MobEA V Workshop: Mobile Web in the Developing World
8 May 2007
WWW2007 Conference
http://www.research.att.com/~rjana/Program2007.htm

Notes:

Galit Zadok:
Developing Regions, not the developing world.
Nigera: Biggest selling Nokia is the N70
African countries are

Ubiquitous computing likely to emerge in developing countries.

Universally accessible

There has to be a value by those in the supply chain in order to make it work.

Kirthi Ramamritham
Digging deep into the rural areas.
Relevance of answers to Web queries is low to those in the rural areas.
Emerging nations left out of the process...
Multi-language -- converting from english into hindi, etc...

Village kiosk services: einformation about e-governance, land records, etc...

Proliferation of mobile phones is a boon -- transmit information of relevance to the rural person.

"The problem is the content."

aAQUA -- creating content from the grassroots -- [knowledge management]
Communication framework to allow rural people to ask questions and get answers.
http://aaqua.org - Almost All Questions Answered
Eg: "How to take care of goats during heavy rainfalls" - answers come back from subj matter experts
The answers can be given in a context sensitive manner -- because the location of the user is known...
"How do you make the experts 'own' the answer?" -- get people to own up.
Crop Doctor section:
"My mango tree has this disease [camera phone image]. What do I do?" - then dialog.
Finding the best market prices for your goods -- data also comes from mobile users (currently via SMS).
"Many of the phone sets sold in India are multilingual" [English/Hindi]

Multilingual information access -
question comes in in Hindi, goes to UNL graph, into UNL document. UNL query against UNL repository. Result comes back in Hindi.
Questions often have multiple languages.

Collaboration is required for the project to succeed -- kiosk owners, farmers, experts, etc...

Interface is in test on mobile browsers.

What is the business model? "We let the business model evolve." Kiosk owners charge 5 - 10 rupees for a page of answers. In rural areas this doesn't work so they have supplemented with advertising. [ad-funded] Company makes money through SMS. Some entrepreneurs have built businesses around it. We did the right thing by not having a business model on day 1.

Charles McCathieNevile

[gave run-down of the workshop held in Bangalore last December]

"We are still selling black and white sms and voice only phones" - Nokia
A lot of work on using SMS based systems to access the Web.
Opera believes in a baseline where people have Web access - Opera Mini.
Report was produced (http://www.w3.org/2006/07/MWI-EC/exec_summary.html)

"SMS-based web is out of scope for W3C."

Ken Banks: There was a debate of "what constitutes the Mobile Web" -- is SMS part of it? Is skype on your phone part of the web?

"Efficient XML Interchange" could be important. As a user paying for data, efficiency is good. Big role W3C plays.

Mobile Web Initiative is built to make the Web more mobile friendly.

Michael O'Farrell: Have you defined the mobile web?
Galit: You've got to put it out there and see what is successful.

"There is a mobile web in the developing countries - as a part of the Web as a whole, it's a significant piece."

For example, putting children's exam results online -- they wanted to move from SMS to mobile web browsers.

Michael O'Farrell: Do we know how many browsers are in the market?

~150 million mobile phones
< 50 million PCs

No statistics on 150 million.

Nokia phones announced recently -- low cost phones -- do not have a browsers.

Ken: older phones have WAP browser.

When price came down from 5000 to 3000, huge jump. Now it's less than 1000.

Prepaid is more dominant.

Output of the workshop: Stephane Boyera from W3C is moving forward with an initiative and requesting EU funding for "promotion of the mobile web in developing countries." Other followup: mailing list that has been quiet.

Mobile Web Initiative work should be pushed to work harder on cases relevant to the developing countries.

DKA: Is W3C well placed to provide a coordination role between actors in this space for collaboration?
CMN: It's an open question. Not clear to me how much they're doing to fill this role right now. EU money could help them start next year.

Michael O'Farrell from dotMobi Advisory Group (MAG)

 

[presentation on dotMobi]
http://advisorygroup.mobi

Developer forum - dev.mobi, ready.mobi

Dan dotMobi play a role to collaborate with W3C, GSM Association, etc... in "mobile web in the developing world".

This is a "commercialization opportunity" to take the technologies, standards and new ventures to the local/regional markets

dotMobi AG could take one perspective -- we need to partner and collaborate with other organizations

What are the new business models - developed countries vs. developing countries? Could new commerce / advertising models play a role in subsidizing the consumer costs? Commerce /Advertising featuring government initiatives and socio-economic benefits - from education/training to health-care/monitoring to grants/credits for information exchange, commercial market transactions, etc.

