If you don’t already know it, Raspberry Pi is not a low-cost computer. It’s an ultra-low-cost computer (see photo below). And it was the subject of a recent demonstration and discussion workshop (see links for video) for CDI members in Manchester. This focused on the development-related potential of Pi and its add-on interface ”Pi-Face”, which is being developed at the University of Manchester by Andrew Robinson.
Although credit card-sized, Pi is a fully-functioning computer. Hook up a keyboard, mouse and monitor and away you can go with Linux and, for example, OpenOffice. And, as noted, it is ultra-low-cost. The actual production costs will depend on scale, with the economics catching even Raspberry Pi Foundation – the non-profit creators – by surprise. Expecting they might eventually ship around 10,000 Pis, they have already shipped more than one million.
At those sorts of production scales, costs for Pi could be reduced to around the US$15-20 mark. Adding a keyboard, mouse and Pi-Face will stack less than US$2 on top, and looking at similar products it is likely that a small screen can be produced for US$15. Of course, cost is not the same as price but we are talking of a complete computer system that will likely cost less than US$35 to produce and perhaps US$50-60 to buy. Just the Pi-plus-Pi-Face combination could be supplied to developing countries for as little as US$25.
In many ways, its key attributes are those of a mobile phone (not surprising since it runs with the same ARM chipset you’ll find in many mobiles):
- Very low cost puts it into the category of “semi-disposable” device, and a ready addition to many other innovations without breaking the bank.
- Its robustness and low maintenance requirements make it particularly suitable to harsh developing country environments.
- Its small size and portability make it suitable for applications that other computers can’t reach.
- It has very low power consumption, so can work more easily in electrical off-grid environrments.
But it’s not a mobile phone, and you can’t use it for calls and text. What it does do is connect readily to a host of other devices. And, unlike a mobile phone, it is easy to customise, using common open source software and “tinker-able” hardware components. All run by a .org not a .com organisation.
Raspberry Pi may just fizzle and die, without much effect on international development. But the potential is certainly there for it to paradigm shift ICT4D. The mobile phone explosion has shifted ICT4D’s emphasis towards the “C”, with widespread acceptance that “m-development” models will dominate. Raspberry Pi could shift us back towards the “I”; towards the computing and data processing and automation that were the origins of ICT4D in the 1970s and 1980s but which have fallen by the wayside.
At present, Pi is a solution looking for development problems, but three application areas spring to mind:
a) Micro-enterprise and household computing: providing access to standard computing applications not for the community but for the individual enterprise and household. Add an Internet connection and we might call it not OLPC (the One Laptop per Child initiative) but OTPH: a one telecentre per household approach that moves us beyond community computing models.
b) Technical education: the prime motivation behind Pi was to reignite interest in computing as a subject among schoolchildren. There’s a great thirst for IT education in schools, colleges and universities in developing countries but budgetary constraints are a major barrier (see earlier blog entry on revising computing curricula in Africa). Pi can help to overcome those – the possibility is that it could do all the OLPC does at half the price, and allow kids to open the box and play about much more, learning how IT works.
c) Data collection and automation applications: there’s a trickle of new electronic applications for development – smart motor controllers that save power and extend motor life, low-cost health monitors, water quality and climate change measurement devices, field-based agricultural sensors. Raspberry Pi could turn that trickle into at least a stream if not a flood.
The promise of Pi, at root, is to enable a new ICT4D innovation paradigm: one in which Pis are widely used and understood within developing countries, and in which grassroots innovation is really possible for the first time in the ICT4D domain (see earlier blog entry on grassroots ICT4D innovation). There’s no reason the same informal sector micro-entrepreneurs who now fix mobile phones can’t also work with Raspberry Pi. But they can customise and adapt this technology much more than they can a mobile phone. It can therefore be appropriated far more by the base of the pyramid.
Pi also allows a new model of collaborative innovation: that done working alongside base-of-the-pyramid consumers. Large firms, university departments, social enterprises can now afford rapid, mass prototyping – trying out and iterating quickly through many different models until they find one that works.
