At HP Discover’s annual conference I was interviewed about this project. They have posted the video online via http://hpsw.co/Af9s2J6 I hope you will enjoy it and rate it.

Thank you, Julian Harty.

I gave a tech talk at the Google offices in Zurich, Switzerland about the projects as some Google staff were keen to know more about them. Here is a copy of my presentation slides TechTalkAtGoogleZurich (13 Sep 2013)

I hope they’ll provide another perspective on the projects. Much of the information is also available in other blog posts on this site.

Julian Harty

It started with an invitation to speak at a conference in Bangalore on Software Testing in July 2013. For various reasons the trip included 2 weekends which I wanted to use wisely. Before the trip I was introduced to Raj, of a small Christian organisation called The Fishermen Trust. His organisation helps various schools around the Bangalore area and in Tamil Nadu state amongst other work.

Ahead of the trip Raj and I had a couple of discussions by Skype and email about some of the challenges these schools face. In terms of technology many of the schools had at least one computer for the children to use, some had over twenty computers in a classroom. However, none of these computers were being used by the pupils. The computers were typically old computers donated or scavenged from corporations, with an old version of Microsoft Windows installed. However they’d not been prepared or commissioned for the schools to use, the teachers lacked the skills, understanding or knowledge of how to use them at all, and in most cases there were major problems with the mains power.

I managed to obtain around 100 copies of an excellent software known as Zoombinis. The Zoombinis is educational software devised as various games. Each game requires thinking skills  and presents problems to be solved to ‘save’ small creatures called Zoombinis. There are at least five variations of Zoombinis software, most of our copies are of Zoombinis Mountain Rescue.

We agreed to spend both Saturdays visiting schools so I could get involved first hand and learn as much as practical about ways to help the schools and the pupils. I’ll cover the school visits and the findings in additional blog posts here.

About the Zoombini Games

There are three titles in the Zoombinis series of games. Of these, the Logical Journey was the original title and the best in my view. Several companies acquired the original game and released refreshed versions of the title, sadly their changes did little to improve the product. The newer versions and games are in wider circulation, the original being from 1996 is hard to find these days. The original is the most tolerant in terms of working on a variety of computers.

Here are links to Wikipedia articles about each of the games.

http://en.wikipedia.org/wiki/Zoombinis_Mountain_Rescue

http://en.wikipedia.org/wiki/Logical_Journey_of_the_Zoombinis

http://en.wikipedia.org/wiki/Zoombini_Island_Odyssey

When I started this project I naively bought some 3 watt solar panels from a supplier who advertised them as being capable of charging iPhones, and the latest Android smartphones, etc. As the e-readers, the Kindles, require less power than these devices I hoped these panels would be able to easily provide sufficient power to charge the Kindles, even in the UK.

Cutting a long story short, the panels wouldn’t charge the Kindles except occasionally to keep me on-the-hook so I kept going for much longer than I otherwise would or should have done. They didn’t even charge the devices in the peak of the day in Kenya, on the equator. Also their voltage regulators were unreliable, and lost about 1 watt of the 3 watts the panels were claimed to deliver.

To make matters worse, the kindles often froze because the voltage (at around 4v via the regulator) was borderline to trigger the charging circuit which drew current and caused the voltage to drop, then the charging stopped. This allowed the voltage from the panel to increase which restarted the cycle. Eventually these cycles seemed to cause the kindle to freeze.

I discovered Solar panels are considered constant current devices while we want something closer to constant voltage at a given minimum current. Although I considered creating my own homebrew regulators with large capacitors to even out the power delivery, etc. I abandoned the idea as others had tried better solutions and mine was unlikely to work.

Indirectly I was encouraged to try connecting the solar panel, without a voltage regulator to an inexpensive USB battery reserve power device intended to provide portable emergency recharging for mobile phones. Internally their electronics enables them to charge their internal battery @3.6 volts DC while delivering a steady 5V DC output once they’d been sufficiently charged. These devices seemed to do just what I needed for this project and my initial tests are promising. I’ve tried a variety of these portable battery units, including several with integral small solar panels rated at 0.7 watts, which isn’t enough in practice.

I also discovered http://getvolt.dk/en/ an innovative business for concert goers who want to keep their smart phones and iPads, etc. charged beyond the reach of mains power.

Overall, the combination of an external 3 watt solar panel  and a smallish USB battery unit seem to be a practical solution for recharging the other electronic devices we’re using, including the Kindles and the 3G WiFi routers.

