Reidar Conradi, IDI, NTNU
Reidar.Conradi@idi.ntnu.no, reidar.conradi@yahoo.no
Reidar Conradi, IDI, NTNU, 2 Jan. 2008:
Episodes in and Comments on the Development of Internet
- NB: p.t. not quite complete??!!
Preface
This document started as some personal Comments on Terje Rasmussen's book on Internet
decision making processes (2007), see Appendix X and
[Rasmussen07]. This endevor, however, soon demanded further and
delightful "studies" in odd hours to better acquaint myself with the
Internet TCP/IP protocols and the story behind. My notes from the
last half year therefore constitute the bulk of this document, which
has been uploaded on the Web via my home web-page. Any comments on
contents or form are solicited and appreciated!
0. Table of Contents
1. Introduction
2. ARPA startup and the ARPANET
3. Prelude on Packet Switching: how, why, and by whom?
4. First Wave, 1967-1972: first ARPANET development
5. Years 1965-1972, Comments on openness and independence
6. Second Wave, 1973-1990: New TCP/IP protocol
7. Technical remarks on the new TCP/IP protocol
8. Third Wave, 1991-2000: the World Wide Web (WWW, or just web)
9. Fourth Wave, 2000-now: "New times" for interactive,
distributed multimedia work
10. Some current and future issues and challenges for Internet
11. Literature and References
12. Appendix A: Internet institutions
13. Appendix B: Mini-glossary of IT terms
15. Appendix D: The Technical Computer Revolution
16. Appendix E: Gallery of Persons related to Computer- and
Internet Technology
1. Introduction
The Internet and the services it supports, especially email and the
Web, has made IT essential for most human and societal activities. And
it has happened in only 30-40 years. What can be learned from this
development, seen from a philosophy of science perspective?
For instance, great innovations can never be planned. Just think
of agriculture, the wheel, the alphabet, iron production, typesetting of
books, electricity, telephone, computers, mobile phones, and
similar important innovations. Both the Internet and Web were developed
for other purposes than their actual use today,
respectively, mutual use of computer resources (telnet login service)
and navigation in hypertext-connected,
scientific documents at CERN (the Web).
Indeed, the Internet and Web together
rank as one of mankind's most
applicable and penetrating technologies. New work modes and usage
patterns pop up every day, like net shops and e-businesses,
web-search applications,
and "social computing" phenomena like Open Source, Wikipedia,
Youtube, and Facebook. The difference between information consumers
(watching common TV series) and producers (writing own blogs) is
diminishing. Furthermore, the fusion of Internet and mobile technologies
proceeds at full speed, with GBs of local storage and 10 GIPS
computing capacity in mobile handsets.
Berkeley professor Manuel Castells claims in an invited keynote to
ICSE'2001 (International Conference on Software Engineering in
2001 in Limerick, Ireland), that the Web has reduced the cost of
becoming an international organization by two orders of
magnitude. We are now witnessing a massive turn-around in how
we interact and organize production and consumption of both hard
and soft artifacts.
The work mode of the Internet creators was a combination of extreme
openness and rough consensus making. Yet most established
stakeholders, like the large IT and Telecom companies (IBM, ATT),
standardization bodies (ISO, CCITT), and international organizations
(UN, ITU), were passive. This was due to a mismatch between their perceived
technical and economic interests and the potential benefits of the
new net protocols. Now, many private and public actors want to have
a hand on the steering wheel. So, who should control and evolve common and
international technology that represents the core infrastructure of
our societies? That is, how to combine peer-driven technological skills
from above, with democratic representation from below?
As a tribute to the creators of the Internet and web, let us
paraphrase Churchill's speech to the RAF pilots
during the battle for Britain in 1940-1941:
Never have so few meant so much to so many so fast.
2. ARPA startup and the ARPANET
A comment on terminology: we will refer to ARPA, not DARPA as it now
is called. The name has changed as follows: ARPA (starting in 1958),
DARPA (1972), ARPA (1993), and finally DARPA (1996).
- 1957-1961: 4 Oct. 1957 with Sputnik, 10?? April 1961 with
Juri Gagarin - the "Sputnik shock".
- 1958, National Defence Education Act:
more math and science in schools, student financing support.
- 1958, Advanced Research Project Agency (ARPA): established for
basic and applied scientific/technological R&D, in order to
indirectly strengthen national
military defence, and to mainly be performed at universities.
The ARPA budget was 250 mill. USD in 1958, soon increasing to
2 bill. USD per year, and with 3.2 bill. USD in 2007.
All space-related activities went to the newly
formed NASA in 1960.
The "Mansfield law" of 1973 redirected ARPA and other
defence funds away from the universities. ++??
- 1961, Kennedy speech 25 May 1961: to go to the moon before 1970.
- 1962, Information Processing Technology Office (IPTO):
established at ARPA,
with dr. J.C.R. Licklider from MIT and
Bolt, Beranek & Newman (BBN) as the first visionary leader.
Licklider regarded computers as possible
extensions of the human mind.
He wanted computers to facilitate human communication, not just
perform traditional numerical calculations.
He also
initiated and funded a set of Centers of Excellence in Computer
Science on very generous terms,
e.g. at MIT, CMU, Stanford, University of Michigan,
University of Illinois, and UC Berkeley.
- 1967 - June, ARPANET work started: The first grant of 0.5 mill. USD
to ARPANET was awarded to Stanford Research Institute
(SRI) - on behalf of several partners -
and with Elmer Shapiro from SRI as project leader.
Sources: [ArpaNetxx] and [Leiner99] for more information and inside stories.
3. Prelude on Packet Switching: how, why, and by whom?
- General on packet switching in data networks:
Packet switching means
decomposing messages into smaller, fixed-size parts (packets),
which are sent separately over a network. This has gradually replaced
traditional, analog "circuit switching" in telecom.
In the latter, the whole route
between two callers had to be pre-reserved during an entire phone
single ticket for a plane between Trondheim and Oslo only when you need
to fly, instead of reserving a seat on the plane for a full year
at a time - as for royal seat cabins in opera houses!
- More on packet switching:
As mentioned, this means splitting a requested message or file into
smaller parts, or packets, of standardized size, typically 128 bytes.
A packet has typically its private and unique ID,
a message ID and corresponding sequence number to facilitate
disassembly and assembly,
a sender and a receiver address field,
some anonymous digital content, and a checksum for error control.
The packets are sent individually over a network towards a receiver.
The receiver checks each incoming packet for possible "garbled"
contents by using the checksum, and sends back an acknowledge
message (a receipt packet) with either "OK" or "Garbled" indication.
The receiver will reassemble each message from its received packets,
which at worst may arrive in arbitrary order.
The sender will attempt to resend packets being
"Garbled" or "Assumed-Lost" (no acknowledge before "time-out"),
possibly through another network route. Note that the acknowledge
messages also may be lost etc.!
To allow efficient "streaming" of time- and sequence-critical
data, like voice and video, a single acknowledge
message may be sent back after a entire "train" of packets.
See later comments on all this for the TCP protocols.
- 1961-1969, the three independent pioneers of package switching:
- Paul Baran et al. at RAND, 1962-1965:
located in Los Angeles. They had an US Air Force contract
to study and suggest possible
solutions for extra reliable phone transmission,
e.g. in case of nuclear war.
Their technical report from 1964 was however not followed up by
the Air Force.
Digression: The RAND Corporation
had to close shop when president Nixon refused further
public contracts to it, after learning that
RAND-employee Daniel Ellsberg had
leaked the Pentagon Papers to the New York Times and the Washington
Post in 1971. However, RAND quickly reappeared as
Information Sciences Institute (ISI) at the private
University of Southern California. They are still there.
- Donald W. Davies et al. at NPL in UK, 1965-1967:
NPL stands for the National Physics Laboratory in
Teddington, UK. Davies and co-workers
here specified and coined the term packet
switching in 1965, and implemented it in a small NPL Net in
1967. This lead to the packet-switched nets Mark I and
Mark II in 1970-1986, used also outside NPL.
- Leonard Kleinrock at MIT/UCLA, 1961-1983:
The real breakthrough for packet switching came from
Leonard Kleinrock at MIT (1957-1961) in his
Ph.D. Thesis Proposal: "Information Flow in Large Communication Nets",
Massachusetts Institute of Technology, July 1961.
Kleinrock showed, by analytical modelling and by computer simulation,
that packet switching would be inherently much more efficient and
robust than circuit switching, as universally adopted by the telecom
community at that time.
After graduation from MIT, he took up a job as an assistant
professor, later full professor, at UCLA. He and his Ph.D. students
were to play a central role in the next two decades, in building
what later would be termed the Internet. These Ph.D. students
include one of the two TCP/IP "inventors", Vinton G. Cerf (Ph.D.
1972 from UCLA, later at Stanford etc.), as well as Jon Postel
(Ph.D. in 1974 from UCLA, later working at ISI), and Steve Crocker
who would be working in the new Network Working Group,
plus David Crocker.
The role of Kleinrock and his own research group was primarily to monitor,
model and control the performance of the new network, e.g. by
first predicting and later measuring actual throughput,
reliability, and robustness in actual use.
- Reflections on packet switching:
Kleinrock's work on packet switching was fundamental for the ARPANET
managers to "dare" to ground the new network project technically
on such a new and unproven data transport mode. They had
certainly many other risk factors to worry about. However, by 1967 a
simple packet-switched net was operational by NPL in the UK.
Indeed, the operational network surpassed their pre-calculations by
a factor 3 in performance, using the TCP protocol in 1975. Indeed,
the rapid success of Internet to replace existing and partly
proprietary network protocols - such as IBM's SNA or DEC's DECNET
- was primarily due
to its outstanding transport performance. That is, the Internet protocols
were so much better in most regards, that the competitors were
just blown off the field. This lead to fast, de-facto
standardization and broad acceptance in the IT research community, and
gradually by the telecom and computer industry, and in related standards.