We are interested in industry collaboration and public-private partnerships -- including hosting a task force, round table, syndicated market research programs, local venture/initiative support, etc...

Current MAG Task Forces:
    Browsing - S. Prashanth, Akmin
    Email - Lauri Hirvonen, Nokia
    Commerce - Michael Keefe, Carlson Marketing
    Advertising - Eric Eller, Millennial Media

Pending task force on "Mobile Internet in Developing Countries."

Interest from Afghanistan, Africa, South America, India ...

Justin Thorp

"World Digital Library" Initiative
http://www.research.att.com/~rjana/mobea2007/ppt/wdl-developing_world.ppt

[Shows video of world digital library prototype / mockup.]

Looking at low cost PCs and other options, and Mobile Phones

Kenya stats:
1999: 15000 mobile phones
2004: 3.4 million mobile phones
last 18 months: 5.6 million phones
few hundred thousand households with electricity

Talking to International Children's Library -- they are looking at how to make those books available to children through mobile phones.

One approach: making primary source content more mobile-friendly

Language is a big challenge. They are looking at 7 different languages for launch.
(English Spanish french Portuguese Russian Arabic Chinese)
We have a "translation memory" system -- volunteer translators around the world?

Just prototyping mobile interfaces now.  Prototype of the web site that's due to UNESCO in September.

How much data? A lot. Partnering with library of Alexandria.

Collaborative features -- bookmark and talk to others that are interested in same content you're interested in.

Ken Banks, kiwanja.net

http://www.research.att.com/~rjana/mobea2007/ppt/MobEA_V_presentation_kiwanja.pdf

"Mobile Web isn't important. It's getting information via the phone that's important."

"Appropriate technologies" need to be considered based on cultural context, the environment, language, user-need, geography, infrastructure, etc.

Defining the mobile web. - we need some idea of definition, i.e. do we include SMS, MMS, non-browser apps (such as Skype)

Who is the target: rural farmers, or health-care professionals? Single approach or multiple approach? Who are the customers?

Internet uses in Africa -- News/current affairs, relationship building, eGovernment, email/im, career-based

Cultural angle of various projects often overlooked.

  e.g. SMS alert system for communities "lacked trust" -- they didn't want to give their mobile numbers up.
  IVR information service in Uganda -- also failed - people didn't want to talk to a machine (we put up with it, they didn't)
  Wind-up radio - effort in reducing the size unnecessary based on initial user reaction trials in Soweto, S.A.

Will "One Laptop per Child" succeed?
Galit: example of dust storms in western Africa ruining a VAIO laptop. Sony - "not supposed to be used in such environments"
"This is not a laptop project -- this is an education project." - Negroponte

Shows slide of phones on sale in a bazaar in Pondicherry (all low-end)
Reference info from Galit on N70 as most popular phone in Nigeria.
"I come from a poor country. In these countries, having S60 is a sign of prestige"

Technology alone is not the answer -- it is an enabler.

DKA: Web as a social phenomenon as well as a set of technology.

Positives and negatives with mobile web.
Inappropriate content.

Michael O'Farrell: Operator can sometimes help out with the inappropriate content concerns.

Cell broadcast -- why isn't it being used?
Guillermo: In [Voda] Netherlands, they are using this, but in general the experience is different across all handsets.

E.g. cantenna.

Phones are shared in developing countries, and are not so much the "personal device" we see them as

Shows: "SIM-Pill" -- GSM device on the side of a pill bottle. Sends a text whenever the patient takes a pill.

SMS "Hub in a box" software (good example of an appropriate, empowering "plug and play" technology) - FrontlineSMS (www.frontlinesms.com ) - used successfully in Nigerian election (ref: BBC article) and awarded from WSIS.

[Lunch]

Jalal Mahmud
http://www.research.att.com/~rjana/mobea2007/ppt/MobEA-V-JMahmud.ppt

Adding semantic tags to the page in order to allow the context browser to pick out the meaning of the items on the page (e.g. item list, item). [how does this relate to microformats?, DIAL?]

Capturing relevant content on a sequence of browsing.

Prototype mobile (PocketPC) browser called CMo - context-driven mobile browsing.