One of the objectives for this project is to gather sufficient evidence about the project. The evidence can then be used:

• to enable the project to be assessed in terms of the impact, applicability, etc.
• to help us improve the project and the outcomes e.g. in terms of the effectiveness, and efficiency of what we do
• To help others to learn from our experiences and our work so they (e.g. you) can build on the good bits and learn from anything else that’s relevant.

I recently read a brief report written by NESTA a UK based organisation. I recommend you read it.  http://www.nesta.org.uk/library/documents/MakingEvidenceUseful.pdf I found the simple Nesta Standards of Evidence figure is effective to help focus the work we’re doing. My impression is we’re working around Level 2 at the moment.

Level 1 states: “You can describe what you do and why it matters, logically, coherently and convincingly” while Level 2 states: “You capture data that shows positive change, but you cannot confirm you caused this”. We are capturing data, however we don’t have enough data to assess the effects of providing the devices to the schools yet.

One of the first major challenges for this project is how to address payments. Very few people in rural Kenya have bank accounts or credit cards. They don’t have much money either, the region is classified in financial terms as being at the bottom of the pyramid[1]. All this means we need to find alternatives to the ‘register your credit card or bank account for automatic billing’ approach often used in the USA, UK, etc.

In my view, Kenya has one of the most advanced micro-payment systems globally, through the m-pesa service and similar mechanisms from the various mobile network providers in Kenya. And micro-payments via mobile phones is ubiquitous in Kenya, including in the rural areas.

Publishers, rightly need to be paid for their work, unless they choose to provide the material without charge. So we have a conundrum, how to pay for the materials for the teachers… And for this project we shouldn’t assume the teachers, or their schools, can afford or pay for the material. A typical teaching book costs around $10 on Amazon.

We initiated the project by creating accounts for the schools where we provide some credit of £25 (around $40 USD) using gift certificates. The school teachers can then choose to use the credit to pay for several books, and there are lots of books, including the ‘classics’ available free-of-charge.

We need ways to defend against uncapped spending, a Kindle user does not need to authenticate themselves to buy items from Amazon, so they don’t need to enter a password or other security information – money is simply deducted from the credit balance or any linked credit card or bank account. Because we use gift cards and we don’t configure the Amazon accounts with either bank accounts or credit cards, the maximum financial risk is the remaining balance from the gift certificates.

We would like to find ways to enable teachers, and other people with access to micro-payment services in Kenya, to be able to contribute towards or entirely pay for eBooks and other items e.g. newspapers, journals, etc. The most popular micro-payment service is m-pesa from Safaricom, the other Kenyan mobile network providers have similar services. The report from iHub Research[1] provides some useful background information on the use of micro-payments. We have approached Amazon and various publishers and online book selling sites aimed at the Kenyan market to explore these possibilities. They would help to enable the teachers, and people involved in the schooling in the country, to take more ownership of the project.

Wishlists are part of the Amazon service. They provide an elegant way for whoever wants an item to track these requests and to share the requests with other people. In our case, the teachers are encouraged to add items to their wishlist, that’s linked to their Kindle account. We, and other people who would like to support the project can then pick books from the wishlist and pay for them. The requestor (a teacher in one of the schools in Kenya) then receives the book the next time they use the device to connect online. There are some restrictions for Amazon accounts on the Amazon UK site that make it harder for people to buy books for another Amazon account. As far as we know Amazon accounts on the USA site don’t have this restriction.

We would prefer the ability for people to buy Kindle books as gifts for other Amazon accounts, i.e. accounts that work like those on the USA site, as it makes the approval and purchasing more decentralised and enables it to scale further. For instance, a sponsor could fund various books for a school by picking them from that school’s shared wishlist, without any need for intermediaries.

Using wishlists is an effective way to spend the very limited funds wisely, particularly for shared devices, and currently each of our devices may be shared with around 10 teachers at a school.

Another effective mechanism is the ability to download a sample of Kindle books. The teacher can then read the first section of the book to help decide if the book is worth ordering. However we have noticed some samples have very little of the book’s content, possibly for copyright or other commercial reasons? When there is little content the teacher is effectively buying blind – i.e. without being able to read a useful sample first.

[1] The following report, written by iHub Research in Kenya, covers mobile phone usage at the base of the pyramid. It provides a good overview of many relevant factors for this project. It can be downloaded from the World Bank blog here. Other web sites have links to it too.

As an engineer I realise a common risk is to have single points of failure in any system. Therefore, on of the design goals for the project is to reduce the number of single-points-of-failure. In practice this means we are seeking ways to share the knowledge, the support, and how paid content is both selected and paid for.

Our approach of using parallel paths also helps reduce single-points-of-failure.