4. First Wave, 1967-1972: first ARPANET development
- The early ARPANET was based on a set of special IMP computers (Interface
Message Processors), using the NCP (Network Control Protocol) for
data transmission in the 50 KHz national network between the IMPs.
Each connected ARPANET
site would get an IMP, a minicomputer the size of a refrigerator,
which would be locally connected in an ad-hoc manner
to some of the main computers of the institution -
typically delivered by DEC and IBM.
The IMPs were chosen in 1967 as part of an overall network solution. They
were provided by the BBN company and its project manager Frank Heart
after a public call-for-tender, on
which no telecom company bid.
The four original sites to receive an IMP were, by Dec. 1969:
Stanford Research Institute (application:
Augmentation of Human Intellect by
using NLS hypertext system - Douglas Engelbart),
UC Los Angeles (Network Measurements Centre - Leonard Kleinrock),
UC Santa Barbara (Culler-Frie Interactive Mathematics centre -
Glen Culler and Barton Fried), and
University of Utah (Computer 3D-graphics - Ivan Sutherland and
Robert Taylor).
As mentioned, the UCLA team included graduate students Vint Cerf,
Steve Crocker, David Crocker, and Jon Postel.
In 1977 the number of IMP-connected sites was 111.
- Late 1971, NCP user-to-user extensions: the first NCP was only for
IMP-to-IMP traffic. Since all users resided on IMP-connected
mainframes, and facing similar contact "problems" to their local
IMPs, an extended NCP user-to-user protocol was devised. The two
"end-parts" had to be developed for heterogeneous mainframes. This
provided useful experiences with data transmission and protocol
implementation on rather diverse platforms, and influenced profoundly
the later TCP design in 1972-1978.
They also discovered that the fundamental IMP-to-IMP transmission
was wrongly reckoned to be fault-free, with the effect that the
transmission service "hung" on transmission faults. Furthermore,
the NCP could not address (or name) machines on other nets, only
those on the ARPANET itself. So it was time for a rethnking of
issues.
- Main user services were remote login (telnet) for cross-use
of computer resources (the original rationale of ARPANET) and
general data transport (by File Transfer Protocol or FTP,
the latter developed in 1971-1973).
- Development and standardization of network protocols and
associated technical issues relied on publicly available
Request for Comment (RfC) documents. These were started by
Steve Crocker, and assisted by David Crocker, in the NWG in 1969, and were
primarily written and discussed by the later Internet
Engineering Task Force (IETF). The RfC-s were edited,
distributed and managed by Jon Postel from
their conception in 1969 and for almost 30 years to come,
first at UCLA and later at ISI.
The SRI had the role
as the Network Information Center or project archive,
lead by Elizabeth (Jake) Feinler.
The RfC-s were initially sent out to all requesting or
interested parties by postal mail, later by the net itself (ftp,
email, web). This meant that any new or revised
protocol specification or corresponding
architecture / design would be carefully scrutinized by dozens of
experts serving a peer-reviewers, whether formally
engaged in the ARPANET project or not.
This emphasis on openness and teamwork
set a precedence for later initiatives on Free or Open Source (see later).
- 1967-1983, IETF meetings: Two physical meetings per year in
the pioneer years, typically to discuss protocol issues.
- 1968-, NWG: A more practical Network Working Group offered
mutual implementation assistance, although it paradoxically
also served as a "Theory" group with its many graduate students.
- 1969 - Oct. 29, First successful ARPANET test: between Stanford
Research Institute (Douglas Engelbart) and UCLA (Leonard
Kleinrock). UCLA managed to send the first three letters in a
login command - namely "L" "O" "G" - before the connection crashed.
But OK in the next attempt.
So this is the first birthday of the Internet.
- 1970, December: end-to-end connection through the NCP.
- 1972, October demo at ICCC'72:
Breakthrough demo of ARPANET, at the IEEE
International Computer Communication Conference (ICCC'72)
in Washington DC, with over 1000
attendants being shown the wonders of distributed computing.
- 1972 - May, Email invented: by Ray Tomlinson at BBN,
by combining SENDMSG,
based on earlier work on shared mailboxes, and TYPNET to send
data files (later using FTP). The "@" ("at" sign) in
email addresses was also invented then. That is, the first unintended
innovation in the ARPANET project was a fact! By July, an
upgraded version was implemented
by ARPANET director Roberts himself, and included
facilities for filing, fetching, and forwarding of individual messages.
Two years later, 75% of all net traffic was email, regulated by
a separate SMTP high-level protocol.
- 1971-1973, French Cyclades/Cigale research network:
used for research, lead by Gerard LeLahn from INRIA.
5. Years 1965-1972, Comments on openness and independence
- Comment 1 - Creative and cooperative work processes,
carried out by dedicated and altruistic craftsmen
The "community spirit" around working with RfC design
documents and related implementations paved the way for generous
and free sharing of software in the Unix community since the early
1970s. This culture again was adopted and systematized by the Free
and Open Source Software communities in the early 1990s. As a
consequence,
100,000s of software components are today publicly available, e.g.
on licensing terms such as BSD (do what you want) or
copyleft (continue to be free). And most importantly, this altruistic
attitude and policy to
openness has so far "guaranteed" that
the resulting Internet has been
regarded as an open resource for everybody until now.
Indeed, the ARPANET/Internet culture was very different
from the one in its ultimate sponsor, the DoD, although maybe
close to the one ruling in ARPA before 1990.
- Comment 2 - A "distributed" and "open" net:
A fundamental design principle for Internet has been
a "distributed" approach, independence from the large
telecom equipment & service providers, and to practise
open protocols, plans and work practices.
That is, no single, strong actor should be able to take "control"
in a "coup" or similar.
The net champions are also easy to mobilize to "rescue" the net.
And does private vs. public ownershop mean anything here?
- Comment 3 - Toolset for distributed development:
The ARPANET development teams were distributed, and necessary
coordination support were gradually put in place ("bootstrapped")
over a growing network. Initially ftp and later email were applied;
later came modern version management tools, web,
and groupware tools such as wiki. The Open Source community
has later been a pioneer in developing and using such tools,
e.g. Subversion tool for versioning, Gentoo for release management,
Bugzilla for defect tracking, ...
- Comment 4 - ARPA's and NSF's trust and generous funding:
The funds for the entire ARPANET and Internet development are of
course substantial over the 23-year period 1967-1990, say 20
mill. USD per year - financing perhaps 200 researchers per year?
In relation to ARPA's and DoD's total budget (3 vs. 500 bill. USD per
year), we are however speaking of pocket money.
The research grants were also administered with great freedom
for the researchers and demanding not too detailed proposals. So all
in all relatively "easy" money.
- Comment 5: Innovation is seldom planned:
There are few cases where a research project has
resulted in so much plain usefulness and thus value,
indeed for all mankind. As for all
"great innovations", the application area of the final innovation
was not the one originally foreseen. That is, ARPANET was
conceived to facilitate cross-use of computer resources between
the universities, but this goal faded quickly. Instead we
have got a superb, intercontinental, open-for-use digital highway,
whose value has been amplified thousandfold
by "killer applications" like email and web. And all this has
happened in 30 years!
6. Second Wave, 1973-1990: New TCP/IP protocol
See Policy issues in 2nd Wave.
- 1972-73, network integration start-up: gradually several
different packet-switched nets popped up:
ARPANET (1970), SATNET (1973) over
the commercial INTELSAT satellite between USA
and Europe - involving UK/Norway, AlohaNet (1971) in Hawaii using
fixed radio links, and others.
Then, how to interface an increasing and
clearly unmanageable set of heterogeneous networks, where each of
these could differ in:
To agree upon one unifying network protocol to replace the
exiting ones was deemed unrealistic and even presumptuous, as
their owners and users were relatively happy with the quality of
service offered. As we will soon see, a network
meta-protocol on top of unchanged ("as-is"), existing
protocols was the answer. NB: It turned out that the NCP itself
could not be used "as-is", e.g. for network addressing.
- 1972, INWG: an informal International Network Working
Group, created after the 1972 ARPANET Conference in Stanford and
with American members (ARPANET team, plus Bob Metcalfe with the brand new
Ethernet LAN from Xerox), and British members (from NPL in Teddington and
UCL in London), and French members (Gerard Lelahn from Cyclades net).
The idea was to approach the network integration problem with an
open mind, and learn as much as possible from previous network
protocols, their requirements and assumptions, design and
implementation principles, and preliminary usage
experiences. Especially the French experiences strongly suggested
"end-to-end" user control to be able to support content-specific
policies upon errors and other abnormal situations,
e.g. during film streaming or to "censor"(!!) sensitive contents.
In the ARPANET, all error handling was simply delegated to the
NCP protocol (assumed 100% reliable - but not so),
i.e. outside the (main) computers where the users and
their applications were hosted.
Possible competition from the ongoing ISO standardization effort
with a proposed seven-level architecture called OSI (Open
Standards Interface; still not implemented), and the concrete
"competitor" X.25 packet-switching protocol from 1966, was quickly
solved. OSI turned out to be a paper dragon, and the new
"meta-protocol" could simply use most of the X.25 "as-is" as an
underlying transport facility. So the "clash" between the ARPANET
team and the mighty telecom community and their bureaucratic
standardization committees boiled away. That is, matters solved,
let us do something new and better instead!
- 1973, September - Sussex University: important NWG meeting, in
conjunction with another event.
- 1973-1974, new TCP protocol: was defined
for end-to-end connections between computers [Cerf74].
This was designed by Robert E. Khan (known as a loofy philosopher) and
partly by Vint G. Cerf (as the practical architect), with both being
advised by Bob Metcalfe and Gerard LeLahn. This meant phasing out
the IMPs (happened in 1983) and using e.g. the ISO X.25 telecom
protocol for actual packet transfer.
