Shikkhok.com - An Altruist-built,Ultra-Cheap MOOC for Pennies:Building an Open Content Education site for RuralSouth Asian Studentsby Ragib Hasan, University ofAlabama at Birmingham
Shikkhok.com – an Ultra-cheapMOOC for Rural South AsianStudentsWinner of 2013 Google RISE Awardby Ragib Hasan, University of Alabama atBirmingham firstname.lastname@example.orgKeywords: Education, MOOC, IT4DIndustry/Category: Education, e-Learning
1.A Real World Project with high impactLow-income and rural students in South Asia with limited knowledge ofEnglish do not have access to quality education.How can we provide top-quality education at a very low cost tothe millions of students in rural Bangladesh and India?•Goal: To design Shikkhok.com - a free Massive Online OpenContent (MOOC) education site at an ultra-low cost , specifically forrural underprivileged students•To allow students with No Internet Access to be able to benefit fromShikkhok’s educational content•Development Timeline: Summer 2012 - present
(pronounced Shik-khok), is aBengali language word that means,literally,“One who teaches”
2. Stake holdersDr. Ragib Hasan, founder and developer of Shikkhok.com,Assistant Professor of Computer Science,University of Alabama at BirminghamVolunteer educators:Researchers, Educators,and Scientists spread allover the worldBengali speaking students in all levels inBangladesh and the State of West Bengal(India), with-Limited English proficiency-Limited access to higher education-Living in geographically remote villages
2. Stakeholders – how they benefitStudents in rural Bangladesh and West Bengal (India) will benefit from the Shikkhok.comproject because• Students in rural schools do not have access to quality study material or teachers,and lag behind students in cities• There are not enough openings for higher education in colleges and universities, somany students drop out after school• Education in non-public institutions is expensive – not all students can afford it dueto poverty• State of the art topics are hardly covered by many educational institutions• Existing e-learning projects are based almost invariably on English language, and thelanguage barrier prevents students from benefitting from them.• Finally, almost all (if not all) the MOOCs use the Internet as their deliverymechanism, which may not work for rural students. Shikkhok provides an alternatesupply chain mechanism to deliver it’s teachings.
2. Stakeholders - Background• Bengali is the 4thlargest language in terms of native speakers(250-300 million speakers in Bangladesh and India)• Students in rural areas often do not have access to qualityteachers, books, or good schools.• Higher education opportunities and content is scarce inBangladesh and India• Only 50,000 opening in Bangladeshi universities and collegesfor incoming freshmen, while there are more than 300,000eligible students• Many students drop out due to lack of cheap highereducation opportunities or extreme poverty
2. Stakeholders - Background: InformationTechnology to the rescue …• While regular computing devices are notcommon/affordable in rural areas, Mobile phones andhence Mobile internet have significantly high penetration inBangladesh, even in rural areas (100 million mobilesubscribers as of early 2013, in a 160 million population)• A mobile-optimized Bengali language MOOC can serve asan alternative education platform for rural and non-traditional students• And an innovative non-Internet based delivery mechanismcan allow rural students with no internet access to get highquality education
2. Stakeholders - Existing MOOCs arenot enough• Coursera.com has 208 courses,ALL provided in English language• The Khan Academy’s excellent online educational videos are also inEnglish• Unfortunately, Bengali translation of Khan Academy’s videos are notpopular among the students in Bangladesh and India (most videolessons have an average of only 100-120 views in 1 year. Example:http://www.youtube.com/playlist?list=PL58BD1F917975C9BE).• Anecdotal reasons include mismatch between the lessons andacademic syllabus in Bangladesh/India, cultural mismatch/”lost intranslation”/artificial and literal translation• As a comparison, Shikkhok.com’s Culinary arts course videos received an average of 300+ views within 1week of publication (https://vimeo.com/user14642276/videos/sort:plays/format:thumbnail)
3. Project Requirements• Education medium must be in Bengali• Content must be highly optimized for mobile phonebrowsers with limited and slow data plans• Lessons must be short, include both text and multimedia,and have easy-to-use student registration, feedback, andevaluation schemes• Must be highly-available, low access times even inBangladesh and India• Must be designed, delivered, and publicized at a very lowcost, and provided to students for free• Must not depend on the Internet to deliver content.
3. Project Requirements – Identifying theChallenges• Technical: Identifying the best design principles• Team: Organizing and coordinating a distributed team• Stakeholder: Getting effective feedback and attentioninformation from the users
3. Project Requirements – Identifying theChallenges?• Cost: Popular MOOCs such as Coursera.comhave millions of dollars in venture capital funding.• Coursera itself has $22 million funding• Such funding is unlikely for educating ruralstudents in Bangladesh and India• Marketing/advertising such a site to the masses isalso expensive.
