Transcript Academy_Infrastructure_Chris Marsden

OII Workshop
14 September 2015
@ChrisTMarsden
Topics include:
PAST AND PRESENT
1.
1.
2.
How did the Internet infrastructure develop?
How are Internet governance decisions made now?
2. NATIONAL/GLOBAL:
1. To what extent can national governments and
supranational bodies control the Internet?
3. LAYERS:
1. How does the underlying structure and markets
behind physical network affect the way we use web?
4. FUTURE:
1. Where will investment come from to build next
generation networks? What are policy implications?
Resources: Current & Future Policy
1.
Brown and Marsden (2013) Regulating Code Chapter 7
1.
https://mitpress.mit.edu/index.php?q=books/regulating-code
2. Mueller, Milton et al (2013)
1. Internet Security and Networked Governance in
International Relations 15 International Studies Review 1
1.
http://dx.doi.org/10.1111/misr.12024 pp86- 104
More polemical & policy oriented
 Evgeny Morozov (2011) Foreign Affairs JANUARY 3
 Freedom.gov - Why Washington's support for online
democracy is the worst thing ever to happen to the
Internet

http://academy.oii.ox.ac.uk/page/infrastructure/#tab-id-2
How are decisions made currently
relating to Internet governance?
 What are the key areas of disagreement?
 What are the European perspectives?
Resources: how governance started
A Prehistory of Internet Governance
1.
Ziewitz, Malte and Brown, Ian, (2011) RESEARCH
HANDBOOK ON GOVERNANCE OF THE INTERNET, at
SSRN: http://ssrn.com/abstract=1844720
1.
Brief History of the Internet
2.
Barry M. Leiner, Vinton G. Cerf, David D. Clark, Robert E.
Kahn, Leonard Kleinrock, Daniel C. Lynch, Jon Postel,
Larry G. Roberts, Stephen Wolff (1998)
1.
1.
http://www.internetsociety.org/internet/what-internet/historyinternet/brief-history-internet
The Framing Years: Policy Fundamentals in the
Internet Design Process, 1969-1979
3.
1.
Braman, Sandra (2010) at http://ssrn.com/abstract=1989650
PAST & PRESENT
Basic features of Internet
architecture and how it is governed
 The Internet was established as a network of networks,
 national borders & centralised oversight?
 Argument around decision-making processes
 and entities involved in its governance.
 Governments have a vital role to play in enabling
 and encouraging development of infrastructure
 Pipes and routers
1962: Paul Baran Packet Switching
 RAND series of project memos –
 resisted internally and by Department of Defense
 Licklider, Pouzin, Kleinrock, Davies, Nelson similar
 Baran left RAND to form Institute for the Future
 RAND tries to rewrite history but never a nethead place
 Disclosure: Senior Analyst at RAND Europe 2005-7


IFTF http://www.iftf.org/future-now/article-detail/iftfmourns-paul-baran/
http://www.iftf.org/our-work/people-technology/the-humaninternet/forecasting-the-internet/
From central to distributed control
Telcos never liked the Internet
 AT&T’s Jack Osterman reacting to Baran:
 ‘First it can’t possibly work, and if it did,
 damned if we are going to allow the
 creation of a competitor to ourselves.’
This is a fake meme – not RAND
18 Original Nodes on Internet 1971
Defence Advanced Research
Projects Agency (DARPA): Network
ISOC: “It was from the RAND study
that the false rumor started”
 “claiming that the ARPANET was somehow related to
building a network resistant to nuclear war.
 “This was never true of the ARPANET,
 only the unrelated RAND study on secure voice
 “Work on Internetting emphasized robustness
survivability, capability to withstand
 losses of large portions of the underlying networks.”
 Still controversial?
1969: social, economic & privacy
problems were predicted
Don’t ever let anyone say:
“If only they had thought about
the problems it would cause”
They had – in the 1960s….
Baran in ARPANET use projections:
“broadband information network”
 On the Impact of New Communications Media upon Social Values
 Law & Contemporary Problems, Vol, 34, No. 2 (1969), pp, 244-254
 “Legislation, Privacy and EDUCOM,”
 Bulletin of the Inter-university Council, Vol. 4, No. 6 (1969), pp, 2-4
 Notes on a Seminar on Future Broad-Band Communications,
 Working Paper WP-1 (February 1970). iii + 21 pp.
 The Future of the Telephone Industry, (1971)
 The Future of Newsprint, 1970-2000, (1971)
 Potential Market Demand for 2-Way Information Services to the
Home (1971)