Visiting Norwegian professor Dag Belsnes from Univ. Oslo in 1975
helped with formal specification of TCP to discover potential errors.
- 1974 - end-of, first TCP implementation ARPA
awarded a contract to implement the
new TCP, to a joint team between Stanford (ARPANET: by Vint Cerf),
BBN (radio net: by Ray Tomlinson), and
UCL in London (SATNET: with Peter Kirstein). In addition David Clark
at MIT implemented TCP on the Xerox Alto and later on an IBM PC,
demonstrating that the protocol could run on non-mainframe also,
- 1975, ICL in London: visiting Norwegian researcher Pål Spilling
helps in testing new TCP between several nets.
- 1977, October 20??: TCP successfully binds together
three computer networks: ARPANET (phone lines), SATNET (satellite)
and AlohaNet (radio between fixed places in Hawaii) - reckoned as
the second and real birthday of the Internet.
- 1978 - March, separate IP protocol layer: taken out,
splitting the "old TCP" protocol into the present TCP/IP
- a work credited Khan and especially Cerf.
The new IP layer handled network address translation (from
symbolic name to numeric IP address), routing of packets,
and interfacing to the
more low-level transport protocol, e.g. X.25. However,
packet disassembly and
assembly remained in the TCP layer to be able to
recover more intelligently from transmission errors.
- 1972-2000: New control institutions to manage and oversee
network structure and use. E.g. the Internet Architecture Board
(IAB) from the late 1970s, to monitor and coordinate ongoing
design and maintenance of protocol standards.
Other control institutions were the Internet Control
Board (ICB), which became the Internet Change and Control??
Board (ICCB), and finally the Internet Governance
Forum (IGF) under UN, first chaired by Phil Gross in 1985.
- In the 1980s:
Increased use of fast LANs (Local Area Networks e.g. using
Ethernet technology) to connect personal workstations / PCs, and
where each LAN has a gateway (router) to connect to other gateways.
- 1975-: growing use of the Internet, due to "killer
applications" such as email (from 1972) and later the web (from 1993).
- 1973, Norway joins to get its NORSAR "seismic listening
ear" on the ARPANET via a satellite link, and arranged by military
research director Yngvar Lundh.
- 1976, SATNET:
using packet switching and the commercial INTELSAT satellite.
- 1977: Coupling of ARPANET + SATNET + Mobile Radio Network
to make the first, de-facto "Internet".
- 1979, USENET: from UUCP (cheap Unix mail).
- 1979, BITNET (Because It's Time Network):
by NYC UNiversity and IBM,
mainly for email traffic and news bulletins.
- 1980, TCP/IP as military standard: named
MIL-STD-1782, and later a separate MILNET in 1983
- 1980 - 27 Oct., virus attack: causing ARPANET to halt.
- 1981, Cheap CSNET for academia:
as civil counterpart to ARPANET.
- 1982, EUnet (European UNIX Network):,
by EUUG (European UNIX User Group).
- 1982, Norway and ICL move their connections:, via TCP/IP
over SATNET.
- 1982, First use of the word Internet: meaning a
connected set of networks.
- 1983 - Jan. 1st, old NCP protocol and IMPs abandoned;
all must use TCP/IP towards ARPANET.
- 1983, ARPANET split: into MILNET and NSFNET,
the latter being upgraded to also serve as
a supercomputing network from 1985.
NSF leader Stephen Wolff then closed off
the net for commercial use. This predictably triggered
commercial Internet providers to enter the market.
- 1983, EARN: European Academic and Research Network using TCP/IP.
- 1983, DNS (Domain Name System): set up by Univ. Wisconsin
and David Mockapetris to maintain a distributed and
hierarchical set of name servers with addressing and routing
information.
- 1984, JANET in the UK: to be used for all higher education, and
similar plans for NSFNET from 1985, i.e. not exclusively for
computer scientists.
- 1984: Part of CSNET upgraded: to be part of NSFNET.
- 1980s: Rapid growth of network:
5,000 net servers in 1984,
28,000 net servers in 1986, and
100,000 net servers in 1989.
- 1980s: US government pays computer industry 20 mill. USD:
to implement TCP/IP on all their computer models,
to be ready by 1990.
In addition there is
TCP/IP freeware delivered as part of Unix systems,
e.g. the BSD (Berkeley Source Distribution) ones.
- 1985-1995: ARPA de-facto out of the ARPANET project:
NSF takes over the 200 mill. USD
of accumulating operating costs for what is now the Internet.
- 1985: Three-day Internet workshop: 250 industrial persons from
50 companies showed up.
- 1987: SNMP protocol (Simple Network Management Protocol)
defined for network management, to update distributed
tables with network addresses and routing information.
- 1988: National Research Council committee: Towards a National
Research Net, supported by Al Gore.
- 1994: National Research Council committee: Realizing The
Information Future: The Internet and Beyond.
- 1995, Oct. 24: Federal Networking Council (FNC): passes a
resolution to define the Internet.
- 1990: CSNET + NSFNET + EUNET: becomes the Internet, and
ARPANET is history.
All new NSF grant holders must use the new Internet, as
military contractors must use MILNET.
- 1992, NFSNET: opened for commercial use by NSF, reversing its 1985
decision on this.
- 1995, NSF out of NSFNET: after spending 200 mill. USD on it
since 1983. Now commercial actors and possibly other
public entities must step in.
- late 1990s, commercialization: and increase of nets
and Internet providers, fuelled by exploding web usage.
- 1996-, Gradually more "resistance" from the telecom community:
e.g. regarding domain addressing, naming standards,
net protocols and standards, and a push for more closed code.
That is, they have "slept in class" and find themselves outrun.
For instance, IP telephony is much cheaper than traditional
telephony, although both are digitized.
Policy issues in 2nd Wave:
- 1972; 1998 (remade) - IANA (The Internet Assigned Numbers
Authority):
IANA was set up to assign new domain names etc. in dialogue
with regional authorities. Jon Postel handled this function almost
singlehandedly in 1972-1998 (until his death).
From 1998 IANA has been managed through ICANN, see later.
- 1995, IANA (The Internet Assigned Numbers Authority):
IANA oversees global IP address allocation, DNS (Domain
Name Server) root zone management (.com, .edu ...), and other
Internet protocol assignments. IANA delegates local
registrations of IP addresses (names and numbers) to the
Regional Internet Registries (RIRs), and the one for Norway is
www.nordid.no.
It is itself managed by ICANN.
Jon Postel at ISI managed the IANA function (with different
names before 1990) from the start around 1972
until his passing in October 1998. After his death, his colleague
Joyce Reynolds, managed to let the IANA function be placed under
ICANN. In November 2003, Doug Barton was appointed IANA manager.
In 2005, David Conrad was appointed as IANA manager.
- 1996, IGF (Internet Governance Forum): UN institution
to give strategic advice on how to run and evolve the
Internet. Last meeting in Rio de Janeiro, Nov. 12-15, 2007.
Also Internet Annual Meeting in Los Angeles, Oct. 29 - Nov. 2, 2007.
- Sept. 1998,
ICANN (Internet Corporation for Assigned Names
and Numbers): is established as a mixed public/private
institution to deregulate acquisition of domain names and
to oversee their evolution and use.
It reports to the US Department of Trade (which accepted ICANN
in Feb. 1999), and has its legal status according to
Californian laws.
But its daily operations are still delegated to the previous IANA.
Ester Dyson and later Vint Cerf volonteered as
the first ICANN top managers, and Cerf even sits
on the board through 2007.
In September 2000 there is the first
international election of new board members in
ICANN.
7. Technical remarks on the new TCP/IP protocol
- TCP is best seen as a network integration (meta-)protocol,
glueing together possibly heterogeneous networks, thus the name
"internet" (reusing slides from Cerf's talk in Trondheim 21 Nov. 2007):
- TCP utilizes and oversees packet transmission (see earlier on packet
switching in general) over any communication service -
so IP on everything (see below)!
- TCP does have facilities for "windowing" to allow data streaming for
demanding media like sound and video (high transfer rate,
little "jitter").
- The net carries any digital content - no interpretation or tailoring.
- End-to-end transmission implies neutrality and user freedom;
all domain knowledge and "intelligence" at the topmost
application or user level, not at the bottom or core transport level.
- Radio supplies mobility; fiber/cable/DSL supplies speed.
Observation: The new routers can be compared with the old IMPs, and the
LAN-connected PCs with the old mainframes.
So instead of connecting computers and thereby humans, we are
connecting networks (often LANs) and thus even more humans.
- Design simplicity:
- Next best is usually OK:
Do not strive for costly perfection; just running code.
Accept that some packets may be lost - and later resent.
- Sophistication on the outside (the ends); simple and general
solutions on the inside:
This is the opposite of the telecom solutions. We do not know
the transmission contents and applications of the future, like
the web before 1991, so do not sub-optimize prematurely.
- Step-by-step routing policy:
Each node knows only about its neighbors and the general
"direction" of major hubs, e.g. towards "whom" should I send
packets to USA or Italy? It does not know the full network
topology or current traffic patterns. The task of each
intermediate network node is simply to send an incoming packet
one step further in the assumed best "direction" of the
receiver, as there is no pre-planned path for a packet.
Network transmission errors and temporary/partial network
blockages will be discovered dynamically, with consequent
rerouting of resent messages. On the day of the 9/11 terrorist
attack in New York City, the Internet was much more available
than other telecom services (mobile and stationary services),
due to flexible rerouting algorithms.
Incredibly enough, this minimalistic, step-by-step
transmission policy works very well wrt. capacity, reliability and
robustness - although there is no guarantee in advance that a
working connection will be found. This policy has been
met with considerable dismay from the telecom profession! So
when sending an email message from Norway to a site in USA, a
hundred networked computers may be involved to assist in the
transport, but basically not knowing what is happening!!