As yet, of course, this is promise not reality, and one can foresee plenty of issues around everything from distribution through support and training to growth in e-waste. But the international development impact of Raspberry Pi – good or bad, large or small, paradigm-shifting or incremental – is up for grabs. Over to you.
Thanks to the kindness and openness of “Goran Podolski”, I’ve been allowed to see the balance sheet of an East European gold farm, selling virtual currency for the online game World of Warcraft (WoW). A summary of its operations follows, but gold farming is a complex and unusual process, and more details on it can be found in the online report: Current Analysis and Future Research Agenda on Gold Farming.
What and Who: this is an informal (i.e. unregistered) micro-enterprise run by Goran, who is a student living in a major city in Eastern Europe. He has been playing in-game himself to make gold and he also paid two other people, whom he knows via personal connections, to play characters on his account. He has taught them how to play.
Making Virtual Gold: because WoW players have their characters on a specific European-based server, they can only buy gold that has been produced by a gold farmer on that server. Goran has therefore placed his gold-farming accounts on two (out of a possible c.250) servers he thinks are particularly “high-yield”, i.e. active and popular with players. His high-level characters have particular in-game “professions” – abilities – that are used to make or gather items (e.g. gathering herbs using high-level Herbalism and then making items with high-level Alchemy, or making items with Inscription). These items can then be sold in-game, typically via the game’s auction system, thus producing in-game gold. (He has chosen this method as more productive than what he sees as the approach used by many Chinese gold farms of fighting non-player characters, picking up items dropped when the NPCs are killed, and then selling these for gold.)
Selling Virtual Gold: most of the virtual gold has been sold to merchant-portals – the Web sites through which the great majority of players buy gold. The two key portals used in this case are www.pusada.com and www.offgamers.com (OGM). Players who wish to obtain in-game gold will buy from these merchants via their Web portals.
OGM are Malaysia-based and the sales process is undertaken via chat on their portal or via mobile phone calls. When a customer indicates through the Web site that they are ready to buy, Goran is sent the in-game name of their character by OGM, and OGM calls the customer on their mobile. Goran and the customer log into the game and the gold is transferred from Goran’s to the customer’s character: Goran sends screenshots to OGM as confirmation for his payment. If Goran is late, OGM will call him on his mobile, so this is not quite the anonymous process it might appear. OGM outsource some parts of their sales work (e.g. customer chat) to China. Likewise Pusada – who are US-based – outsource to sub-contractors in other countries. Unlike OGM, Pusada headquarters staff never go in-game but only broker the sale via IM chat.
Speaking English is thus a key part of Goran’s added value. His sub-contractors do not speak English and so, for example, are unable to sell the items they produce in the WoW auction houses to make the in-game currency; Goran has to do that himself. Likewise, thanks to his English, Goran is able to make money in a different way – by acting as the broker for a local gold farming workshop, selling their gold to the portals above and also to www.net4seller.com and (now defunct) www.mmoinn.com. He charges the local workshop a commission of about 12% for this service.
Financials: Revenues are affected by two main forces. First, the price that merchants charge to the final consumers. This has been steadily falling as virtual gold continually devalues against real-world currencies. For example, from March 2009 to January 2010 the US buy price of 1,000 WoW gold depreciated from US$14.50 to US$7.00. Second, merchants price and pay in US dollars (even though the virtual gold is sold in euros to the European player-buyers). During periods when the dollar appreciates against Goran’s national currency, this further erodes revenue. From March 2009 to January 2010, the US$ rose by around 10%. The result was a fall in Goran’s revenue from US$7.10 per 1,000 WoW gold in March 2009 to US$3.37 in January 2010.
Costs are made up of three elements: the cost of the World of Warcraft game accounts; game cards which cover the cost of a certain amount of in-game play; and the wages paid to his ‘sub-contractors’.
