of these appear next to each other). So you’re faced with something like this:
And it’s left to you to figure it out. In the above, your guess is as good as mine whether those all appear in the same menu or in two different ones (perhaps the first 3 in a Fry menu and the last in an Other menu). Oh, and did I mention that they don’t even agree on the symbol? In some, it’s &Fry (which gives the end user the line under the Fry), in others you have to to ~Fry and… oh, you get the idea.
So to a half-baked idea, we’ve added a haphazard localization process. Great. Oh, did I mention the guidelines? The ones which say you should put lines under letters with descenders (something dropping down like gjpqy)? Which is usually fine in English with its 26 letters. But Gaelic only has 18 guys, 16 if I have to cut out letters with descenders and even less if I’m instructed not to use thin letters (like ilf). Do I look like the Brahan Seer? I won’t even start on the difficulties that arise in locating said string when you see a wrong access key on screen in testing.
I did take the time to make sure that the most visible ones don’t overlap in programs like LibreOffice and Firefox. But several layers down, to be honest, I can’t be bothered. So I had to remind myself that the nice person who filed a bug on my behalf with a list of them for some several-layers-down-menu-about-frigging-proxies that they’re not responsible for the general mess that are access keys and not to bite their head off. In the end, I did post a condensed version of my reasoning – the fact they’re mostly pointless and as a result, that I don’t have the time and manpower to fix something which the translator shouldn’t have to fix in the first place.
Honestly, don’t people ever take a step back and think?
Tags used in my posts
- Language Technology
- Minority languages
- Operating System
- Predictive Texting
- Scots Gaelic
- Search engine
- Social networking
- Speech Recognition
- User Testing
- RT @Scientists4EU: It also devastating for the NHS as they buy equipment like MRIs, drugs, devices, etc ... half their purchases come from… 13 minutes ago
- RT @flashboy: Bloke who wrote Article 50 says the idea that Scotland would have to queue to get into the EU is “balls” https://t.co/2BCS6ZQ… 5 hours ago