Lift09 — Vint Cerf [en]

Story of the talking dog. Excitement, but nobody is paying attention to what the dog is saying.

Vint feels like the talking dinosaur.


Lift09 141

The Internet works because there are standards, and people cooperate to work together around those standards.

Jan 1983, 400 computers, official launch of the Internet.

Now: 542 mio hosts, 1.464 mio users. Doesn’t count computers behind firewalls.

Lift09 138

Other relevant phenomenon: 3.5 bio mobiles have come into the system. The first experience of the internet for many people in the world is going to be through mobile.

More internet users in China than in the US

Lift09 139

Asia biggest number of users, then Europe. Hard to make predictions about Europe as it keeps adding countries.

Astonishing: the kinds of things that people put on the net! Fridges and picture-frames! (Sounded about as useful as an electric fork…)

steph-note: internet-enabled picture frames sound cool, to share photos with family around the world

Internet-enabled surfboard. Surf the internet while waiting to surf the next wave, with wifi hotspot on the beach 🙂

Vint seems to have the house of the future: sensors all over the place, network, monitors temperature of the wine cellar for example, and sends and SMS when something is wrong (happened to him when nobody was home, so got one SMS for every five minutes for the next three days of his trip, and when he got back the cellar was a bit warm…)

For the moment, all Vint can tell is that the lights have been on in the wine cellar, but not if somebody took wine. RFID: could put a chip on the bottle, so would know if it walks away. But you could drink it and leave the bottle! So put something in the cork… steph-note: wild!

This year is a significant year of change for the internet. moving to run in parallel with the IPv6 something or other (steph-note: fuzzy for me)

In the seventies, nobody could decide how much address space was needed for this “experiment”. Someone wanted variable length fields, too much computing power. 128-bit addresses (3.4 X 10^38 sounded like a preposterous number of addresses to ask for for an experiment).

The experiment never ended… we’re living in it.

Non-latin characters in domain names — that’s happening now. Hard to integrate that in the current domain name system without disrupting it.

Multi-core chips. (steph-note: technical stuff I’m not following, about clock speeds and chips and stuff and how this relates to the internet)

Conventional relational databases are not scaling up to the sizes people are looking at today. Petabytes of data…

Bit-rot problem: it’s 3000, can you interpret a 1997 Powerpoint? Big big problem. Application software needed to interpret our bits not available anymore. Need to maintain access to application software after support is dropped.

Before: computing utility = big building. Today, big buildings with lots of computers and people use the internet to access it = Clouds. Cloud makers don’t usually worry about dealing without other clouds. No vocabulary to talk about other clouds. How does one cloud communicate with the other? How do you tell Cloud B to protect the data that’s just been copied to it in the way that Cloud A was protecting it?

Privacy and acceptable behavior: how will we agree on what privacy is online, and what is acceptable or not there?

Big clash between copyright policies and the structure of the internet. Legislators and technologists will have to come together to figure out stuff that will hold in a world built by copying.

Digital libraries. Most works will hopefully be available in online form.

In 100 years we’re going to say “can you imagine that at some point, we had books that did not talk to each other?” How do you navigate a dynamic world of books.

Non-Google project here: Interplanatary stuff. Point-to-point transmission is ok if you don’t have to talk to too many devices or spacecraft. Design a space communication system that is as rich as the internet. Why not use TCP/IP? doesn’t work because the distance between planets is astronomical… it takes 3.5 minutes for a signal to propagate to mars (20 minutes at the furthest). TCP/IP is not designed for 40 minute delays. Other problem: celestial motion. Planets have the nasty habit of rotating and we haven’t figured out how to stop that yet.

Disruptive and highly delayed environment. Devised new protocols. Went to test them in tactical military communications because it’s also a highly disrupted environment. DTN.

DTN transfered way more data than TCP/IP, and the marine corps ran away with it. Where is my experiment?

NASA: Deep Impact Testing. launch a probe and get data, but the spacecraft is still going round the sun, so they used it to test data transmission from and to it. Neat!

This summer they upload the protocols to the space station. August, another craft. October, another, so three nodes. By the end of the year, will have formally qualified the interplanatary protocols, and they’ll be able to offer them to standardize communication in space. => interoperability between space missions, if desired!

Next step: interstellar network. But… today it takes long to get over there to the other stars. So need to work on a propulsion system to fix that. Lot of work to be done!


  • Are you happy? Yes, internet shows people are willing to come together and collaborate. And the WWW has demonstrated that sharing information is power. Happy to be at Google, because they’re too young to know “you can’t do that” and so they just go and do it. The reason things didn’t work out 25 years ago might not be true anymore.

  • Can we keep the internet open? Amazing pressures in the network today. At the time, academic geeks who were happy to work together. Pressures to try and control the network and the way people use it. Not necessarily all bad. Privacy questions. Protecting people. Legal system needs to be adapted. Tension between the open internet and being so shut down that nothing is possible. Somewhere in between the network is openly accessible, things can be tried out. Committed to keeping it as open as possibly.

  • Semantic Web, will it become reality? You should ask TBL… Was feeling sorry for TBL because the idea of the semantic web was moving as quickly as IPv6 into the public internet… Link = “something over there that is of interest”. What if we could add a “semantic hyperlink”? Jaguar can be a car, operating system, animal… (steph-note: this is what wikipedia disambiguation pages do) More hopeful.

  • Is Google the real Big Brother? Doesn’t think so and hopes it never does become it. Helping people manage their information. How well is the information managed and protected? Google recognizes that separate access and privacy is important. E-mail is always read by programmes. The one that puts ads in Gmail just does pattern matching.

5 thoughts on “Lift09 — Vint Cerf [en]

  1. Nice set of notes! Thanks. Very interesting. Been randomly reading your contributions for quite a while.

  2. Whee, thanks! Very honoured. I’m sorry the rest of the blog is down. Will try and get it back up this week-end, but I have so much even more urgent stuff (don’t even try to start guessing) to deal with that I just can’t take care of it before…

  3. Trés bon résumé – Vinton Cerf doit vraiment être un personnage hors du commun ! Je ne manquerais pas de faire circuler ce lien pour mes étudiants – J’espère aussi que tu pourra récupérer tes données car ton contenu est bien cool – Thx for sharing !

  4. Merci! Pas de souci pour mes données, en principe j’ai tout (sous forme de divers exports wordpress). Du coup je suis allée jeter un oeil sur ton blog — j’ai la chance de ne pas tellement avoir besoin de ton dernier article, parce que google se jette sur mes articles à peine sont-ils publiés! Mais sait-on jamais… je le garde sous le coude 😉

Leave a Reply

Your email address will not be published. Required fields are marked *