3. Project Requirements – Identifying theChallenges?• Overcoming the language barrier: Studentswith limited English language proficiency cannotutilize existing MOOCs such as edX, udacity, orCoursera, so how do we ensure maximum impactfor such students?• Finding teachers: How to gather teachers withthe right expertise and technical know-how?• Reaching stakeholders: How to publicize anddeploy content to the intended audience?
4. Our Solution• Explore Human Computer Interactionprinciples and methods to effectively reach therural students• Take extreme penny-pinching measures todevelop the platform at a low cost• Use social media marketing strategies topublicize the service to the target audience• Utilize non-Internet based supply chains todeliver content to the rural students
4. Solutions - Methods usedDesign– Use an iterative model for creating the most effectiveuser interface which has to be mobile friendly, less-graphics intensive, and suitable for both smart and non-smart cell-phone browsing– Follow a User Centric Design methodology byconstantly evaluating user responses to lessons andmodifying teaching tools accordingly
4. Solutions - Methods usedDevelopment– Use rapid prototyping and design methods to developcourses (lessons and lectures augmented per userfeedback and view counts)– Use ultra-low cost and open source tools in acrowdsourced model– Use Social Media marketing for free, leverage thepower of cloud to distribute content
4. Solutions - Methods usedEvaluation:– For evaluation of lecture style and content, measureuser responsiveness and attention span for each lecture(use webpage stats to calculate how long users stayedat each lecture page, how many users came back toview further lectures, i.e. user retention)– Measure user engagement by correlating lectureviews with participation in quizzes associated withlectures
4. Solutions - (Ultra-cheaply) DesigningShikkhok.com• Over summer 2012, we rapidly developed Shikkhok.complatform• Total development cost: only US $15.00• Total number of registered students = 20,000• That is, cost per registered student = US $0.00075 only!• Total number of courses designed = 25• 4500 lecture views per day, from 3000 unique visitors
4. Solutions - (Ultra-cheaply) Designing Shikkhok.com• To minimize development costs –• Adapted open source CMS to provide authoring platform• Mobile-optimized front end• Host all media/videos on free online repositories such asYoutube, Dropbox, imgur• Use Google forms and embedded scripts to automate userregistration and MCQ quiz processing• Cost: Domain name: $5/year, 100 MB low-cost host: $10/year(Development (mostly wordpress theme tweaking) done by one volunteer for free)
4. Solutions - (Ultra-cheaply) DesigningShikkhok.com• Site design and graphics: Crowdsourced viaSocial network contacts (received 5 submissionfrom a volunteer within a few hours of request onFacebook)
4. Solutions - Social Media is extremelyeffective Getting content and volunteersTo gather a team of volunteer teachers:• Posted a request on Facebook• 10 volunteers signed up in 1 day• Two courses were developed by day 2• By week 2, 5 courses were running• By week 8, 15 courses were started• By month 8, 25 courses running, with 5 coursescompleted
4. Solutions - The power of Clouds foreasy deployment• Used Amazon’s CloudFront Content DistributionNetwork (CDN) to cache and distribute content,and to reduce bandwidth costs• CloudFront improved loading times in South Asiaby 50-60%• Only incurred $0.20 in two months for deliveringcontent to 50,000+ unique visitors across theworld
4. Solutions - Design principles andstrategies for online education via amobile phonePlain text (not multimedia) is still the king of content• Users of mobile phones have to pay per-KB, soless images is better• For videos, youtube based low-res streams anddownloadable 3gp formats work the best
4. Solutions:A Innovative DistributionChannel• A major challenge was to create a non-Internetbased distribution channel to reach rural studentswithout Internet access• Solution: Develop innovative distribution channels.
