Published on October 31, 2007
Lifeline:A Knowledge Network Spanning the Rural Digital Divide: Lifeline: A Knowledge Network Spanning the Rural Digital Divide Workshop on Knowledge Networks FAO Headquarters Rome, Italy Feb. 8-10, 2005 Contact: [email protected], [email protected], [email protected] The Vision: The Vision Reduce hunger and food insecurity …by providing direct access to knowledge and know-how to the people who need it, i.e., farmers in villages Create a global infrastructure that connects end-users with existing knowledge and experts Providing access to essential knowledge Eliminate gatekeepers to knowledge Reduce asymmetric information Create efficiency and transparency in the supply chain Improve farmer’s share in the value chain Make information intrinsically available Social empowerment and accountability Access Data Information Knowledge Action Opportunities February 2005 Workshop: Consultative Workshop: February 2005 Workshop: Consultative Workshop Format: Bring together experts from academia, industry, and FAO to study the feasibility of creating a global infrastructure for directly linking end-users to experts How would it work (stakeholders and design)? What would it take (effort and expenditure)? Will it be sustainable? Workshop Goal: Create a plan of action to realize the Lifeline knowledge network Design a system architecture for Lifeline Determine a roadmap for deploying Lifeline R&D required Indicators for success Pilot projects/trials proposals, extending to global deployment Knowledge Networks Stakeholders and their Needs: Knowledge Networks Stakeholders and their Needs Rural Communities – individuals and representatives Learn about and get better prices for inputs (fertilizer, seeds, pesticides, etc.) and outputs (agro-products) Reduce transaction costs significantly Reduce damages from catastrophic conditions (weather, pests, etc.) Rural Service Providers – public and private Increase supplier and consumer bases Improve supply-chain management Governments Create rural information networks whose value and utility extends beyond agriculture Data collection Information dissemination, etc. “Bill of Rights” for the Information Society: “Bill of Rights” for the Information Society Getting the right information to the right people, in the right timeframe, in the right language, with the right granularity - Jaime Carbonnel (c. 1998) Farmers typically unable to tap into even Google-like sources of information Often illiterate or functionally illiterate Lack connectivity, especially data connectivity Rarely have relevant information accessible Internet and existing databases/portals give too much raw (and unverified) information Achieving the “Bill of Rights”: Achieving the “Bill of Rights” Right Information Rural users ask Lifeline network any question they want Right People Sharing relevant information with all potentially impacted people, not just those who ask questions Right Language Spoken language interaction in all languages “Knowledge Officers” for collecting, transcribing, translating, analyzing, creating, indexing, and disseminating information Right Timeframe Response within minutes if answer exists in the database or within a few hours to days if it requires expert input Right Granularity Synthesize an actionable summary from multiple documents as an “encyclopedia on demand” Bill of Rights can only be exercised if Lifeline network is accessible and affordable Example Applications of Lifeline Network: Example Applications of Lifeline Network Hierarchical structure spanning districts, regions, countries, etc. Outside experts interact with higher level Knowledge Officers Builds up an ever-increasing multimedia database Can provide static (e.g., best-practices) as well as dynamic (e.g., weather, prices, etc.) information Innovative mechanisms and processes for information digitization, exchange, analysis, and dissemination Must also strengthen local/neighboring and peer-to-peer interactions Farmers themselves are often the best source of locally relevant knowledge An illiterate farmer goes to a Village Knowledge Officer (with a computer connected to a multimedia database) and asks a question in his or her local language The KO retrieves answer from local Multilingual database within minutes 80 - 90% of the time For the remaining 10 - 20% of the time the KO puts up the question to a higher level office and gets an answer back, typically in less than 24 hrs 100s of domain experts populate the databases, both as part of their jobs and as volunteers (say, 2 questions per week) Multi-level Information Flow - An archetypal scenario Knowledge officers and Domain Experts: AV data collection, Transliteration and Transcription Information Retrieval Translation, Information Retrieval Verification of Query-Answer Relevance And RFP to domain experts Knowledge Management & Coordination (national level) Domain experts: Volunteer to answer at least 2 questions a week (or part of job responsibility) District State Knowledge officers and Domain Experts Knowledge Management & Coordination (global) World Nation Expertise of Knowledge Officers Village Roles of Knowledge Officers: District Region/Nation (sub)continent Records question of the end-user in audio-video format. Enters text transcription of the question. Searches local language database for answer Need not be knowledgeable in English. Enters translation of questions. Searches multilingual database for answer Sends answer after translation to lower level If question not among FAQs or automated system, sends to higher level Picks questions of critical nature and validates the answer provided at lower level If critical or unanswered question, puts up request to experts even if not paid for by end-user Same as next level up, but with the range of analyses broadened to the region/subcontinent level 3,000 people Transcription (and possibly Transliteration) 300,000 people Translation and Information Retrieval 0.3B people Knowledge Management & Coordination Roles of Knowledge Officers 30M people Verification & RFP from Experts Global Brings experts to where their knowledge is needed. Mobilization of resources towards their need. Identifies and triggers initiatives to