- IP on everyting - once more:
Ardent TCP/IP supporters claim that this protocol can be run
atop of most transmission carriers, such as children's iron
wires to connect paper membranes in empty food cans, or by
using a pigeon post service, as for Baron Paul Julius von
Reuter's news service in 1849 between Aachen in Germany and
Verviers in Belgium.
- Overview of network layers:
- User application/service: telnet, FTP, SMTP (email), HTTP (web), ...
- High-level transport: TCP
- Medium level transport: IP
- Low-level transport: X.25
- Physical transport: by cable, radio (mobile: GSM/UMTS),
optical fibre
- Example of transmission flow
- assuming two computers C1 and C2, each its own LAN with routers R1
and R2:
- Send from computer C1 to local router R1 in current LAN.
- Send from router R1, via many others in possibly
different nets, to router R2.
- Send from router R2 via its LAN to computer C2.
- Or inversely back, from C2 to C1.
8. Third Wave, 1991-2000: the World Wide Web (WWW, or just web)
- The web resides on top of the Internet.
The web was invented - of all places - at
CERN in Geneva by Tim Berners-Lee (British) with help from Robert
Caillau (Belgian), and made public on 6 August 1991. The web
couples (in its orginal design) hypertext documents with the Internet, and
has initially three innovations:
- html (HyperText Markup Language),
- URL (Universal Resource Locators, such as
www.idi.ntnu.no/~conradi/index.html), and
- http (Hyper Text Transfer Protocol) to connect a local web
browser with a remote web server.
Surprisingly enough, Berners-Lee had argued in vain to have CERN
interested in developing a prototype. A joint paper by Berners-Lee
and Caillau was also rejected at the Hypertext conference in Autumn
1990. Success does not come easy!
Almost immediately, the web took off, helped by freeware
net browsers such as Mosaic from Univ. Illinois at
Urbana-Champaign in 1993, later becoming a product by the new
Netscape company. It was a simple http extension to let the
downloaded texts also serve as executable programs - so-called "cookies",
written in an interpreted (i.e. directly executable)
programming language like Java or PHP.
This enables a web server to instrument a web client, which is
running a web browser, to behave as a universal
"eye" or interface into the web. This opens up for all kinds of
applications, such as net shops, information portals, blogs,
wikipedia, and not to mention search engines. It also opens up for
ten-thousands of virus infections!
- The semantic web 2.0, using XML:
This extended web seeks to exploit a "generalized" html, called XML or
eXtended Markup Language,
to define more of the syntax and semantics of web-stored information.
More precise and tailored processing of such information
will then be possible.
Digression: IBM in Germany developed around 1980 a
general markup language SGML, but nobody saw its potential.
Berners-Lee successfully simplified the flexible markup codes in SGML into a
fixed set in the new html, e.g.
<font ...> (start using a new font),
<table> (start of a table),
<ol> <li>item1 <li>item2 </ol> (an ordered
list with two entries), and
<b> ... </b> (put some text in boldface font).
Other researchers perfected the dormant SGML into XML, where the
user can define her own markup codes, like <PersonNumber> or
<BankAccountNumber>. This leads to Web 2.0, and a plethora of free
support tools to decode and process XML documents. Such documents
are often very large text files (GBytes). Each file contains a
definition section (a schema) with new markup codes defined in XML,
and a data section using these codes. An XML file typically holds
information about a set of invoices, a team of football players,
books in a library etc.
- 1995 - April, W3C consortium: to overlook the
evolution of web-related standards, with Berners-Lee as general
manager. Located at MIT, supported by INRIA in France and Japan.
9. Fourth Wave, 2000-now: "New times" for interactive, distributed
multimedia work
Basic ICT infrastructure:
we have got cheap and powerful laptop PCs, PDAs/ mobile-phones,
Internet, LAN, wireless WiFi and fast UMTS/GPRS nets to stay
connected with high bandwidth at nearly at all times and places
(nomadic computing), not to mention web++ on these devices.
Construction tools:
In addition comes a large set of common and partly
freeware construction tools for e.g. text processing,
spreadsheets, graphics/images/video etc. - and
with use of standardized data formats and transfer protocols.
For instance, a TV director can now compose and edit a whole TV
program on a normal laptop, sitting on a street cafe, and sending the
final program via the local
WiFi net to the "home TV station" for further broadcast or
network dissemination.
Twenty years ago, such an editing process
would have required a "studio" of the size of a 50m2 apartment.
Another example: due to the penetration of laptop PCs and local
WiFi nets, our university's specialized computer rooms and reading
rooms stand almost empty. The students sit "everywhere" in
small groups, and work and interact through their computers.
All are equal in the name of ICT: The pervasive access
and versatile use of ICT has caused several authors to claim that
the earth is flat [Friedman06].
This means that all countries and companies now stand on an
equal foot wrt. competitive selling of brainpower and in production
of ICT-enabled and ICT-reliant services and products.
This "flatness" will accelerate
outsourcing not only of labor-intensive physical products, like
clothes, but also of many services that by their nature is
labor-intensive, like legal advice.
Many "low-end" services have been outsourced to low-wage places like
Bangalore, e.g. "call centers"
to help you fill in IRS forms, "hot-line" help for PC users (e.g. how
can grandmother write a postcard before Christmas?), or
standard booking of plane tickets. More "high-end" services include
specialized medical diagnoses and outsourced software development.
This illustrates the new interaction patterns and media made
possible through the web, and enthusiasticly adopted by young
people (less than 30!). The ICT support for symmetric and diverse
participation seems groundbraking. Any "producer" can upload on
the web their own texts, drawings,
photos, speak, music, videos, and combinations of these - being
written and sent from mobile devices - and later being
downloaded from the web by some "consumer".
Anybody can thus be a writer or graphical artist - no more "quality
control" by book publishers or art galleries.
The relevant web resources will be own
blogs, the new FaceBook site (already subscribed by
100,000s of Norwegian users), home videos by Utube etc.
The possibilities and implications of creative commons
("dugnadsbasert felleskunnskap"), as an extension of OSS, also
deserves some thought. The common artifacts
can be almost anything: (young) people's
blogs and email messages, arbitrary personal notes and essays,
online discussions, home photos and videos, software components,
construction engineering and architectural designs,
books, scientific papers, ...
This raises issues of privacy, copyrights and ownership, business models
with attached services, quality (not offending others - use
moderators?), ...
Of web-enabled tools we should not forget the newest and most
spectacular innovation of them all, the textual search engines
a la Google. These operate internally with a flat, textual search
space - regardless any pre-existing structure (document format, contents,
topic, origin, internal data organization). This search space is then
(magically) "indexed" ca. once per day. The query responses are
sorted after asssumed relevance, and we usually get
a response time of a textual query of less than a second.
The three most frequent search words are "sex", "god" and "work" -
in that order [Friedman06].
For instance, it takes me shorter time to use google to get the
phone number of my colleague down the hall (searching among 20
billion webpages) than to navigate and search in the university's
web portal. Or I can easily find the home address of my colleague
through her phone number.
A more sophisticated use is to analyze the text of new articles in
the Medline database, where 2100 are added every day (!). It is
infeasible even for a top medical researcher to keep updated in
her field of specialization and interest.
Dr. Rune Sætre, a NTNU Ph.D. graduate from 2006, made a
text-comparison tool called GENETUC [Sætre06]
to analyze in which articles and by which authors
the same human gene were described in
the Medline articles. He thereby coupled hundreds of researchers
that had been unaware of each other's existence and overlapping
research. This "meta-research" lead to almost ground-breaking
results. So let the earth be one.
Most written knowledge from the last 15 "post-web" years
are already on digital form, and thus searchable by common
search engines. A similar
situation is for the (mostly unstructured) internal archives of
large organizations (StatoilHydro, NTNU).
So they ought to apply modern search technology on their own data.
For the "pre-web" books, there
are ambitious scanning initiatives by Google, EU (Gutenberg
project), Norway (Runeberg effort) and other actors
to make available and later searchable, most older books,
magazines, newspapers etc. - mainly from
paper-based libraries and publishers. This raises, regrettably (?)
the issues of copyrights to books less than, say, a hundred years
old. So, is this medium-old literature
"owned" by their living writers, their heirs,
their publishers, or by the greater society (at least eventually)?
A united world: Regardless of copyrights, mankind will
soon be connected to each other and to most existing
knowledge - in completely new ways by just a few decades.
What will this combined "access" mean? Let us live to try it out!
10. Some current and future issues and challenges for Internet
Some more technical aspects:
(mainly from Cerf's talk at NTNU 21 Nov. 2007):
- IPv6 supplies 128 bits address space:
otherwise IPv4 runs out in 2011.
- Increasing capacity in the core and the edges:
i.e. more optical fiber in the middle and better broadband and
threadless communication for the end-user.
- Mobility: coupling to WiFi/UMTS - so watch films everywhere!
- Broadband: gives choice about up/down-load symmetry.
- Security and authentication: very sensitive areas.
- Space: extending the Internet in space, but long latency times
(hours, days).
Some policy aspects:
- P1. Democracy and commercialization vs. technocracy:
How to balance slow and representative control vs. effective
hacker-community management of
how the Internet standards should evolve, and
how IP addresses and names are assigned:
That is, we need both legislative, executive and
controlling institutions, organized by public or private organizations,
p.t. lead by The US Trade Department,
by political elections / appointments (a la ICANN)
or by peer-respected and
self-recruiting technology experts (a la old IANA attn/ Jon Postel)?
- P2. Free competition and equal net access to foster innovation:
All users shall have the same and unhindered
access, thus enabling and promoting
innovation. No censorship like in China. But OK to
pay more for higher net service quality (speed, reliability)?
- P3. Bundling of net-infrastructure and contents providers,
and their services:
The broadband or cable-TV
providers should not deny us certain services or contents,
partly caused by shared ownerships (cf. P2 above and P4 below).
