5 minute overview

From Alistair Mann / csi18n
Jump to: navigation, search

North America's comic book market was worth $700 million dollars+ in 2013. What would it be worth to a publisher that they could market test 'any' foreign language version without the cost of paper, printing and distributing? 1,442 movies were made in 2013, grossing an average $7 million. What would it be worth to a film studio to have their movie accessible to the 95.2% of the world whose mother tongue is not English? I click on a British English use of "colour" and change it to "color", and everyone using that App in American English sees the spelling too, without interfering with those who see "colour" or "farb" or "farbe" or "颜色" or "warna", without downloading one of 35 language packs or even wholly different programs.

csi18n (From crowd-sourced internationalisation) uses existing Language negotiation facilities within HTTP to enable your clients to upload the translation of your material that makes sense to them. Subsequent clients are then presented with material appropriate to their language. You get to expand your market, they get a more useful service, and the both of you get it for free.

How it works

If you have produced a game, a website, or an App, a smart TV, or anything that can connect to the Internet, your users can upload text translations of material you choose to make available, and which can be immediately reused with others who use the same language without recompilation of your product or roll out of service packs.

This service is secure, comes with locks, moderation and sanctioning facilities. It also comes with inbuilt intelligence such that the more broadly acceptable the translation, the more likely it is to be used.

The service is accessed over HTTP.

HTTP

HTTP, the language of the world wide web, contains within it the ability to ask for a particular language to be used in a reply.

Example of headers supplied by a browser. Here, British English is requested, with Klingon as second preference

A normal computer will be localised: it will have been told to use English vs German or any other language when originally manufactured or installed. When you use a browser, this knowledge is embedded in the HTTP request used to obtain a webpage, even though it is often ignored. You can check this with your own browser here - have a look for the "Accept-Language" header.

HTTP is not restricted to browsers and websites, it can also be used directly by office software, games, media players - in fact any device that can connect to the Internet.

HTTP also contains within it the facilities to create, edit, delete and retrieve resources on remote servers. If you have a thing that can connect to the Internet, it can make use of this service.

A simple walkthrough

You create a service using English. Alice, who is the only English-speaker in an office of Russian speakers, wants her colleagues to use your service too. As you use csi18n, she can upload the Russian translation for any English you use. Her colleagues visit: the HTTP request for Russian appears, and the reply back is not in the English Alice first read, but immediately in the Russian she uploaded, without waiting for you to have your service professionally translated.

This creates network effects as your service is now more useful to Alice, and Alice's use of it brings more clients to you.

Wisdom of the crowd

A mock up of a video being subtitled from within its player

You have a video player that offers subtitles of movies such as TED lectures. Alice and Bob both offer translations of one video into Spanish. The csi18n makes both available and your player allows its users to choose one translation or another, perhaps even down to the level of individual words or sentences.

Carol comes along, and prefers Alice's translations. Dave prefers Bob's. Eve also prefers Bob's. At this point the csi18n service realises Bob's translations may have broader acceptance, so when Frank watches the video with Spanish subtitles, he is twice as likely to see Bob's translation as Alice's.

In semi-randomly choosing among competing translations this way, the csi18n service will issue broadly acceptable translations in much the same way Wikipedia issues broadly acceptable knowledge.

Smaller crowds

An institution wishes to translate academic papers into other languages, but wishes to heavily restrict who may do the translation, perhaps to given scholars. The csi18n service supports this too: they may assign subusers who are permitted to make changes as well as provide for some of their number to have moderation power: editing, deleting and even user sanctioning. Privacy is as enhanced as quality - no others may read the work until such time as it is judged ready.

Wider audiences

A car manufacturer wants to market a car which in English they'll call the Nova. "No va" in Spanish means "It doesn't go". The csi18n service will allow the audience to improve and approve such suggestions, none of which will appear before the manufacturer checks them out.