control “epidemic”-like problems 3 Billion people Knowledge Analysis and Inference Village (All numbers shown are for rural, developing country populations = beneficiaries) Design Assumptions: Design Assumptions End-users interact at Village Knowledge Centers Equipped with a networked computer and basic A/V equipment Staffed by a Knowledge Officer Humans are intrinsic to Knowledge Networks (raw information knowledge!) Domain experts provide answers to previously unanswered questions Many answers converted into an “encyclopedia-on-demand” video documentary at higher-level centers centers and dubbed into local languages Lifeline Network also available for direct access browsing by literate and networked users Sample Workflow: State/Nation KO Regional and Global KO District KO Village KO Sample Workflow Number of Knowledge Officers: Assume One Level-1 Knowledge Officer per village (cluster) 1 Min of video allowed per question. = ~ 5 min. of prep/processing time by KO1 = ~12 Questions per hour Only 15% are unanswerable, and forwarded to KO2 Requisite higher order KOs required to answer the questions at each step of the hierarchy Assumes 85-90% hit rates (being able to answer the question without sending the question up Assumes 5 – 10 minutes per question total time Repeats at each hierarchy Factors in seasonality, working hours, etc. Information stored as a combination of text, graphics, audio, and video Corresponds to tens of thousands of TeraBytes (TB) of storage per year Number of Knowledge Officers Global 20,000,000 people 1,000 people 400,000 people 0.5B people 3B people Number of officers needed at different levels Village Business Model: Business Model Model viability based on Per question costs ~ 20 cents (net) Can be <10 cents depending on assumptions such as hit (success) rate, ability to synergize infrastructure, etc. Marginal costs are much lower 6 questions per capita (rural, developing country) per year, or 0.6 questions per family per week 1 million rural Knowledge Centers established Less than one per village Includes all the remote hardware, including telecom systems Network could be utilized for other development needs as well! Would be adding billions to the rural economy, but only costing a little over $1/year per person Total Budget (Annual) indicative—steady state conditions: Total Budget (Annual) indicative—steady state conditions Scaling Options: Scaling Options Distributed deployment - grows over time Begin with “low hanging fruit” Selection by geographic coverage Selection by socio-economic coverage Selection by infrastructure availability (telecom, power, etc.) Can begin with one village per district first Initial deployment will have limited answers in the database Will impact economics during “teething period” Actual villages vary in size (distribution) “How many unconnected villages are there?” What’s a village? How many are there, of what size? What connectivity is or isn’t available? Assume min. 3,000 people (or other threshold) required to have a Knowledge Center Lose a small percentage of the population but a large number of villages – work in clusters of villages instead Issues: Hardware/System : Issues: Hardware/System Lack of user literacy let alone e-literacy! Robustness: Repairs/warranty/operating conditions and physical security Even a simple water pump lies idle in parts of the world due to lack of parts/maintenance Telecom availability Bandwidth is key to multimedia applications Even affects system (database) design Electricity availability Else requires standalone power (expensive) Do the technology solutions (language, interface, expert systems, multimedia searchable databases, etc.) exist? Will take several years to build up appropriate vocabularies, syntax models, etc. Seeking perfection should not delay good enough What database(s) structure is optimal and multimedia searchable? Distributed? Unstructured? Issues: Business and Policy: Issues: Business and Policy Financial Viability Average numbers may be misleading Some big farmers (or agribusiness) contribute most to the output How do we determine user and operating priorities? Willingness to pay? Timeliness? Importance – as per whom? Intellectual Property issues Risk of amplification of existing divides Issues: “Trust”: Issues: “Trust” What is a “wrong” answer? How do we know or find out if it was wrong? Subjective answers are harder to monitor than objective (“numbers”) ones What is done about it? Issues of liability and recourse Danger of monopolies, misinformation, etc. How to oversee the KOs? How do we train the KOs? How to sensitize and “train” end-users? Incomplete or incorrect questions will lead to similarly problematic answers System must become incorporated into user decision-making More Than Just Internet Connectivity: More Than Just Internet Connectivity This is infrastructure plus information “Bill of Rights” Potential additional uses of Lifeline network Other development agencies and institutions E.g., Health Internetwork (WHO) today Doesn’t include access Doesn’t cater to the rural poor Public and private users FAO should not wait for them (or the “market”) Market Mechanisms Don’t Always Work: Market Mechanisms Don’t Always Work Some infrastructure requires governmental push E.g., Rural Electrification or the National Highway System in the US Developing countries need concerted (global) effort Rich get richer in a global economy Lifeline network levels the playing field Population still growing in developing countries with less land available and increasing deterioration of soil productivity The longer we wait, the harder the challenge Conclusions and Actions: Conclusions and Actions At an annual cost of ~$1/rural capita, creating a Lifeline knowledge network is an optimal use of global capital Stakeholders and countries must agree on a deployment plan R&D Pilots and demonstration (100 deployments) Selected regions and geographies (10,000 deployments) Increasing densities (100,000 deployments) Global (1,000,000 deployments) Minimum penetration target (% by year 20_ _): “20 by '10, 50 by '15, 100 by '20” Detailed information available in the spreadsheet…: Detailed information available in the spreadsheet… Thank you!: Thank you!