Posted on May 6, 2011 · Posted in FHSST

I just wanted to quickly dump a few thoughts about a big event we’re about to run. I’ve been slipping behind in my blogging and thought I’d inflict a quick core dump on you.

Tomorrow (7th May 2011) we are holding a translation hackathon at the University of Stellenbosch. The objective is extremely ambitious, translate the Grade 10 Mathematics and Physical Science FHSST texts into Afrikaans in one day. We’ve been discussing this possibility for a long time and we finally thought we’d just have a crack at it.

The pressure had been mounting after we were given some airtime on an Afrikaans radio station, Radio Sonder Grense. We were contacted by a number of people wanting to help translate. We had 3 people start doing this virtually but we didn’t make much progress. Most were unfamiliar with the platform and the books and it wasn’t our focus so they got a little lost and gave up. It allowed to start thinking about the process and begin preparing some documentation and workgroups for translation on Connexions.

Then we finally had someone volunteer who was familiar with markup languages and felt comfortable translating the raw Connexions XML, Carl Scheffler. Carl translated 3 chapters into Afrikaans and gave us the first real estimates for how long a chapter should take someone who is comfortable looking at a markup language and confident to do the translation, our average chapter length should take one man day for a first pass translation.

This was the information we needed to proceed, we contact Thinus Booysen, a programming lecturer at Stellenbosch just to see if he’d be prepared to mention it to his class. Thinus jumped straight in and practically organised the participants, formal approval and a venue. So the die was cast!

Given our experiences with regular hackathons, the failed translation attempts and the fact that the translation hackathon is to be a completely new process we’ve tried to put some things into place to mitigate the risk of failure but also ensure that the event has the most meaning for the participants. It is a once-off event so we must get it right.

Event Structure

We know that we need to process people very effectively when they arrive. I’m sure this is obvious but still a challenge, everyone needs extremely clear, simple instructions; they need to be quickly and simply made aware of all the necessary processes, structures and they need to be made as comfortable as possible.

The layout of the room and other visual cues assist in this. They must be channeled to one person who makes sure they’re made to feel welcome, relevant and able to participate. The team must be either introduced to everyone (possibly impractical) or very easily identifiable.

We’re not going to begin with a presentation but actually create a sprint-like working atmosphere to get the ball rolling. The introductory presentation will actually be delay until 1.5 hours after start time by which time everyone should definitely have started working. It seems that if people have to wait until a critical mass has arrived and then sit through a presentation, the uncertainty about the actual process (regardless of the presentation info) builds up and people become frustrated. So our refined instructions should allow everyone to get started on actual translation work without a presentation or much background information.

Community Aspects

We want people, many of whom have never met, to feel as much part of a team and like they are achieving something. To help people get a sense of the team and its progress we’ve developed a small tool to produce a webpage that iterates through all chapters highlighting with a progress bar the state and showing a photograph and name of the translator.

This will be projected on the screen and is linked to a progress spreadsheet which will be continuously updated. Every time it is changed a small script will update the webpage. Anyone will be able to see at any given time how we are making progress as a team who is working on what chapter. This is useful because many chapters build on concepts etc. from previous chapters so there will need to be interaction between translators to answer questions like which word was used for this, how precisely was this defined etc.

The progress webpage will be live here from 9am (GMT+02:00) tomorrow.

Focus Groups

One thing we know is no matter how well prepared we think we are things can go wrong, so we tested our process with a small focus group. The focus group lasted an afternoon and had people who had never used the site before involved. It gave us a great perspective on how ready we were and how volunteers would behave.

One area we’ve had some problems is the performance of the site, this has led to us prototyping other editors etc. We tried to commission a lightweight editor that would allow translator to only edit the proper text and forbid changes to the markup. Unfortunately this hasn’t been completed in time. This is the one tool that could really streamline translation and we feel that it is worth building even after our event is complete.


There are two primary risks, both technical. There is a very real possibility we’ll have editing issues on the site (there seems to be a dodgy proxy-server between us and Connexions that regularly caches a corrupt javascript file and it doesn’t respect refresh requests). To mitigate this I will download all the modules from Connexions early tomorrow morning and people can just translate in any text editor.

The other risk has to do with technical aspects related to the language used. To mitigate this we do have a number of Mathematics and Science educators coming who teach in Afrikaans. It will be important to iterate rapidly in the first hour having them give each translator feedback on a section or two to get them into the right groove.

Anticipated Outcomes

In addition to translating a lot of content we hope to create a mini-documentary video that highlights both the process, the successes and the volunteers who made it possible. We’ll be doing a video interview with everyone who is willing and then try to put up a video in the next week.

If all goes well we’ll have two Afrikaans textbooks next week and a video of how it all came about.

About the Author

Mark Horner is the CEO of Siyavula Education, a social enterprise working in the school sector in South Africa. While working as the Shuttleworth Foundation Fellow for Open and Collaborative Resources, Mark was able to transform the Free High School Science Texts (FHSST) project, which he co-founded, into Siyavula Education. In this process, openly-licensed, collaboratively authored textbooks have been printed and distributed nationally in South Africa. Working at the intersection of community, openness and technology; Mark intends to leverage this success to make Siyavula an innovative, technology provider in education that works effectively as part of the education community to ensure better learning opportunities for all. A recent notable event being the delivery of Siyavula's textbooks over Mxit, the most popular mobile chat solution in South Africa. Mark has a PhD in physics from the University of Cape Town and conducted his research at Lawrence Berkeley National Laboratory in California on the results from the STAR experiment at Brookhaven National Laboratory in New York. His work is carried out in the belief that the liberation of information and support of education in South Africa will lead to a peaceful and prosperous future for all South Africans.