A consequence is that we get an expensive and
limited (even no) access to alternative use or contents, such as
less-popular TV channels or films.
As an illustration of mixed interests , look at the company
"Norges televisjon" - established to build up a new digital, land-based
broadcasting network in Norway -
This company is a mess of conflicting owners and interests (NRK, TV2,
Telenor, ...).
- P4. Media ownership restrictions (not directly Internet-related):
Shall infrastructure providers (ex. Telenor) be allowed
to own or make special deals with content providers
(ex. TV2, newspapers, film distributors) - i.e. bundling.
Ex. Telenor has leaped from traditional, well-regulated public
phone services to almost "turbo capitalism" in the media world.
Similarly, shall media companies be allowed to own or control
other media, like Schibsted Group owning parts of TV stations,
newspapers, book publishers etc.?
Number of users per summer 2007 - and counting:
(source: Cerf's slides of 21 Nov. 2007)
- Now 1.24 bill. people being connected (18.9%); 50 mill. people in 1997.
- Now 489 mill. hosts; 22.5 mill. hosts in 1997.
- Now 2.5 bill. mobile phones and 1.1 bill. immobile phones.
- Now 1 bill. personal computers: desktops or laptops.
- Now 20 bill. web pages (used by Google).
11. Literature and References
- [ArpaNetxx] authors??:
"Arpanet history",
http://www.dei.isep.ipp.pt/~acc/docs/arpa-Contents.html.
- [Cerf74] Vinton G. Cerf, Robert E. Kahn:
"A Protocol for Packet Network Intercommunication" (original
unified TCP),
IEEE Transactions on Communications,
Vol. 23, No. 5, pp. 637-648, May 1974.
- [Cerf07] Vinton G. Cerf:
Vinton G. Cerf, Google vice president and Chief Internet Evangelist:
"The Next Generation Internet",
guest lecture and media show at NTNU, 21 Nov. 2007 (.ppt).
- [Friedman06] Thomas L. Friedman: "The Earth is Flat",
Penguin, London, 2006,
ISBN 978-0-141-02272-7, 600 pages, 175 NOK (July 2007).
- [Hira05] Ron Hira, Anil Hira: "Outsourcing America",
American Management Association, New York, Sept. 2005,
ISBN 0-8144-0868-0, 236 pages.
- [Kleinrock66] Leonard Kleinrock: "Communication Nets; Stochastic
Message Flow and Delay", McGraw-Hill, New York, 1964.
(Out of Print) Reprinted by Dover Publications, 1972.
(Published in Russian, 1971, Published in Japanese, 1975.)
- [Kleinrock75] Leonard Kleinrock:
"Queueing Systems. Volume 1: Theory",
Wiley-Interscience, John Wiley & Sons,
448 pages (hardcover), 1st edition (January 2, 1975).
ISBN 978-0-471-49110-1.
- [Kleinrock76] Leonard Kleinrock:
"Queueing Systems. Volume 2: Computer Applications",
Wiley-Interscience, John Wiley & Sons,
576 pages (hardcover), 1st edition (April 22, 1976),
ISBN 978-0-471-49111-8.
- [Leiner99] Barry M. Leiner, Vinton G. Cerf, David D. Clark,
Robert E. Kahn, Leonard Kleinrock, Daniel C. Lynch,
Jon Postel, Larry G. Roberts, Stephen Wolff:
"A Brief History of the Internet" (v3.2),
http://arxiv.org/html/cs/9901011 (.html), CACM, Feb. 1997 (50th
anniversary issue).
Computing Research Repository, CoRR cs.NI/9901011, Jan. 1999.
- [Rasmussen07] Terje Rasmussen: "Kampen om Internett" (in
Norwegian), PAX, Oslo, 229 sider, ISBN 978-82-530-2999-3, 2007.
- [Saltzer84] Jerome H. Saltzer, David P. Reed, David D. Clark:
"End-To-End Arguments in System Design",
ACM Transactions in Computer Systems, 2(4):277-288, Nov. 1984.
- [Sætre06] Rune Sætre:,
"GeneTUC: Natural Language Understanding in Medical Text",
dr.ing. thesis, NTNU 2006-59, ISBN 82-471-7866-4 (printed),
ISSN 1503-8181, 155 p.
Advisor: Tore Amble, NTNU;
Current employer: NTNU and Tokyo University.
- [Turing36] Alan Turing: "On Computable Numbers,
with an Application to the Entscheidungsproblem",
Proceedings of the London Mathematical Society, Series 2,
41:230-267, 1936.
12. Appendix A: Internet institutions
- ARPA (from 1958): Advanced Research Project Agency, in DoD.
The DoD "research council".
- ITU (from ??): International Tele Union, with CCITT
standardization body.
- W3C (from April 1965): The International Web Consortium with
Berners-Lee as director.
- IAB (from 1998??, Feb. 27): Internet Architecture Board.
Set up by Jon Postel to oversee many technical/managerial
issues, and is again decentralized into almost 20 task forces (the
IETF being the most important), and even subtask forces.
- IETF (from 1970??): Internet Engineering Task Force.
A key institution until the late 1970s, when must of the
protocol design work was completed. Now reporting to the IAB.
- IANA (from 1972): Internet Assigned Numbers Authority.
Assigns new domains in cooperation with national authorities.
- ICANN (from 1995): Internet Corporation for Assigned Names
and Numbers.
ICANN delegates to and oversees the work in IANA,
under contract to the United States Department of Commerce (DOC).
- IPTO (from 1964): Information Technology Project Office.
A suborganization of ARPA in the 1960s and 1970s.
- ISOC (from 1992): The Internet Society.
A public foundation to oversee the work in IETF, IAB and
partly IANA. ISOC has also a mission to promote openness in
Internet use and evolution.
- NWG (from 1968): Network Working Group.
Shares practical implementation information.??
13. Appendix B: Mini-glossary of IT terms
- GIPS: Bill. no. of machine instructions per second.
- MIPS: Million no. of machine instructions per second.
- BSD: Berkeley Source Distribution, special lisence type for Unix.
- LAN, Local Area Network (often 10 MHz): like Ethernet, to
link together computers within a building.
- TCP/IP, Transport Control Protocol / Internet Protocol:
The main Internet protocols.
Alan Turing proposed in 1936 a general computing machine, the
"Universal Turing Machine" or UTM [Turing36]. It is essentially
a finite-state machine extended with an infinite external memory.
The hypothesis is, that such an UTM
can simulate any computer, or generally any other UTM.
Such a UTM has:
- Sufficient external tape storage - for symbols:
A UTM has a potentially infinitely long (i.e. will never run out)
external tape, organized as a linear sequence of discrete "cells".
Each cell contains a symbol from a finite alphabet.
The alphabet includes the blank symbol ("B"), serving as a
default, initial value in all cells, and for simplicity just two more
symbols - "0" and "1". A sequence of e.g. eight tape cells may
then have the contents "0100B10B".
Parts of the tape may be given a non-"B"
initial contents, serving as input data to the UTM.
- A tape head to read/write symbols and to move the tape:
The tape head is positioned at the current cell of the
tape, and can read or (over)write the contents of this cell.
This contents is for convenience
abstracted by a current symbol variable.
The head can also move the tape forward or
backward in single cell steps.
- A basic print facility:
The UTM can print the value of the current symbol and any
explanatory text, such as "HALT", onto some unspecified, external device.
- A finite-state machine:
This is the core of the UTM. It consists
of a predefined and unchangeable ActionTable or matrix,
where each slot may contain a 5-tupled instruction serving
as a "program". There is an associated
execution engine (a "CPU") to interpret this program, done by one
instruction at a time - see next point.
The finite-state machine uses two variables:
the current state (often called a state register) and
the mentioned current symbol to represent running input data.
A special start state-value serves as the
initial value of the current state.
There is usually an explicit halt state-value, upon which
the UTM stops its execution.
- An instruction format:
Each table slot or matrix element is either empty or
contains an executable instruction. The latter is
a 5-tuple with parts called e.g. E1-E5:
- E1: a possible tape move-command - indicating a one-step
Left-move, Right-move, or No-move of the tape
- in short a L, R, or N.
In case of a L or R, the value of the
current symbol will correspondingly change, i.e. an
implicit "read"-operation after each head move.
- E2: A possible tape write-command to
change the contents of the current cell, possibly
changing the value of the current symbol accordingly
(i.e. an implicit "read"-operation after a write-operation).
- E3: Another possible move-command - either
L, R, or N as in E1 - and
possibly changing the value of the current symbol.
We can thus move the tape before and after each
read/write-operation, where reading is implicit and
writing explicit.
- E4: Possibly a print-command
for execution after E1-E3 -to show either
the current symbol or the text "HALT" externally somewhere.
- E5: Finally, update the current state
with a new, specified state-value.
Note 1: None, one, two or three of the tape operations in E1-E3
can be specified and thus executed per instruction.
Note 2: If E1-E4 are all empty, only a state change may occur in E5.
Note 3: In Turing's original paper, the UTM state included the entire
execution history including all written tape cells.
Note 4: The current symbol is usually "B" at start, but this
depends on what input data we let the tape contain initially.
Note 5: All anticipated "final" states may
be set to arrive at this common halt state,
usually after printing a unique message.
Note 6: Attempt to execute an empty instruction
is an error and will probably lead to halting.
Note 7: There are then three possible outcomes:
"erroneous" halting being close to a software fault!!,
"successful" halting via possibly different execution paths, and
"undecidable"!! due to still ongoing execution.
Note 8: Finally,
there is a specified procedure to initialize and start the
machine.
Note 9: Wrt. instruction
format - a 3-tuple is also possible by merging E1-E3.