Nominally, Goran’s payments to those who farm the gold for him are paid on the basis of the amount of time they spend in-game with his characters rather than, for example, on the basis of what they produce in terms of either items or gold. (He tried the item-based approach but it became too complex due to the variety of items produced and different prices earned; pay-per-gold is not possible since the players themselves are unable to sell their items in-game for gold.) He paid just over US$7.00 per eight-hour working day, which was almost exactly the same as the minimum wage for a young person in this country. However, as noted, this is nominal – over time, the pay would be more akin to a certain percentage of the overall profits made during a month.
The figures can also be used to get a sense of the mark-up being charged by the merchant-portals. In March 2009, Goran was paid an average of US$7.10 per 1,000 WoW gold, and the average sale price from portals was around US$14.50: a roughly 100% mark-up. This was very similar in January 2010, when Goran was paid an average of US$3.37 per 1,000 WoW gold by the portals, while their sell price was around US$7.00.
The balance sheet below shows a sample nine-day period in January 2010 during which ten sales were made: all but one via the merchant-portals. Below the revenues earned from these sales, Goran’s calculations for total number of hours his sub-contractors spent in-game are shown. Hours are converted to days and paid at just over US$7 per day with account and game card expenditure then added. The balance is shown at the end, indicating a daily profit for Goran just greater than the national minimum wage.
|East European Gold Farm Balance Sheet: 9 Days in January 2010 (for World of Warcraft)|
|Date||Quantity sold||Price per 1,000 gold||Total Price||Buyer||Overall|
|Day 1||20000 g||$3.00||$60.00||pusada.com||Average Prices||$3.37|
|Day 1||2000 g||$3.73||$7.45||offgamers.com||Total sales in gold (thousands)||66.80 k, gold|
|Day 2||3000 g||$3.73||$11.18||offgamers.com||Total Sales in USD||$224.85|
|Day 2||3000 g||$3.73||$11.18||offgamers.com|
|Day 3||2300 g||$3.73||$8.57||offgamers.com|
|Day 4||10500 g||$3.25||$34.13||pusada.com|
|Day 5||5000 g||$4.20||$21.00||Individual player via MSN|
|Day 7||14500 g||$3.25||$47.13||pusada.com|
|Day 8||3000 g||$3.73||$11.18||offgamers.com|
|Day 9||3500 g||$3.73||$13.04||offgamers.com|
|Character||Level 80 Alt|
|Days||Hours||Mins||Character Time-play Recorded|
|Character||Level 65 Main|
|Days||Hours||Mins||Character Time-play Recorded|
|Character||Level 80 Main|
|Days||Hours||Mins||Character Time-play Recorded|
|Total hours worked||115.55|
|Total days worked||14.44|
|Pay per day||$7.14|
|Average Daily Net Profit||$8.47|
Postscript: By March 2010, Goran had expanded his gold farming business to employ nine farming sub-contractors and a business manager to oversee them all, helping to pay his way through university. By March 2011, though, Goran has – at least for the moment – stepped out of the gold farming business; in part to focus on his studies, but also because profitability is increasingly difficult. At the time of writing, the advertised buy price on www.offgamers.com was US$1.00 per 1,000 WoW gold (i.e. less than one-third what it had been a little over a year previously); the advertised sell price was US$1.54. This means the sub-contractor model would no longer work well. Goran has been making some money by renting his WoW game accounts and a GatherBuddy bot (for automating in-game activity: like gold farming, this is also against game rules) license on a monthly basis to his former sub-contractor who was doing some gold farming himself, though finding it ever-harder to make a profit and, at the time of writing, likely to drop out because prices were so low. Goran is currently offering his full support to US calls for revaluation of the renminbi!
BoPsourcing – the outsourcing of work to bottom-of-the-pyramid communities – is on the rise. Outsourcing used to mean sub-contracting work from one big firm to another nearby. Then, with offshoring, the contracts crossed borders. With BoPsourcing, the contracts cross several income strata as well.
BoPsourcing initiatives can be found that are run by:
- Governments: such as the IT component of Kerala State’s Kudumbashree initiative which has created more than 2,500 jobs for women from below-poverty-line urban households
- Social Enterprises: such as Anudip Foundation which set up its first ‘MERIT’ (mass employment through rural IT) Centre in rural West Bengal in 2010. (There are suggestions that some 200 BoPsourcing initiatives are currently being run by social enterprises and NGOs.)
