WEBVTT
Kind: captions
Language: en-GB

1
00:00:00.350 --> 00:00:05.450
Twitter was set up to support 140 characters.
And in the English alphabet, that's easy to

2
00:00:05.450 --> 00:00:10.100
understand: a character is a letter, number,
space or punctuation mark. People more or

3
00:00:10.100 --> 00:00:14.030
less agree with computers there. And if it
was twenty years ago, that's exactly how the

4
00:00:14.030 --> 00:00:17.170
system would work. That far, no further.

5
00:00:17.170 --> 00:00:18.970
But now, we have Unicode.

6
00:00:18.970 --> 00:00:23.190
Mind you, it's still fairly straightforward
in some languages. East Asian languages, for

7
00:00:23.190 --> 00:00:28.260
example - Chinese, Japanese, Korean -- "one
character" is a glyph, a number, a space,

8
00:00:28.260 --> 00:00:32.629
or a punctuation mark. But since the language
is denser -- each of these characters encodes

9
00:00:32.629 --> 00:00:37.309
more information than an English character
-- you can fit almost twice as much information

10
00:00:37.309 --> 00:00:39.609
into each tweet.

11
00:00:39.609 --> 00:00:41.670
And then, it gets complicated.

12
00:00:41.670 --> 00:00:46.079
Take Arabic, for example. What counts as an
Arabic letter? First of all, the shape of

13
00:00:46.079 --> 00:00:50.550
Arabic letters change significantly depending
on where they are in a word. Watch what happens

14
00:00:50.550 --> 00:00:56.649
as I take the Arabic for "Arabic alphabet",
and hit backspace. Arabic's right to left,

15
00:00:56.649 --> 00:01:00.519
remember. The characters change in order to
be consistent with the rules of the written

16
00:01:00.519 --> 00:01:05.119
language, and the diacritics disappear separately
to the letters they're next to.

17
00:01:05.119 --> 00:01:09.979
In Vietnamese, on the other hand? Each of
those counts as one character.

18
00:01:09.979 --> 00:01:12.240
Backspace, and away they go.

19
00:01:12.240 --> 00:01:17.359
It's at this point that most British programmers,
myself included, throw up their hands in defeat

20
00:01:17.359 --> 00:01:21.310
and just use existing code by some other generous
soul who's already worked the problem out.

21
00:01:21.310 --> 00:01:25.299
Or if they're lazy, they just say, well, no-one's
going to use this who doesn't speak English,

22
00:01:25.299 --> 00:01:26.729
so we don't need to worry about it.

23
00:01:26.729 --> 00:01:28.020
(MOUTHS) Yes you do.

24
00:01:28.020 --> 00:01:33.390
Hmm. Unicode has a single character for some
English ligatures, like "ffi" - notice how

25
00:01:33.390 --> 00:01:38.179
the letters there are smushed together to
make them look better to the eye. Some programs

26
00:01:38.179 --> 00:01:42.539
will automatically add those in for you. So
you copy and paste your text from that into

27
00:01:42.539 --> 00:01:45.749
Twitter, and suddenly you're saving characters.

28
00:01:45.749 --> 00:01:50.119
People would count that as three characters.
Unicode, and therefore Twitter, and pretty

29
00:01:50.119 --> 00:01:55.350
much every computer program? Just one. The
greatest example of this I could find is the

30
00:01:55.350 --> 00:02:01.530
Arabic for "peace be upon him". Unicode has
a single character for this, and Twitter will

31
00:02:01.530 --> 00:02:06.020
treat it as counting for just 1 of your 140.
Which is handy, if you're a devout Muslim

32
00:02:06.020 --> 00:02:08.149
and want to talk about the prophets on Twitter.

33
00:02:08.149 --> 00:02:13.670
So. What counts as a character? Well, it's
complicated. Computers see things differently

34
00:02:13.670 --> 00:02:17.050
to people. And let's be honest: unless you
have a professor who's setting their essays

35
00:02:17.050 --> 00:02:21.250
by character count instead of word count the
only time it'll really matter for most people...

36
00:02:21.250 --> 00:02:23.110
is when they're trying to tweet.

37
00:02:24.340 --> 00:02:26.340
[Translating these subtitles? Add your name here!]