- A machine execution cycle, summing it all up:
The 5-tuple instruction in the slot
identified by ActionTable(current state, current symbol) determines:
1) The actual tape and print actions.
2) The next current symbol value, taken from the tape
(an external input value).
3) The next current state value (a UTM-internal value).
So by sequentially executing one instruction at a time during a
machine instruction cycle, the UTM execution engine
can execute its pre-made program.
A modern computer executes in a similar way billions of memory-stored
instructions per second (several GIPS), as mentioned elsewhere??.
Many UTM simulators have also been programmed on modern computers
and with good support facilities - for those that want to try this out.
Remark on UTM variability:
When a UTM shall tackle a new problem or task, the finite-state machine
table must be filled up or "programmed" with relevant problem-solving
instructions.
When merely another data set or example shall be analysed, the machine
table must be initialized once with an existing program for the task at
hand, while the input data on the tape varies between executions.
A classical challenge will be to first put a description of
some formal language on the tape and then a statement in that
language. The UTM task is now to analyze the possible truthfulness of
the latter statement, by expecting a Yes or No after some years!
Lastly and most importantly, Turing was in 1936 able to prove,
that such a UTM may not always stop to execute its program, the
so-called halting problem. However, that a UTM may simulate
any other UTM or even any other computer, is p.t. only
a very well justified hypothesis, but nobody has so far managed to
outline a computer that cannot be simulated. So "proof" by
generalization from overwhelming pluraltiy, since no falsification so far.
15. Appendix D: The Technical Computer Revolution
- 1946, ENIAC at Moore School, Pennsylvania:
first electronic computer, but programmed by a fixed plugboard.
- 1949, EDVAC at Moore School, Pennsylvania:; with stored program -
after an idea from John von Neumann.
- 1948, "Baby" Mark I at Manchester University: also stored program.
- 1949, Mark II at Manchester University: with larger program.
Later becoming the Ferranti Mark I,
the world's first commercial computer.
- 1949, EDSAC at Cambridge, UK: similar to EDVAC.
- 1947 - 16 December, ATT's Bell Labs: William Shockley, John
Bardeen and Walter Brattain built the first practical point-contact
transistor.
- 1950, ATT's Bell Labs: develops modem
(modulator-demodulator) to convert from analog to digital
signals and v.v.
- 1951-1953, First mainframe, 24-32 bits addresses:
from IBM, UNIVAC, CDC, Burroughs, ICL, Bull, ...
- 1961-67, First minicomputer, 16 bits: PDP-I and
PDP-7 from DEC, NORD-1 from Norsk Data.
- 1975, First microcomputer, 8-16 bits: The MITS Altair 8800
was based on the Intel 8080 CPU (a successor of the Intel
4004 and 8008. It was mainly
used for programming in Altair BASIC (MicroSoft's founding
product!) and for computer games. Many similar microcomputers
followed, like the Commodore.
- 1978,1982 First graphical workstation, 24-32 bits:
Alto from Xerox with local Ethernet (3-10 MHz)
for internal demos, Lisa from Apple (Time magazine's
"Machine of the Year" in 1982) for common use.
- 1979, First mainframe as a mini, the Vax, 32 bits: by DEC
and with the VMS OS, later as a one-card MicroVax.
- 1981, First desktop PC, 16 bits: from IBM with Intel 8086 processor
and MS-DOS from Microsoft, 640 KB memory.
- 1984, First Unix workstation, 32 bits: from Sun
Microsystems, with a Motorola 68000 CPU.
- 2001, Oct. 1st, music-playing IPod from Apple: can have 4 GB
of external storage at a teenager-affordable price.
Applies MPEG format to represent music and video.
- 2006-2007, laptop PC: comes with 200 GB
external disk and 2 GB internal memory, weighing 2 kg, at 1000 USD.
- 1971, First microprocessor from Intel: the 4004
chip with 2300 transistors,
aimed at electronic pocket calculators, but mostly used in
computer games, and with 0.11 MIPs (million instructions per second).
- 2006, Intel's Pentium processor: has two parallelized
computing cores and yields theoretically 20 GIPS. Furthermore, its
Itanium processor has 1.7 billion transistors.
- 1992, GSM mobile telephony, 8 kHz (2nd Generation mobile):
Global System for Mobile Communication (GSM) was designed,
simulated and tested by
SINTEF/NTH in 1982-1988. Has replaced the previous NMT-450 and
NMT-900 (1st Generation).
SMS (Short Messsage Service) was added in 1994.
- 2004, UMTS mobile telephony, 200-1000 kHz (3rd Generation):
Universal Mobile Telecommunications System (UMTS)
is under deployment in most OECD countries.
- 1997-2007, IEEE 802.11, 10-100 MHz: for threadless WiFi,
mainly inside buildings and inner city areas.
- 2006, An optical fiber: can transmit 1 Terabits/second, and
48 fibers are twined in a cable, so
6 Terabytes/second for the cable.
The above examples are
all on the hardware side. The ACM
now operates with 14 knowledge focus groups
in their Curriculum for CS studies:
- Discrete Structures (DS)
- Programming Fundamentals (PF)
- Algorithms and Complexity (AL)
- Architecture and Organization (AR)
- Operating Systems (OS)
- Net-Centric Computing (NC)
- Programming Languages (PL):
e.g. syntax and parsing, semantics and compiling.
Languages like Lisp, Fortran, Algol,
Cobol, Pascal, Modula, Ada, Smalltalk, C, C++, Java, and Php.
- Human-Computer Interaction (HC)
- Graphics and Visual Computing (GV)
- Intelligent Systems (IS)
- Information Management (IM): including databases.
- Social and Professional Issues (SP)
- Software Engineering (SE)
- Computational Science (CN)
16. Appendix E: Gallery of Persons related to Computer- and Internet
Technology
Computer (pre-)pioneers 1930-1960
- 1928, David Hilbert (German, 1862-1943):
On a mathematical congress in 1900, Hilbert launched 23 unsolved
mathematical problems as challenges for the mathematical community in
the century to come. The 10th problem dealt with diophantic equations, close
to computability.
By mid-2007,
12 of these have been resolved, 4 remains unresolved, and 7 are partly
resolved, disputed or reformulated (source: Wikipedia).
In 1928 Hilbert also posed the Entscheidungsproblem (in German) or
decision problem (in English), as an elaboration of the 10th
problem above.
This problem "... asks for a computer program that will take as input a
description of a formal language and a mathematical statement in the
language and return as output either "True" or "False" according to
whether the statement is true or false. The program need not justify
its answer, or provide a proof, so long as it is always correct. ..."
(cited verbatim after Wikipedia). As we can see below, this problem
was quickly resolved as not feasible
in 1936-1938 by Alan Turing and Alonso Church.
- 1931, Kurt Gödel (Austrian, 1906-1978):
proves that any formal logic system - powerful enough
to express simple arithmetic - is potientially:
- Incomplete: i.e. there are statements/theorems being true,
but which cannot be proven so within the given system. Just consider the
previous phrase - "there are ... given system." - which gives "trouble"
whether it is false or true!
Thus, truth and provability are not the same.
- Inconsistent: but this cannot be determined within the
context of the system.
- 1936, Alan Turing (British, 1912-1954):
In the seminal paper - "On Computable Numbers, with an Application to
the Entscheidungsproblem" [Turing90] - Turing proposes a general computing
machine, later termed the "Universal Turing Machine" (UTM).
An UTM is a finite-state machine with an infinitely long tape
serving as extra memory, see
Appendix B: More on Turing Machines.
Turing hypothesized that an UTM could simulate any other
computer, or indeed any other UTM. Turing then proves that such a UTM may
not always stop to execute its program, the so-called halting
problem. This means - see below for
Alonso Church - that a UTM may not always be able
to determine by itself whether a mathematical statement is true, or
whether two programs are identical. Indeed, the Power of the
Machine can be proven limited at the outset.
By this, "Turing reformulated Kurt Gödel's 1931 results on the
limits of proof and computation, replacing Gödel's universal
arithmetic-based, formal language with a simple, formal device - the
Turing Machine" (source: Wikipedia). No wonder that the
American Association for Computing Machinery (ACM) has named its
most prestigious award after Turing - close to being
the Nobel Prize in Computer Science.
The award was, by the way, given
to the Internet creators Kahn and Cerf in 2004.
- 1938, Alonso Church (American, 1903-1995):
worked with Lambda calculus and extended
Turing's theorem. He proved that it is not a
mechanical process to determine whether a general
mathematical statement, i.e. with recursive functions, is
true or false - but a human may possibly do it. This is the
"entscheidungs" problem.
- 1945, Vannevar (pronounced "van-NEE-ver") Bush (1890-1974):
publishes in the
Atlantic Monthly the article As We May Think, where he
visionarily describes an automated library system called memex.
Bush was primarily a science administrator, and proposed a plan
for the Government to fund basic research at the universities, via
an upcoming National Science Foundation (NSF from 1950) and the
military (ARPA from 1958). Source: Wikipedia.
- 1948, Norbert Wiener (1894-1964): invents the field
Cybernetics to use technology to extend human capabilities.
- 1956, Darthmouth Artificial Intelligence Conference: discusses the
consequences of the exponential growth of technology.
- 1962, Marshall McLuhan (1911-1980): coins
the term global village in his book The Gutenberg Galaxy,
where everybody is connected somehow. McLuhan was a Canadian communications
theorist, who is mostly known for the phase
the media is the message.
Famous ARPANET Managers
Some "ARPANET managers" in rough chronological sequence are:
- Joseph Carl Robnett Licklider ("Slick", 1915-1990):
both engineering and psychology degree.
He came to ARPA from project MAC, via BBN.
He served as visionary IPTO chair in 1962-1964, with
grand plans for the Intergalactic Computer Network (now Internet?).
In 1968-1985 he returned as
director of project MAC and professor at MIT.