Trying to solve problems -- making browsing faster, easier, decreasing scrolling, etc...
Now trying to make a web transactional system using the notion of context.  Using contextual browsing with domain-dependent knowledge-base.

Their analysis indicates that interaction time is greatly reduced because relevant information is presented immediately.

Algorithms used heavily depend on structure and organization of the page.  Mostly 80-90% accuracy was found.

Looking to extend minimo mobile browser with CMo functionality.

Question: What about failures?

We keep the state and are able to recover from failures in the client in the latest version.

Question: What about page weights (comparing filtered / unfiltered)?

Yes, we are filtering out advertisements, for example. [this could be an issue with this model]

Question: Why are you porting all the server side stuff to the mobile browser?

We are trying to make it compact so you don't need to have huge database/knowledge base.

Yael Schwartzman
http://www.research.att.com/~rjana/mobea2007/ppt/MobEA_Schwartzman(3).ppt

Presenting use of CAM system in Guatemalan highlands.

Asobagri organization

"Internal inspection" - to control and maintain external certifications (such as Fairtrade, organic, etc...)
CAM-RANDI system uses 2-d bar-codes on a booklet with a mobile phone.

Challenges of free text entry meant they had to convert free text entry questions to multiple choice.

Results: about the same amount of time. 3 out of 4 inspectors found CAM as easy or easier than paper forms. Advantage: provide audio and image evidence, lighter carrying load, rugged cam booklet, easier to talk to producers. They used CAM as a mediator between themselves and the producers [interesting parallel to Ken Banks's discussion of how using a microphone in a mobile phone lowered barriers...]

CAM-DPS: Delivery Processing System
Again using 2-d bar-codes

Error rates went down by 50% using mobile entry (CAM) vs. using excel.
Can be used without consistent power. Can be used in mobile settings. Audio prompts can reduce errors.

Mobile technology used in this way helps farmers making them more competitive.

Architecture: bar-codes open an appropriate application on the phone.

Question: What are the operational aspects? How easy is it to deploy? Are they thinking of making it widely available?

CAM Browser is very localizable. The operations -- it took about a month to do it for this trial.
It will be open source.

Julia Kuck, Universität Bonn

Web Service & service discovery in developing regions.

Sharing computing power and information -- executing complex computing tasks remotely.

How to find adaquate Web Services?

Scenarios for the developing world -- health care and e-learning.

User context - split into static context and dynamic context information.
Privacy issues -- full context never sent to the server -- user can choose which context information gets sent.

"Client development on mobile device emulators"
Mobile browser can trigger web service discovery.

"Enabling user collaboration within different situations" through feedback.

Guillermo Esteve
Vodafone R&D Spain

Which devices will enable access to the Internet in these regions?

TRIZ - G.S. Altshuller - theory for problem solving - "Systematic Innovation"

Generating ideas about the perfect device to access the Web in the developing countries.
Contradictions: mobile devices are costly. ADSL chipset is $6, WiFi is $11, GSM is $33 (royalties $17), Bluetooth $6, UMTS $54

Topics brainstorm brought to the table:
Software-defined radios
Combined phone and walkie-talkie -- if phone is 200m away why can't it go straight
Speech to text, text to speech
No screen, just voice
Babelfish -- translation
Separate communication hardware from the mobile device
Wireless charging
Solar charging
Foldable screens
Using gaming consoles as a screen
UNL as a language-independent format
Remember your conversations -- such as sending mail
Phone that only works in your hand (biometrics)
Lego phone -- components based -- what you need on that day
Phone that can capture smell and taste -- if water is drinkable, crop doctor, etc...
Matching buyers and sellers in a market -- ebay model around that
Rugged mobile phone
Self-elevating phone

Best device right now is a PC with WiFi + satellite
But -- it's not scalable

"Phone as a computer."  -- Phone connected to TV with a keyboard -- in the range of $80 - $60 + storage to produce. ARM11.

http://www.flickr.com/photos/torgo/490423640/

Mohammed Jubaer Arif (Univ. of Melbourne)
http://www.research.att.com/~rjana/mobea2007/ppt/SOVoIP.ppt

[Video of Bangladesh TV ad showing how mobile phone is used by fisherman to find best market price.]

VoIP solutions for developing world.

Current VoIP solutions not interoperable.

SOVoIP solution -- "A universal connectivity for VoIP"

Client-side switching between different VoIP systems...