http://oxfordfutures.sbs.ox.ac.uk/boucher-futures-researchlibrary/publications-of-the-institute-for-the-future/index.html
Utopian/dystopian visions of society
 Alvin Toffler (1970) Future Shock
 post-industrial society “prosumer”: merging of roles of
consumer and producer : The Third Wave, 1980
 John Brunner (1975) Shockwave Rider
 Dystopian vision of how to survive in Orwellian ‘1984’
 Accurately predicts many technologies in 2015: ‘worm’
 Don Tapscott expanded on “prosumption”
 The Digital Economy: Promise and Peril In The Age of
Networked Intelligence (1996)
End-to-end: why is that important?
 Telephone network intelligence lies in the network
 Telephones (relatively) unimportant end-user devices
 Switches and routers vital
 Internet switches and routers even more vital
 But end-user devices absolutely critical
 So it’s not a DUMB network with SMART devices
 It’s a SMART network with VERY SMART DEVICES.
E2E offers no guarantees
 “IP is a connectionless or datagram service,
 providing no end-to-end delivery guarantees.
 “IP datagrams may arrive at the destination host
 damaged, duplicated, out of order, or not at all.
 “The layers above IP are responsible for reliable
delivery service when it is required.”
 Unsurprisingly, network engineers have spent the last
30 years trying to work out how to improve things!
NATIONAL-GLOBAL
 To what extent can national governments and
supranational bodies control the Internet?
How does the underlying structure and markets behind
the physical network affect the way we use the web?
1. Key infrastructure issues for
national governments
2. How are these playing out at present
across Europe?
Regulation: networks, markets, law
 Always remember: everything you do online is regulated
 But not necessarily by government…
 Engineering affects speeds/quality of experience
 Software affects browsing/apps
 Choice of content/apps/services/devices affects ecology of
Internet
 Government can steer these forces but they are everpresent
 Adopt a wider view of what regulation means
 But regulation is always political!
Standards Matter
 National responses risk “fragmenting” the Internet
 But it’s already a network of c.40k networks
 Examples: Great Firewall of China, Iran Halal Internet
 See Regulating Code (2013)
Is there a ‘European’ response?
 Yes: de-Americanisation of root control
 Yes: emphasising competition for
infrastructure
 Yes: multistakeholder dialogue
 Yes: human rights e.g. privacy
 (not 1st Amendment)
 Type of triangulation between US & BRICs
 Especially Russia/China moves in ITU/IGF
 Next IGF in Brazil November 10, 2015.
Lectures: More Policy Responses
Governance
 Professor Helen Margetts
Privacy
 Bendert Zevenbergen October
Security
 Dr Joss Wright October
Content
 Dr Victoria Nash November
Economy
 Professor Jonathan Cave November
LAYERS
 How does the underlying structure and markets behind
physical network affect the way we use web?

 Classic model of network of networks focuses on layers
of protocols
 Think of it like an iceberg – we see the ‘content’ layer
 Much of the “plumbing” takes place further down
Content Sitting Above the Standards ‘Iceberg’
Seven layers?
 That’s not agreed – most agree there are at least four!
 RFC1122 says 4
 Ah but ‘physical’ (hardware) is not in RFC1122
 So that would make 5
 ‘Session’ is divided into 3 by this iceberg
 OSI counts session as session-presentation-applications
 Which would make it 7
Is transport important?
 Only if you want your data to reach the other person!
 Vast RFC1122 is the best description of how it works:
 https://tools.ietf.org/html/rfc1122
 “The transport layer provides end-to-end
communication services for applications.
 “There are two primary transport layer protocols:
 Transmission Control Protocol (TCP)
 User Datagram Protocol (UDP)”
How can a network have no
agreement on layers?
 It’s like nations not agreeing on types of law
 But actually many nations have different constitutions,
different ways to ratify international treaties –
 that’s just how nodes in a network operate
 “We agree to differ”
 As long as everyone externally agrees to follow the
internetwork rules, it’s not that important
 “We can still speak the same language”
 The ‘language’ is the ability to recognise layers
 TCP and IP are the most important
What is ‘content’ and ‘applications’?
 Web pages,
 Apps such as browsers and script running inside




browsers
Driven by advertising, cookies
Personal data
Includes video, audio, P2P file sharing
BUT reliant on deeper transport layers
Oldies may remember a 1995 New
Yorker cartoon…now updated
Infrastructure Rules Affect WWW
Development of Infrastructure
 Remember: the WWW is NOT the whole Internet
 Public IP traffic is also NOT the whole Internet
 Internet: public + private, WWW + IP traffic
Regulatory Rules & Competition
Affect How We Use WWW
FUTURE
Where will investment come from to build next generation
networks and what are the policy implications?
1. should consumers pay more?
2. should we move away from open free web we enjoy?
3. who should decide?
IP backbone costs dropping 14-22%
per annum as traffic grows by same
NO ‘explosion’ in IP traffic ~20%
Internet cost-growth roughly same
since 2002
 Which is (more or less) why Internet access + phone line
costs £20/€27 for basic service
 It used to be £10 phone line and £10 broadband
 Now costed as £16.40 phone line and £3.75 ADSL (UK)
 Of course voters have to rent the ‘phone line’
 Competition determines whether basic service is 1Mbps,
10Mbps, 100Mbps or 1Gbps
 Which are simply the speeds of different access
technologies
 1Mbps ADSL (old school) 10Mbps ADSL2
 100Mbps VDSL (short line) 1Gbps (DPCSIS3 or fibre to home)
 No miracles in 1990s engineering NOT GDP/inflation growth
Why this ‘miraculous’ speed
increase/cost decrease?
 Bandwidth provisioning;
 Fibre capacities reaching Pbps transfer on Internet 2 & NTT tests
 Fibre is sand-water; much cheaper than copper to maintain
 Microprocessors (Moore’s Law)
 Double transistors in 2 years; 200 times better in 15 years