4. Solution: Innovative DistributionChannelsOur Approach:Approach 1:• Create short 3gp version videos; put a collection ofcourses on USB sticks, give out to phonevendors/shops in rural bazaars.• Students visiting the bazaars can load the videos ontheir phones for free or for a nominal fee (charged bythe vendors, not us)• (We found this model to be very useful, as rural bazaarphone shops are already used as a distribution hub formusic videos/songs, and people are used to going thereto load videos on their phones)
4. Solution: Innovative DistributionChannelsApproach II• Use ultra-cheap Raspberry PI computers• Each Pi costs only $35.• We put a large number of courses on SD cards on each PI,add a donated keyboard, mouse, and ship this to ruralschools. (No internet needed, we preload everything onthe SD cards, and make a kiosk-like interface easy for evennon-computer users)• The schools can hook the Pis directly with regular TVs, andhave the video lectures delivered to students
4. Solutions - User engagement strategiesthat work …To engage users in easy discussion, integration withexisting social networks is the best strategy:• Using wordpress native commenting: about2/3 comments per lecture• Using Facebook comments: at least 30 “like”and 5-10 comments, questions per lecture
4. Solutions - Marketing strategies:utilizing social media• Social media based “free” marketing campaigns worked verywell• Did not use regular advertisements, rather usedFacebook and Twitter to publicize Shikkhok• Got 3000 fans on its Facebook page within a few days• Each lecture announcement is viewed approx. by 4200people within one hour or so (stats via FB Insight)• Total fans as of May 1, 2013: 9000
5. Our Contributions• We demonstrated that localized strategies work betterthan globalized universal MOOCs (local language basedand cultural context-aware content is more effective)• Unlike Khan Academy Bangla, we did not translateexisting MOOCs, rather developed localizedcontent from scratch, which turned out to be moreuseful to students. (our video lectures viewed manytimes more than the translated content)
5. Our Contributions• We developed a set of tried-and-tested design principlesfor educational content delivery over mobile internet torural students• Evaluated various site design and lecture content todetermine the best possible strategy and contentformats that serve the mobile-internet-using ruralstudents
5. Our Contributions• Our user centric design and constantfeedback/evaluation loops allowed us to detectstrategies that work (mobile optimized video,Facebook Integration) and that do not work (e.g.live sessions with teachers using Google HangOut)• Constant user engagement strategy allowed us toimprove our lecture content (lectures with loweruser retention/attention span are re-written/developed)
5. Our Contributions• Our biggest contribution is the generalized set ofdesign and evaluation principles for thedevelopment of a localized micro-lesson modelthat can be effectively used by e-learning systems inother languages in other parts of the developingworld.
5. Our Contributions - Results – some stats• Since it’s start on August 1, 2012, Shikkhok.com has• 25 online courses on diverse topics such asBioinformatics, Neuroscience, Computer Programming,Finance 101, Calculus, Cloud Computing• Total number of students registered for all courses:20,000 (actual student count larger since registrationisn’t mandatory)• The Computer Security101 course alone has 2000registered students• Total number of quizzes/tests taken: 30,000
5. Our Contributions - Results – somestats• Total unique visitor count in in 8 months: 369,108• Total lecture views in 8 montsh: 1.1 million• 80% visitors are from rural Bangladesh, using mobilephone browsers• Shikkhok.com is getting 2000-3000 unique visitors a day
Visitor data as for 2012 (Q3, Q4) and2013 (Q1, and ongoing Q2)
5. Our Contributions - Results, that matter• Shikkhok.com is the first e-learning MOOC site inBengali language, completely free and open foreveryone• Students from rural Bangladesh and India regularlycontact the founder to express their satisfaction:• “I wanted to study Computer Science, but had todrop out of school due to poverty. Shikkhok.comhas given me the chance to enter the wonderfulworld of computer science once again” –testimony from a user from Jamalpur, Bangladesh
AwardsFrom Google’s RISE award citation:•http://www.google.com/edu/rise/recipients.html“The Shikkhok.com project aims to provide free onlineeducation and high-quality courses in Bengali language to ruraland disadvantaged students in Bangladesh and India.Theybring together educators and researchers from all over theworld for to create content in Bengali language on both basicand advanced topics, to develop a model for ultra-low-costonline education for students in the developing world and toserve as an open, free, and cheaper alternative to traditionaleducational institutions.”
Future goals• To create a complete set of courses for grade 6-10 ofBangladeshi school curriculum• Project timeline: Summer 2013• Teams of teachers already formed in April 2013• Technical content development begins from May 2013• Content distribution and pilot studies in several Bangladeshischools: September-October 2013.• To create a complete set of courses for grade 11-12 ofBangladeshi highschool and college curriculum (Fall-winter 2013)• Reach at least 100,000 students and 100 schools by the end of2013
6. Summary:What did we learn fromShikkhok.com?• It is possible to design successful MOOC e-learning sites at ultra-cheap cost via an altruistic volunteer model (Shikkhok cost only $15to develop and deploy compared to $22 million for Coursera)• Attention to HCI design principles such as user centric design canallow better retention of users and improved attention to content• To reach rural students, focus should be more on non-Internetbased textual content designed for low-bandwidth mobile phonebrowsers• Localized, native language education is more successful than the one-course-fits-all approach by many well-known MOOC sites
To view Shikkhok.com in action• Please visit: http://www.shikkhok.com• Media coverage: Front page news in 10 top Bangladeshi newspapers,including:• Daily Kaler Kantho: http://bit.ly/Rj5Sym• Daily Prothom Alo: http://www.prothom-alo.com/detail/date/2012-09-14/news/289111• Daily Sun (English): http://daily-sun.com/details_Virtual-school-wins-google-award_420_1_19_1_2.html• Covered by Deutsche-Welle – the German International Radio’sBengali language world service http://bit.ly/WXjsyP