- Private Firms: such SourcePilani which operates a 50-person business process outsourcing centre in Rajasthan
All these, and many other examples from India, are onshore BoPsourcing. There are also offshore varieties, such as Digital Divide Data, which outsources from large US organisations to telecentres in Cambodian small towns and villages.
BoPsourcing has no necessary connection with ICTs. Indeed, many agricultural value chains are arguably examples. Here, though, my focus is on ICT-related outsourcing. Examples of contracts include data entry, transcription, digitisation, call centres and ICT training.
One of the key concerns about outsourcing and developing countries has been the potential to fuel income and other inequalities. I could already see this studying software offshoring to India in the 1980s: an economic shearing in which those involved saw their incomes pull far ahead of the bulk of the population. More recent evidence suggests offshoring increases overall wage growth but also increases inequality.
BoPsourcing presents an obvious solution. Channelling the benefits of outsourcing down to the poor can drive wealth creation for those on the lowest incomes, and serve to reduce inequalities.
So. Job done. We can add BoPsourcing to our list of great development solutions.
Well, not quite yet. First because the evidence base is very weak. Second because BoPsourcing comes in nearly as many varieties as Heinz. Figure 1 summarises.
Figure 1: Continuum of BoPsourcing Approaches
There are at least four conceivable models that form the continuum (but feel free to add your own evidence and ideas):
- Exploitative outsourcing seeks to bear down on wages and working conditions in order to minimise costs and maximise profits. The result is an ICT sweatshop that does little to grow incomes, to deliver empowerment, or to reduce inequality. At present this seems more of a bogeyman brandished by those at the other end of the continuum, than it is an evidence-based reality. The potential, though, is certainly present with so many outsourcing firms seeking to drive down costs.
- Commercial outsourcing reflects, for example, the steady diffusion of outsourcing in India and other nations, from cities to large towns to small towns and beyond. Whether this can yet be called BoPsourcing (e.g., forgive the pun, whether BPO is BoP) is unclear. Quite likely commercial operators have to date only got as far as large towns. But the migration trend is clear, and it will reach poorer towns and even villages soon enough. As for inequality, the effect is likely to be as arguable as it is for outsourcing generally: evidence is contested and, unfortunately, fought more by economists pitting ever-more complex models against each other, than on the basis of field data.
- Ethical outsourcing (also known as socially-responsible outsourcing) takes commercial outsourcing and requires that it meets certain minimum standards; typically relating to labour practices but also starting to include environmental issues. The International Association of Outsourcing Professionals has taken a lead on this. This is likely to have some impact on inequality but, again, the extent to which such work really reaches the BoP as yet is questionable.
- Social outsourcing (also known as developmental outsourcing) differs from ethical outsourcing as fair trade differs from ethical trade. Ethical outsourcing involves existing commercial players with either a commitment to or measurement of adherence to standards. Social outsourcing involves new non-market intermediaries who sit between the client and the BoP sub-contractor. Social outsourcing most definitely does reach the BoP; indeed that is its raison d’être. It has already been shown to increase incomes, increase asset holdings, increase skills, and increase empowerment. It is therefore likely to reduce inequality.
True ICT BoPsourcing is on the increase – you only have to monitor the growing number of initiatives to see that. Much of the more commercial end of outsourcing has yet to get this far. It’s more like MoPsourcing (i.e. middle of the pyramid) just now, but cost pressures, supply-demand gaps, ICT diffusion, and growing awareness of BoPsourcing mean this is changing.
The impact of these trends on inequality will depend on which outsourcing model comes to dominate the BoPsourcing business. If social outsourcing wins, then so too will the poor. If exploitative outsourcing wins, the opposite will be true.