- Ivan Edward Sutherland (b. 1938-??): Ph.D. in EECS from MIT in 1963
(25 years old), and recipient of the Turing award in 1988 for his work on
Sketchpad, an interactive graphical editor. He was head of ARPA
in 1964-1966, after Licklider returned to MIT and before Taylor
took over.
He later was at
Harvard (1966-1968), University of Utah (1968-1974) which was an early
ARPANET node in 1969, Caltech (1974-1978) where he founded its
CS department, and
recently adjunct professor at UC Berkeley (2005-2008).
He co-founded Evans & Sutherland in 1968 to develop specialized hardware
and software for advanced computer graphics.
He also co-founded Sutherland, Sproull and Associates in 1980 for
consulting services on computer graphics.
The latter company was in 1990 purchased by Sun Microsystems, where
he now is a Vice President and Fellow.
- Robert Taylor (b. 1932): BA and MA from University of
Texas. IPTO leader in 1966-1970 and ARPANET reponsible
1968-1970. Later he established and managed the Computer Science
Laboratory at Xerox PARC in 1970-1983 (spin-offs: 3Com, Apple, Adobe,
...). Then founded and run the Systems Research Center (SRC) of
Digital Equipment Corporation in 1984-1996, until his retirement.
- Lawrence ("Larry") G. Roberts (b. 1937): Ph.D. from MIT in
1963, and first worked at Lincoln Labs on computer
networks. Inspired by a meeting with Licklider in 1964, he finally joined
ARPA in 1967, and became IPTO chair (after Taylor)
and ARPANET chair in 1970-1972.
The ARPANET project was approved by ARPA on June 21 1968, whereupon Roberts
hired the future designer of TCP/IP, Robert E. Kahn, who had
worked on the Interface Message Processor?? at BBN.
Roberts left IPTO in 1973 for Telenet, the first
packet switching network carrier, which was sold to GTE in 1979 and
became part of Sprint. Since 1982, Roberts has worked in DHL,
NetExpress, Packetcom, and Caspian Networks.
- Robert E. Khan (b. 1938): Ph.D. in 1964 from
Princeton. He then worked at Bell Labs and MIT. He then had a leave of
absence to work at BBN with the ARPANET system design.
He joined ARPA to be the director of IPTO in 1972-1985 and main
responsible for the ARPANET project in 1972-1980??. In 1986 he
established the Corporation for National Research Initiatives,
onto which Cerf joined in 1986-l994.
In December 1997, President Clinton presented the U.S. National
Medal of Technology to Khan and his partner, Vinton G. Cerf, for
founding and developing the Internet.
Similarly, he and his partner got the ACM Turing Award in 2004,
regarded as the "Nobel Prize in Computer Science".
Finally, Khan and Cerf got the President's Medal of Freedom in 2005 by
George W. Bush.
- Vinton Gray Cerf (b. 1943): B.Sc. in Math/CS from Stanford
in 1965. At IBM in 1965-67. MSc in 1970 and Ph.D. in 1972, both in CS,
from UCLA and with prof. Kleinrock as his advisor.
Worked on network protocols at UCLA in 1967-1972 and at Stanford in 1972-76.
He designed TCP together with Robert E.
Khan in 1973-1974, and the same two invented IP in March 1978.
In 1976-1982 he worked at ARPA on Internet architecture and security.
In 1982-2005 he worked for MCI, but on leave in 1986-1994
at the Corporation for National Research Initiatives (run by Khan).
From 2005 a Google Vice President and Chief Internet Evangelist.
He has recently been working on the Interplanetary Internet,
together with NASA's Jet Propulsion Laboratory since 1998.
Co-founder in 1992 of the umbrella organization ISOC (Internet SOCiety),
to serve as a civil and formal "anchor" on top of IAB, ITEF and ICANN
and to create public awareness on Internet issues.
In 2000 he volonteered as chairman of the board in ICANN,
and in 1999 and 2001-2007 as ICANN board member.
In December 1997, President Clinton presented the U.S. National
Medal of Technology to Cerf and his partner, Robert E. Kahn, for
founding and developing the Internet.
Similarly, he and his partner got the ACM Turing Award in 2004,
regarded as the "Nobel Prize in Computer Science".
Finally, Cerf and Khan got the President's Medal of Freedom in 2005 by
George W. Bush.
- Barry M. Leiner (1945-2003): Ph.D. in 1973 from Stanford.
At ARPA he managed the Internet Project in 1980-85.
- Stephen Wolff (b. ??), at NSF: Ph.D. in 19xx from XX.
When ARPA gradually reduced its ARPANET effort in 1983-1990, NSF gradually
stepped in on the Internet side.
Wolff from NSF managed the operative and public Internet in 1985-95,
after which it was deemed sufficiently commercialized.
NSF contributed 20 mill. USD for software development
at US computer manufactors in 1985-1995 to support TCP/IP
protocols on their computers.
NSF also upgraded the net to serve as communication
channels to and between US supercomputing centers.
More network pioneers
- Above all, there are the four Internet "fathers":
Robert E. Khan, Vinton G. Cerf, Leonard Kleinrock,
and Lawrence G. Roberts.
- Paul Baran (b. 1926): invented packet switching at RAND in
1962-1965, with "war"-robust routing protocols
(FYI: the latter had no implication for the Internet development).
- Donald W. Davies (1924-2000): co-invented packet switching
in 1965 at NPL, small NPL net in 1967.
- Leonard Kleinrock (b. 1934): MSc/Ph.D. in years 1957-1961
from MIT, later UCLA professor. Expanded queueing theory and applied
it to packet switching and thus laid the base for ARPANET.
Kleinrock published an extended version of his Ph.D. thesis from 1961
as a book in 1964 [Kleinrock64]. This evolved over the years as two
classic textbooks, respectively,
"Queueing Systems. Volume 1: Theory" [Kleinrock75] and
"Queueing Systems. Volume 2: Computer Applications" [Kleinrock76].
- Jonathan ("Jon") Postel at ISI (1943-1998): Ph.D. in CS from
UCLA in 1974. Worked then steadily at ISI, where he edited RfC
documents from 1969 till his death in 1998. From 1970-1972 he
managed name spaces for ARPANET on contract from DoD/IANA and later
from ICANN. Died during heart surgery in October 1998.
- Robert Melangton Metcalfe (b. 1943): co-invented Ethernet
in 1973, using standard coax cables to support a LAN of Alto workstations
at Xerox Parc in Palo Alto, first at 3 MHz and later at 10MHz. Founder of
3Com in 1979 and quit Xerox.
Metcalfe's law states, that the value of a
telecommunications network is proportional to the square of the
number of users of the system (n**2).
David Crocker (b. 194??): PhD from UCLA around 1975.
Early ARPANET pioneer and best known as the author
of RFC 822 that defines the Internet protocol suite.
He is cynical about the size of the IETF vs. that of
the ANSI EDI subcommittee. He is now at Silicon Grahics.
Stephen Crocker (b. 194??): PhD from UCLA around 1975.
Early ARPANET pioneer and initiated RfC and the NWG.
Now working for ICANN on security issues.
Free and Open Software missionaires
- Richard Stallman (b. 1953): BA in Physics from Harvard in 1974,
abandoned Ph.D. studies in AI at MIT in Jan. 1984 after working
there since 1974.
Developed EMACS text editor, QCC compiler, GDB debugger;
initiator of the GNU (Gnu is Not Unix)
"Unix project" and of the Free Software movement where he is President.
- Eric Steven Raymond (b. 1957): Revealed confidential
Microsoft documents, boasting of the quality of Linux, to the public.
Open Source propagantist with the book "The Cathedral and
the Bazaar", but with quite extreme attitudes to weapons and blacks.
- OSS examples: over 150,000 software items in the
www.SourceForge.net portal alone, and 10,000s in similar portals,
e.g. software as:
Lamp platform (Linux, Apache, MySQL, PHP),
Gimp and Trolltech graphical library,
CVC version control system, Gentoo release management system,
Bugzilla bug reporting system.
Netscape => Mozilla in Nov. 2004.
New Firefox browser from a New Zealand team.
Unix pioneers
- Kenneth ("Ken") Thomson (b. 1943) and Dennis MacAlistair
Ritchie (b. 1941):
created Unix operating system in 1969 at Bell Labs on a PDP-7,
later ported it to PDP-11 and gave away the system with full
source code to research institutions.
Digression: ATT tried to patent Unix in 1985, but eventually sold
the Unix rights and software to SCO, who again resold it
to Novell. During the patenting process
ATT found out that all source code had to be made
publicly available as part of the patent!
ATT also discovered trivial copyright violations just by
starting (booting) a desktop computer over
a net of LAN-connected computers!! The patent application was therefore
withdrawn.
- William ("Bill") Nelson Joy (b. 1954):
Ph.D. from Berkeley in 1983, improved Unix implementation on
PDP-11 in 1977, TCP/IP on Unix,
later ported Unix to VAX as the v4.1BSD delivery in 1981.
Co-founder of Sun Micros
ystems in 1982.
- Linus Torvalds (b. 1969): master in informatics from Helsinki
University (1988-1996).
Creator of Linux Unix kernel in 1993 (his share is now
2%!), and indirectly of Open Source Software. Recently at Redhat
which sells Unix services.
Web pioneers
- Tim-Berners Lee (b. 1955): Invented and implemented web at
CERN in 1990-1991, later director in W3C consortium from April 1995.
- Robert Caillau (b. 1947): co-inventor of web at CERN.
Norwegian Internet Contributors
- Pål Spilling (b. 1934):
then at FFI, participated in 1975 in the first intercontinental
performance test of TCP, i.e. to UCL in London from Stanford
University, USA.
- Dag Belsnes, UiO (b. 1941):
helped proving TCP/IP protocol at Stanford in 1974.
- Yngvar Lund, FFI and UiO (b: 1932): ARPANET promoter in Norway around 1980.