Gordon (2015) “I see Moore’s law dying here in the next decade or so”
 Digital storage capacity (Kryder’s Rate)
 about 15% per year.
 1000 times better in 15 years 1994-2009 (anomaly)
 Network effects (Metcalfe’s Law);
 Or more specifically adaptation to humans n x log[n]
 Note: may not continue forever

https://royalsociety.org/events/2015/05/communication-networks-sm/
But mobile is growing fast!?
 1oEB to 2oEB is 100% growth
 1ZB to 1.1ZB is 10% growth
 But 0.1ZB is still 50x greater than 2EB…
 This is hard for qualitative social scientists such as
lawyers (me) and politicians to understand….
 1,000,000,000,000,000,000,000 bytes is one zettabyte.
 Conclusion: mobile is growing, wifi is growing, but
there is no explosion!
Markets: it’s about the money. Mobile
termination rates UK (Source: Ofcom)
Despite what some might have you believe…
So this smart network listens to us?
 It listens to traffic – either by type or individually
 By type – to monitor health of networks
 Occurs all the time in real-time, generally benevolent
 By content – that’s censorship, as humans understand
 Network neutrality is a principle understood to allow
monitoring by type but not content
 Permits users unimpeded freedom of use of:
 Networks, devices, content AND how access providers
filter
Why would access providers filter?
 Spam and viruses: existed since 1970s
 Invented by BBN to test network – Creeper virus
 1980s – hackers’ viruses appear (Elk + cBrain)
 Access providers filter both (spam email often used as virus
transmitter)
 Peak of spam was late 2000s – majority of email 2003-2015
 Symantec analysis – spam <50% of email July 2015
 http://www.bbc.co.uk/news/technology-33564016
 But note developing countries/webmail often much higher
 But you still don’t want half your inbox as spam?
Bringing us up to date
Cables could – and can - be tapped
Bude – GCHQ signals intelligence
base Cornwall
#TEMPORA: Cables like TAT14
NSA & GCHQ civil war on encryption
1.
Tampering with national standards (NIST specifically)
1.
to promote weak, or otherwise vulnerable cryptography.
2. Influencing standards committees to weaken protocols.
3. Working with hardware and software vendors
1.
weaken encryption and random number generators.
4. Attacking encryption used by 4G phones'.
5. Obtaining cleartext access to 'a major internet peer-to-
peer voice and text communications system' (Skype)
6. Identifying and cracking vulnerable keys
7. Human Intelligence division to infiltrate the global
telecoms industry – essentially bribing employees
8. Decrypting HTTPS/SSL: Yahoo, Google, Hotmail/Outlook
2013: Anti-Surveillance Discontent
 Snowden: Booz Allen’s man in Hawai’i
 Revealed extent of surveillance to 2012
 fled to UK imperial entrepot Hong Kong
 Five Eyes extra-legal surveillance – each spying on the
other’s citizens
 Parliaments, NGOs, IGOs examine Five Eyes
 Especially German Bundestag + European Parliament
 Also Brazilian Senate + others
Summary: understand Internet
past to regulate its future
 Internet c.50 years old
 Inventors understood social & economic potential
 Problems were clear from outset e.g. privacy, security
 Layers interoperate making intervention complex
 Regulated by many stakeholders
 engineers’ standards,
 market actors’ behaviour,
 users’ responses (e.g. privacy information, switching)
 Law (least of all, late but significant response)
 Regulators should understand the potential for
encouraging competition and respect for law (e.g. rights)
To what extent can national
governments control the Internet?
What role do supranational bodies play in its regulation?
2. International Telecommunications Union
3. Think about standards bodies too:
1.
1.
2.
IETF, IEEE, ISOC, W3C, ICANN
Not all US-dominated
4. Market actors global – or at least regional
1.
Microsoft, Google, Facebook, Amazon, Cisco, Intel
5. US and EU powerful players
1.
First amongst equals, or more important than that?
2. Very important regulators of competition & privacy
Questions?
 For Hangout 1-2 October