What we did, what didn't ought to have. So I guess this makes me one of the old fogeys of networking - perhaps the strange uncle back from the woodshed, standing here alongside the godfather and grandees of the net, and the young guns. So what I want to talk about is not the Future Internet, or Internet Science, or even Internet Science Fiction (my main hobby - see my blog:) but the way interesting stuff happens, and the way to make sure it doesn't. Interesting stuff happens because of curiosity and random walks through ideas (and encounters with random colleagues, often in other disciplines, or no discipline). Interesting stuff does not happen because of top-down initiatives. Hence the Future Internet programme in Europe (and the US) is almost certainly doomed to produce rather mediocre, small deltas to existing ideas. But enough about me. What was interesting about arriving at UCL in 1980 to do an MSc then a PhD with Peter Kirstein was that he had been written to by the UK funding agency (then called the Science Research Council - sort of the ERC or NSF of the time for Blighty), telling him to "cease and desist this Internet stuff and get on with real networking which would be OSI" - I kid you not - they actually wrote to him Now, this was doubly ironic (not in the Alanis Morrisette sense, but the real English sense, pace, Ed Byrne) in that the SRC didn't fund him - DARPA and RSRE did. Luckily. So he (rightly) ignored them, and went on to be the real Internet pioneer for Europe (and he's still at it deploying satellite access for schools along countries along the old Silk Road, with kids in Kabul and Samarkand now getting at Wikipedia. One of the great things about being at UCL then was that Peter ignored these rules, but he also played devil's advocate to his own work, so we did Big European projects in crazy things like running IP alongside X.25 (CONS) and CLNP (ISO IP) and comparing the performance (and TCP v. TP4:). We built working X.400 and X.500 systems before multimedia (MIME) and Directories (LDAP) became working fare in the IP world. So we did both sides. Just to make sure we were right. And people couldn't accuse us of not trying. Hey, we even ran ATM:) Later on (say 1988) we ran video and audio over IP (I remember listening to Van over Vat in Berkeley, back in 88 as he was building new versions to see how well the audio playout adaption worked over "Long Fat Pipes" as we called our meagre 2Mpbs Internet access link for all the UK (I type this over a 100Mbps ethernet connected to Cambridge's 100Gbps access to JANET - plus ca change, plus c'est les meme reseaux a grande vitesse:) 20 years after doing multicast (and the paper from Sprint saying 10 reasons why no-one would ever do multicast), AT&T and Telefonica run triple-play nets with IPTV being the principle way of delivering video (actually it isn't quite multicast as we knew it, but is sort of). So the first thing I did 20 years after UCL, 12 years ago, on moving to Cambridge was to think "what shall I do that people will tell me is wrong?". Three things: 1. Haggle - with the Intel Research Lablet - we set out to build a system everyone told us no-one wanted. Now people want it - it's a personal delay tolerant net for days when disaster or attack have bought down the Internet. 2. Xen - I actually remember reading though all the old expired IBM VM patents (thanks to IBM for putting them all online nicely readably) to make sure that all the ideas in Xen were already covered so NO ONE could sue - nothing like an expired patent for prior art:) We're now (advertisement) working on a replacement for Xen, as it's too insecure. 3. Raspberry Pi Jim Gettys came over from MIT with two of his cute 100$ laptops (OLPC is an excellent programme). We liked it, but sitting in a pub (where else?) in Cambridge Ebon Upton (Broadcom's hardware engineer who built the RSPI) and I said but 100$ is way too expensive - I bet we could do it for 10 (this "blowing a raspberry" at MIT:) We weren't quite right - it's 25$. But it's cool Oh, no wait - 4 types of torture (_No-one_ expects the Spanish Inquisition:) 4. Mobile Decentralised Privacy Preserving Social P2P Ad Hoc Systems with adverts. So now "we are all Social Scientists" and I study OSNs just like Bala (at AT&T - who wrote _the_ book on WWW and how it really works with Jen Rexford...). But we like the fact we get all this storage and compute power for free. But the cost is a) putting up with adverts (tolerable ) and b) having to cope with the idea that some random Cloud-Company-Employee might do a Bradley Manning on all of the customers' private data and publish it all or c) Cloud-Company-X decides to start charging. So we are building a replacement for all this stuff - friends at Eurecom had a v. good go (see Safebook) but there's more to do with disrupting the incumbent Cloud Mafia. If you tell me I am right, only then will I stop. What's next? well you'll have to ask some of the young guns...