Transcript A review of the Internet Society study: Promoting Local

Demand-side stimulation
Rohan Samarajiva
Nagarkot, March 28-31, 2015
Locally relevant content leads to
increased use of the Internet
But, it needs to be accessible quickly and
cheaply
In recent times
• More attention has been given to create
locally relevant content (Part 2)
• Less attention to the supporting infrastructure
Content providers lack incentive to
host locally
• Cheaper to host overseas (US / EU)
• The Internet makes “virtual” a reality (content does not
HAVE to be hosted in close proximity to its users)
However,
• Access becomes costly (E.g. Int’l bandwidth costs
passed on to users)
• Latency is higher
impacts
Adoption / use
Similar to the case of Europe preliberalisation
• Historically content was hosted in the US
• Before telecom markets were liberalised, hosting
services within the EU was expensive
• Traffic was routed via the US
As traffic and locally relevant content increased;
• IXPs emerged (reduced latency & cost), becoming
large hubs  content was hosted closer to the
hubs
Similar impacts have been observed in Kenya with the introduction of KIXP and access to
Google Global Cache
Case study
Rwanda
• Landlocked
• Dominated by mountains
in the west and savanna to
the east
• Area 26,338 sq km
• Population 10.7 million
(2011 est)
• Median age 18.5 yrs
• Pop density 407 (highest in
Africa)
• GDP per capita 698 USD
• Gini (high) 50.8
• HDI (low) .0506 (151st)
Legal/ Policy Dimension
• Vision: Knowledge based economy by 2020
• National ICT Plan (NICI - 2015) was introduced in 2011 and is specifically aimed at
increasing access to and uptake of Internet services across the country
• Universal Access Fund (UAF), specifically aimed at facilitating access to telecoms
services in rural areas
• Additional tax on telecoms operators
• Subsidize Internet services
• Subsidies of mobile handsets
• Government separately contributes to a “One Laptop per Child” project.
Economic Dimension
In Rwanda there is a large negative externality associated with
overseas hosting of local content
LOCAL HOSTING
OVERSEAS HOSTING
Hosting Costs
(Size of website: 8 GB)
USD 261 (annual)
USD 149.99 (annual)
Cost of importing traffic
(9 Mbps)
-
USD 13,500 (annual)
Average Latency
10 ms
350 ms
- Passed on to ISPs and finally to end-users, for an annual saving of USD 111
The websites that generate less traffic
are much worse off
• Approximately 8 content aggregators
operating in Rwanda
• For over 800 small websites (all hosted
overseas)
• About 80% of the traffic flows back in to
Rwanda ~ 1 Mbps
Hosted in-country is too expensive at present; However, traffic is limited (partly)
because of high latency.  Cost-effective local hosting is needed.
Invention: Google’s Global Cache &
Akamai’s local cache cluster in Rwanda
GGC, containing largely YouTube videos was made available to all ISPs connected to
RINEX in July/August 2013
Analysis
Page Load Time as Round Trip Time
Decreases (Ilya Grigorik, 2012)
• During congested evening hours,
researchers measured peak latencies up to
800-1000ms in Rwanda
Number of game players and average latency (Source:
GamersNights, 2014)
• Liquid Telecom offered to
host their server
• As the latency decreased,
the number of players
increased
PART 2: FOSTERING INNOVATION
How do you get content & apps that
are attractive to Nepalis
• Web content in local languages/of local
relevance
• Most people will access the Internet through
wirelessly connected devices
– Apps become relevant
• How to encourage more, better content
services and apps
Revenue sources
•
•
•
•
In-app advertisements
Pay per download of content or app
In-app purchases
Upgrading free apps to more feature-rich ones
for a fee (freemium), and
• Subscriptions
Payment facility
• Premium SMS
– Ideal because it reaches all mobile subscribers regardless of phone
type
– However, challenging because MNO revenue shares can be very
high
– Nevertheless, rates are negotiable and MNOs are beginning to
realize that high revenue shares are slowing market growth
– Increasing smartphone penetration and mobile money adoption
are rapidly creating alternative payment facilities, outside MNO
control
• Airtime transfers
• Mobile money
• Bank-based payments such as electronic transfers and checks, or debit
and credit cards
Distribution channels
• Availability of an app is dependent on its
distribution network
– Through app stores
– Mobile operators—e.g., through unstructured
supplementary service data (USSD) services
– Through local distribution networks
Incubators & hubs
• E.g., what Ooredoo is doing in Myanmar
– Big bar camps to identify potential
innovators/entrepreneurs
– Selected entrepreneurs are provided work
locations where they can interact with each other
• Provided mentoring so they can develop products &
most importantly viable business plans
• Provided guidance by experienced entrepreneurs
– Possibly matchmaking with angel investors?
Which ones work?
• Or all?
• Any that have been missed?
References
• http://www.infodev.org/infodevfiles/mobile_apps_at_the_base_of_the_pyra
mid_summary_report.pdf
• http://lirneasia.net/wpcontent/uploads/2015/03/Promoting-LocalContent-Hosting-to-Develop-the-InternetEcosystem.pdf