In practice we may well see some messy combination of models, as social and commercial approaches intersect. For the social outsourcing intermediaries, the lure of big clients, contracts and growth may pull them into bed with commercial operators. Conversely, the operators will be attracted by the contacts, expertise and CSR-window dressing that social intermediaries provide. The poor themselves will get jobs, skills and income. Whether they will see the structural transformation necessary to really clobber inequality, only time will tell.
Where are the Amazon and eBay for international development? ICT has delivered new models for business and commerce; is it also delivering new models for international development?
A new short paper, “Development 2.0: Transformative ICT-Enabled Development Models and Impacts” outlines some initial ideas (summarised below), building on an earlier Viewpoints column in Communications of the ACM.
Ideas about Development 2.0 must be initial because ICTs have only very recently diffused to the bottom of the pyramid (with many gaps and inequalities remaining). A first pass suggests three potentially-transformative, ICT-enabled development models:
- Direct Development delivers resources and services without the intervention of traditional development actors; where those resources and services can be digitised. Examples would include Kiva, MYC4 and similar micro-lending platforms.
- Networked Development occurs neither solely through the state and similar agencies nor through the market, but through a mesh of actors and institutions that are connected and can act together through ICTs. Examples include txteagle’s crowdsourcing, and the ‘crowdvoicing’ of e-participatory budgeting. (See also an example of emergent ICT-enabled networks impacting development.)
- Grassroots Development occurs from within poor communities, as a result of ICT-enabled empowerment. Examples include beeping/flashing, and use of airtime as currency. (See further discussion in an earlier entry on grassroots ICT4D innovation.)
These models could only be judged transformative if they are having real and significant new development impacts. Evidence is only just emerging, but five types of impact are starting to be seen:
- Connecting the excluded: providing information and other livelihood assets including social capital that were previously unavailable.
- Disintermediation: cutting out the gatekeepers who prevent access to resources and services, or who charge rents for such access.
- Digital production: enabling those in low-income communities to become producers of digital content, and to develop ICT-based productive livelihoods.
- Digital innovation: enabling those in low-income communities to appropriate technology to such an extent that they start to do new things with it.
- Collective power: enabling communities to bring the power of the group to bear in the service of economic or socio-political agendas.
These ideas on models and impacts, however, leave many questions:
- How should we frame and conceptualise ideas about Development 2.0?
- How can we properly distinguish between an incremental and a transformative effect of ICTs on development processes and structures?
- Where can we get independent, long-term impact data relating to the new ICT-enabled models, as opposed to the current ‘evidence’, much of which is anecdotal and/or written by those with vested interests.
These and other Development 2.0 issues form part of the ongoing research agenda for Manchester’s Centre for Development Informatics.
Are we seeing a phase change in use of ICTs for international development?
The paper outlines some of the emerging characteristics of ICT4D 2.0, based on research from the University of Manchester’s Centre for Development Informatics, and other sources. Feel welcome to comment and add your own observations to this list:
a) New Hardware Priorities: a need for innovation around low-cost, broad-reach terminals, telecommunications, and power. A need to bring the hardware success story of the last decade – mobiles – even more centre stage. The paper also discusses implications of broadband, cloud computing, and individualisation of hardware devices.
b) New Application Priorities: the growth of participatory content creation, and the use of ICTs to create new income and employment for the world’s poor. The paper also discusses implications of FOSS, and the growth of applications to address urban poverty, security, economic growth, and climate change.
c) New Innovation Models: the growing need for – and potential of – innovation that moves beyond top-down, laboratory-type models. This includes collaborative (para-poor) models that work alongside poor communities. It also means greater attention to the grassroots (per-poor) innovation that is arising from within those communities. The paper also discusses the new innovation intermediaries that are emerging in private and NGO sectors.
d) New Implementation Models: based on the limitations of ICT4D 1.0 projects, there will be greater emphasis on sustainability, scalability and ICT4D project evaluation. This will necessitate more process than blueprint approaches to implementation, and better techniques for closing design—reality gaps. The paper also discusses new funding mechanisms and new organisation forms that are increasingly seen.
e) New Worldviews: effective ICT4D 2.0 policies, strategies and projects will require “tribrid” champions. They must understand enough about the three domains of computer science, information systems, and development studies to draw key lessons and to interact with and manage domain professionals. Training programmes and working group formation must reflect this need.
