Diceware in Manglish

A Diceware list in Manglish

Diceware?

Yeah, Diceware is this thing you use to generate really good passphrases. But seriously, if you want to know more, you can find out here.

Manglish?

Ya, Manglish. Apparently it’s an English-based creole spoken in Malaysia? Aiya, it’s like Singlish lah but, you know, for Malaysians.

I didn’t know it was a creole. Really? It just feels strange. Creole. Like, sudah upgrade liddat. Rather than just bad grammar for losers, you know? Not bad lah.

Anyway, the variant I’m doing is primarily more Anglophone and reflects my upbringing (I speak Malay, barely speak Cantonese and Hokkien). I expect actually the composition of the word-list to change as more contributors come in.

Why la?

Oh, why la? Okay, like this: while the vanilla Diceware list is okay for gomen work, it damn annoying to memorize la.

I mean, it’s got words like ncaa (apetu?), boise (hah?) and a&p (er…), and it always felt awkward and strange to memorize.

The Beale list is actually a little better, but sometimes it trips me up as well (I mean, I don’t think at&t’s ever been a part of my life, and okay, got abdul but a lot of the names are gwailo names, and y'all? Hahaha no lah no thanks), and you know, I’d like to see gwailo, or cham, or kopitiam, or yamcha, or uols or some kind of local flavour lah. Why not?

So it’s a little more secure than using Beale or vanilla Diceware?

No lah. If anyone knows you they’ll know you’ll be using the list from this. Plus the list is public, and the list of items is public, everything also public. I want it to be about the same level of security as vanilla Diceware and Beale Diceware, just… easier to memorise. If you’re Malaysian. Of a specific demographic.

That matches mine. Right now.

(shamefaced)

Seriously, come and contribute. This is so bangsar it’s embarassing.

Ok, ok, I don’t need to know. Where are you getting the sources from?

Aha, good question! Some are from the Wikipedia page itself, which is of course of dubious quality since any loser from the NSA can come in and vandalise kau-kau. But, just for your reference lah, it’s this snapshot, okay? Some words I don’t use, because I don’t recognise (no, seriously. kapster? not today lah). That one goes to the src/wikipedia.asc list.

Eh got multiple sources issit?

Yaa. For now lah. Like this:

places.asc: place names lah. common ones.
makan.asc: food and so on, because, obviously.
suffixes.asc: three-letter suffixes that are equivalent to the Klang Valley “lah” and East Malaysian “bah”
wikipedia.asc: the list I got from the wikipedia snapshot here.
lain-lain.asc: anything else that doesn’t fit anything else.

Oh, not finish yet issit?

Yep. The goal here is to hit 7776 words, and based on the Diceware Kit page it suggests I should have a nice list of about 6784 words. That’s a tall order, but if I’m aiming for, I dunno… maybe 20% I’ll only need about 1356 words or so. The rest I can use a list of commonly-used words in English or something. Maybe something from Ladybird or some basic reader scheme. Can what, no shame in using that.

So, the list isn’t generated yet. How lah?

Actually, very easy:

Write bash script to wget either the diceware kit file or the diceware 8K list.
put the list together, sort, uniq, and then use some knit the file using bash.
Plonk all the results to diceware-manglish.asc in the root directory.

Right now, to check out how many words there are in the list, I use the following shell command (I use Debian Linux Jessie).

cat src/*.asc | uniq | wc -l

By rights by the time I figure it out and it comes out okay there’ll be a code excerpt in here for you to see and reproduce. Right now for a rough list of words, run this on your command line (tested on Debian Linux Jessie):

cat src/*.asc | sort | uniq

Eh, how come you don’t include the diceware kit list?

Well, I haven’t asked Andrew G. Reinhold, the fella who holds the copyright to Diceware, but releases it for non-commercial distribution, whether he’s okay with me including his kit into the repo or not yet. I dunno also if I want to. Takpe lah, it’s not big deal for now. wget the files needed first, build it as necessary lah. Still early days.

Can contribute words or not?

Can! Just log in an issue first. Don’t send email. That email is a spam-trap. I should have notifications of new issues activated.

Candidate words should be more than 3 letters long. The diceware-kit got 2 letter equivalents, so no need. Even my suffixes are liddat, three or more letters.

Anyway, generally arr, the rule is, don’t use offensive words. I break it down for you:

No sexist words. That means nothing about female genitalia, especially those that are used to imply someone’s inferiority.
No transphobic or homophobic words. I mean it.
No racist words. This goes double. No point oso in the end I shut out someone because I put in a word that’s hurtful to them, especially if they have to memorise also.
No swear words. That’s right, in the Wikipedia snapshot got kanasai but I don’t put in right? Yes. Aiyo, nothing your 10-year old relative should know lah.
Use your common sense.

You don’t like? Fork. Don’t come at me and whine about political correctness. I got no time for you.

Eh, some of your words I don’t think should be in there lah.

Oh, yes arr? Can log an issue and say also or not? I want to know.

Maybe we can have discussion and see what I can do.

tariqk / diceware-manglish