- Harald Tveit Alvestrand, then at UNINETT (b. 1959): leader of
IETF (Internet Engineering Task Force) in 2001-2005. Before that, he
was a member at the Internet Architecture Board.
The first three were among the 30 persons invited to celebrate the 30th
anniversary of the Internet in autumn 2007.
Re: [Rasmussen07]
Really a delightful, scholarly and captive book! I learned a lot,
and read even more from other sources afterwards, mostly taken from
the Internet/Web - where else! :-)
Below comes a page-sorted list of minor comments, mostly typos.
After that comes some editorial and then some
more general comments, partly pertaining to earlier policy issues:
P1 (rights to decision making e.g. in ICANN), P2 (unhindered net access),
P3 (bundling of net and contents) and P4 (media ownership).
List of minor comments:
- p.28: "crechendo" => "crescendo".
- p.30: Mention also Jon Postel, Steve Crocker and David Crocker
among the UCLA graduate students around 1970.
- p.30: Mention how RAND lost most of their public contracts,
reborn as ISI.
- p.33: The stuff on "entscheidung" (norsk: "beslutning", ikke "skjelning")
and "Turing machine" is not understandable even for experts,
so please rewrite.
- p.37: "skal ut og fly" => "skal ut å fly", twice at the top.
- p.39: Name explicitly the first four IMP sites: UC Barbara, UCLA, SRI
and Univ. Utah - and their projects and contact persons.
- p.42: "webside" => "nettside".
- p.4x: "webserver" => "nett-tjener (and elsewhere).
- p.45: Please explain "LAN".
- p.45: In case of a busy Ethernet-bus, say that the delay time
before another bus-access request, is doubled before each
retrial - in order to ease congestion.
- p.48: MILNET and ARPANET: split in 1983, not 1982?
- p.50,middle of page; "var buffer" => "fungerte som en buffer".
- p.54: Say that Cerf got a PhD from UCLA in 1972.
- p.56: "Sammentenke" => "Sammenlenke".
- p.61: "Tannum" => "Tanum", satellite link station in Bohuslaen, Sweden.
- p.63: "CEC" => "DEC"?
- p.72,nine lines from below: "annen funksjonalitet" =>
"annen ekstra funksjonalitet" =>
- p.91: "QoS" is a "neutral" measure of some aspects
(e.g. reliability, robustness, performance, ...) of an ITC
service. It is up to
the service user to decide whether the reported measure is
acceptable, e.g. by setting a max/min value for some QoS aspects.
- p.91: "header" => "hode"!
- p.94: "selverhvervede" => "...erverv...".
- p.103 Mention that UC Berkeley Ph.D. graduate, Bill Joy,
co-founded Sun Microsystems in 1984.
- p.108: "GNUs not Unix" => "GNU's not Unix".
- p.120: "Open source" => "Open Source".
- p.120: "reliable/reliability" => "pålitelig/pålitelighet" =
"The probability for incorrect behaviour/crashes
(vs. documented requirements) in an IT-system in a given
usage/operative context and time period".
E.g. a probability of 10**(-4) means one hour
of "down" time per year.
Very seldom to get under 10**(-6), and often dominated by
non-IT errors ("force majeur") or human operator errors.
- p.120ff: "gaveøkonomi" => "dugnads.." - word not in English!
- p.127: "Len Kleinrock" => "Leo K..."?
- p.130: Network Working Group (NWG) established in 1969, not 1972?
- p.137: "maskin og programvare" => "maskin- og programvare".
- p.137: The TCP/IP split-up was in March 1978, not 1980.
- p.138: "en enhelhetlig" => "en helhetlig".
- p.138: Almost incredible bargaining with certain domain names:
.am, .fm., .cd etc. !!
- p.153: "Trusted Computing" - forget it!
- p.157: "Creative Commons" i.e. "Dugnads ..." (no English term for this!)
- p.165: "server" => "tjenermaskin", her "adressetjener".
- p.169,middle: "websted" => "nettsted",
"webadresse" => "nettadresse". - And elsewhere.
- p.176: Say explicitly that WIPO was established in 1967.
- p.178: "hverken" => "verken".
- p.180,line nine and eleven from top: "WIPI" => "WIPO"?
- p.180,middle: "dagens ording" => "dagens ordning".
- p.182,middle: "nettets kode" is ambiguous: "programvare" or "codeks"?
- p.184,bottom: "kode er politikk" - similar ambiguity.
- p.185,middle: "mindr" => "mindre".
- p.186,top: "åpen arkitektur best effort" =>
"åpen arkitektur, best effort" =>
- p.186: Confusing use of the word "kode". "kode" in the meaning of
formats/structures exists on all three levels.
- p.186,bottom: "Ogsp" => "Også".
- p.187: Say that in the telecom network, most advanced logic and
features lie in the inner, kernel layers -
but opposite in the Internet with its enduser-to-enduser
architecture, see discussion in [Saltzer84].
So are all the recent QoS
features posed to the Internet (price-dependent quality, high
streaming efficiency, flexible
security/authentification policies, ...)
really implementable by mere
add-ons in the outermost Internet layers?
- p.191: "digitale skiller" => "digitale klasseskiller" (e.g.)?
- p.191: -------- Noter: (mainly not commented)
- p.200,note 47 on object-oriented (OO)
programming by Dahl and Nygaard:
"Deres ide var at programmeringen kunne gjøres enklere og
mer fleksibelt ved at små subrutiner med hukommelse kunne
brukes i programeringen. Tanken var helt uvant og brøt
helt med den sekvensielle skritt-for-skritt-prosessen som
programmeringen innebar ...".
- Well, most non-OO subroutines (like those in
FORTRAN) are organized in a rather "flat" name space.
I will use the word "linear",
not "sequential", about such a list of subroutine definitions.
In OO languages, the subroutines are defined inside
classes that stand in a hierarchy, so the flat name space is gone.
But most non-OO and OO languages are still supporting mostly
sequential (not parallel) execution of subroutine
calls. So your text is imprecise.
- p.200: "Programmene Simula68 og ..." =>
"Programmeringsspråkene Simula67 og ...".
- p.200,note 71: "Java(CLI)" => "Java(C#)" - pronunced C-sharp.
- p.210: -------- Litteraturliste:
- p.211: "CarpenterB" => "Carpenter B".
- p.212: Daling etc.: a PhD thesis?
- p.212: Gilder ... Metcalfes Law" => Gilder ... Metcalfe's Law".
- p.213: "Hamm, Ingrid, Macel" => "Hamm, Ingrid og Macel".
- p.214: "Jackson ... Technologogical" => "Jackson ... Technological".
- p.215: "Moody ... WIRED" => "Moody ... Wired" (as earlier)?
- p.215: "Moscovitis .. et.al." => "Moscovitis .. et al.".
- p.215: "Postel Jon, og" => "Postel, Jon og".
- p.215: "Raymond ... revolutionary" => "Raymond ... Revolutionary".
- p.215: "Reed, David P. Jerome" => "Reed, David P., Jerome".
- p.216: "Reid ... 1000 Day" => "Reid ... 1000 Days".
- p.216: "Reingold ... Cambiridge" => "Reingold ... Cambridge".
- p.216: "Saltzer Jerome H et al." => "Saltzer, Jerome H et al.".
- p.216: "Segaller ... LLC" (Limited Liability Company) =>
"Segaller ... L.L.C." (as earlier?
- p.216: "Streeter ... incoporating" => "Streeter ... incorporating".a
- p.218: ------------- Ordliste:
- p.218: "BBN Bolt Beranek and Newman" => "BBN Bolt, Beranek and Newman".
- p.219: "GNU ..." - NB: It is Linux that is the kernel!
- p.220: "NCP ... ARPANETs program for" => "NCP ... ARPANETs
protokoll for".
- p.221: "Nettsted En samling nettsteder ..." (circular def.) =>
"Nettsted En samling nettsider ...".
- p.221: "Node En side ..." => "Node En nettside ...".
- p.221: "Operativsystem ... ustyr" => "Operativsystem ... utstyr".
- p.223: ------------- Register (over personer og organisasjoner):
- p.224: Heart, Frank (BBN employee): please include.
- p.225,228: IANA: placed wrongly under letter "T".
- p.225: ISP (... Providers) => ISP (... Provider).
- p.226: Joy, Bill (Sun co-founder): please include.
Three editorial comments:
- Protocol example: make a static one to show
OSI-style layering, and a dynamic one to demonstrate
how a single message is split up in standardized packets,
which are sent through the layers, cf. above.
- The glossary ("ordliste") should couple persons better
to institutions, and inversely.
Ex. Vint Cerf worked at ... (whole list?).
- Make a good timeline / chronology with major events indicated.
Four more general comments on policies:
- Discuss innovation vs. technology - to what degree
can such be "planned"?
- Point P1 on technocratic vs. democratic control:
Discuss different stakeholder roles, e.g.
technology developer (academic researcher, hacker, IT industry),
international standardization organ (ITU, ISO),
international "government" (UN, EU),
national government (political actor in office, ministry),
national control body (ICANN, FNC)
Internet provider (Telenor), content provider (Disney),
end-user (any citizen or organization), ...
- Point P2 and P3 on "bbundling": Discuss overlapping roles between
infrastructure/Internet providers (CanalDigital)
and cultural agents (bookstores)
vs. content providers: split these to avoid conflict of interest
(NRK, TV2, Digitalt bakkenett, CanalDigital).
- Point P4 on media power structure: Discuss media owning or controlling
other media (Schibsted, TV2, A-pressen).
So on points P2-P4, stricter national legislation seems in place.
Re: [Friedman06], but also see [Hira05].
++??
- reidar
www.idi.ntnu.no/~conradi/internet-history.html
RC: 28 Nov. 2007, 10 Dec. 2007, and 1 and 4 Jan. 2008.