The paper also discusses the need to move beyond ICT4D mainstreaming, to plan ICT4D policy structure and process as much as content, to engage with the growing “Development 2.0″ agenda, and to shape ICT4D research priorities accordingly.
Above all, it argues, ICT4D 2.0 will require a new worldview of the poor; no longer characterising them as passive consumers but, instead, seeing them relate to ICT as active producers and active innovators.
Is implementation of fiscal policy in developing countries being increasingly handled by mobile phone operators?
That’s what hit me from two presentations at the recent 1st m4d conference; in particular from Adam Denton’s keynote [2MB pdf] on behalf of the GSM Association. (This was full of interesting evidence-based nuggets, some of which I’ve added at the bottom of this post.)
My thoughts on government finances arise from:-
- a) the general poor performance of developing country governments in implementation of fiscal (tax and spending) policy;
- b) Mick Moore’s argument (e.g. in the book, “Taxation and State-Building in Developing Countries”: you can find on Google books) that state-building in developing countries is significantly undermined because governments don’t rely very much – if at all – on their citizens for taxes; as a result, governments find it easy to ignore citizens when making policy.
Mobile telephony could be changing this, albeit in an unplanned and as yet largely unrecognised way. How? Because mobile phone operators are key, and rapidly growing, contributors of tax revenues to government. They average 7% of tax receipts in Africa and, in some countries, are the single largest tax payer.
More importantly, and almost uniquely among tax sources (compared, say, to customs revenue or taxes on natural resource extraction), mobile phone companies derive their revenue from a large and increasing mass of the citizenry of a country (around 30% of the cost of mobile phone ownership goes to pay tax). They are, therefore, indirectly providing the tax connection between citizens and governments in developing countries, the absence of which Mick Moore and others have long lamented.
Previously, if governments wanted to raise or lower the tax burden on their citizens, they had few if any levers. Now they do. Assuming a competitive mobile market, changes in the tax burden on phone companies will directly affect large swathes of a country’s population. Raise taxes on mobile operators: you extract more money from millions of citizens. Lower taxes for the reverse effect.
The fiscal flipside could also apply. Katharine Vincent and Nick Freeland are working on a social protection project in Lesotho. Cash transfers to the poor are an increasingly-popular social protection tool to address poverty and wider vulnerabilities. However, delivery mechanisms are difficult. One possible solution? Get government to outsource delivery of the cash/credit to mobile operators, who would effect the transfer via mobiles. Mobile operators would thus be implementing government spending policy.
That may be some way in the future. However, we do need to recognise the extent to which mobile phone companies are mediating – and could mediate – fiscal relations between governments and citizens. This one more sign of “Development 2.0″: the way in which ICTs are reworking core development structures and processes.
Any other examples of phone operators as outsourced Treasuries would be welcome.
Other nuggets from Adam Denton:
- 10 years ago, Manhattan had more phone lines than all of Africa. Today, Africa has more phone lines than the US and Canada combined.
- GSMA states that every US$1.00 invested in mobile telephony generates an average US$0.80 in tax revenues for developing countries.
- GSMA predicts that 80% of broadband delivery in Africa will occur via mobile; but that may require release of spectrum currently used by analogue TV.
- Population coverage in Africa of mobile networks grew from 10% in 1999 to 60% in 2007, and is predicted to grow to 90% by 2012.
- Mobile operators provide 34% of the revenues for universal service funds but receive only 5% of disbursements (fixed line operators get the rest). In 2006, more than US$6bn in universal service funds was collected; only US$1.5bn was disbursed.
- (A personal conclusion): we know the menu for mobile and telecom and ICT policy quite well already; the interesting question today is not what should be in policy but how and why policy is made and implemented.
And, yes, I do realise that the GSM Association has a particular worldview. If you wish to challenge their figures with alternative evidence, please feel free to comment. If you’d like more of the same, then see: http://www.gsmworld.com/newsroom/index.htm