Final Session: Round Table Discussion around Key Issues

Context, user context, session context -- personal context, privacy issues, cultural context
Access to information - device-agnostic
Contribution of information

User-generated context, but different from "You-Tube"

Information that is of use to .001% of the population? (Does this map onto the "long tail" idea?)
No different than having traffic information available on the Web.
Price of corn in one place is necessary to a small number of people but aggregate of price of corn in a number of places is highly valuable.

Socio-economic impact of that information to that territory, region, etc...

What is the "ripple effect" of that kind of information (as in the fisherman example)

Where the information will come from is about the intent.

Tradenet has just started in West Africa -- are they recruiting people to provide that information?

If someone is getting paid to create that information then the subsidy of the network or device comes back to the people who find that information valuable. Alternative model for mobile network subsidization.

People are not aware of these solutions being available. Word of mouth / viral is important. Education is a factor but trust is also important -- people need to trust the solution.

Philippino farmer online marketplace -- linked to an Internet café franchise side-business.

How do you build trust -- you have to talk to the community leaders. You have to sell it to those people before you can sell it to the grass roots guys.

"It's easier to develop a technology than to deploy it."

People adopt technology out of fear of being left behind.

Peer group adoption drives adoption -- viral.

Perception of cost of use...

Fraud management -- in any system there are fraudsters. [reputation system a-la-ebay]

All you need is one bad experience/person to muck up the system. Identification could play a key role (through the mobile?) But mobiles can be shared.

Other information types:
Education -- social impact -- health and medical -- eGovernment

If you give incentives to the knowledge base. Reward program -- currency doesn't have to be cash. Commercialization and business models may have to be completely different in developing regions.

Situation in developing regions:
Infrastructure isn't there -- landlines, etc... Word of mouth flies very fast
One criticism of finding market prices -- if the farmer finds a better price but there is no road to get there then it doesn't help. But just building awareness is very powerful (knowledge of price disparity can drive prices down.)

Technology has improved roads in one way -- corruption is sometimes a problem. An anti-corruption law has required you to provide info of how money was sent. Many govt. agencies have started to have web sites to show how money was spent -- transparency.

In one village there were no roads and villagers offered their sweat-equity.

Citizen-journalism -- mobile phone blogging. Examples from St. Petersburg, Lebanon of use of mobiles and mobile blogging as a political activism tool and a journalism tool.

Market prices in a rural area - could be integrated into mobile payment scheme? Yes, but this is a long way off because price is "set by how you shake hands."

A lot of people in the world that "do not exist" -- not in any census. The more the Internet can provide accountability for these people the better.

What about the "voice web"?  More recent phones (e.g. N70) don't have support for multiple languages -- this is a retrograde step. So language is still a big issue. Education of the industry itself is required.

Text to speech is happening (e.g. voicemail). Multi-language in same message.

Multi-lingual is extremely important, particularly in India -- English doesn't cut it.

Voice files -- sound files -- stored by themselves instead of (or in addition to) transcoded. Similar problem with determining what people say in different dialects.

Collaboration is a key factor.

Definition of the Mobile Web -- what are the parameters, what are the possibilities?

Last words:
Brain power going into this is impressive.
Collect the different business models and ideas.
Hope to collaborate and bring a device or new thing before next year.
Keep it simple, end use of it is going to be simple. Complex system won't be used.
Importance of understanding the user context. Not just technology for the sake of.
"In developing countries" should be bigger than "Mobile web."
Enabling any business  -- if the access is expensive then it won't be a viable solution. Carriers need to listen.
Collaboration and keeping the end user in mind from the beginning is the only way to succeed. Looking for collaborators.
You have to be there.
Huge disconnect between what industry is saying what we're saying. How can we get larger engagement from industry?
Cultural context is key - working with GSMA, etc... innovative business models
Collaboration needs to be driven
Economic uplift, social uplift and transparency
Replicability -- things need to be able to be replicated

Motto of Israeli high-tech industry: without a target, you don't move forward

Collaboration note: we should leverage NGOs, GSMA, W3C, etc... NGOs have a trust factor.
"BCT" Bhagvatula Charitable trust-- first non-profit to ever start in India -- http://www.bctindia.net/

"FAHAMU" -- forefront of using mobiles to report abuse in Africa, etc... Organizing a workshop in Kenya.
dotMobi MAG wants to do this and wants to convene a round table

---
In December there is a ICT Development conference in (ICD4D) in Bangalore. December 16-18th. Could organize a workshop or session.