Notes only – no narrative or analysis – Vint Cerf’s keynote at WWW2010 in Raleigh, NC.
I’m the guy who likes to work on the underground plumbing that your tools run on. I’m going to talk about the problems below the level that you are dealing with. Anecdote about highway engineer advising race car driver on strategy to win race.
Picture of the connectivity of the net 1999 – would look the same in 2010 – collaborative entity. I still think it’s pretty amazing to see that this system can be run in such a distributed way.
Number of machines on the net continue to climb – internet domain survey. Probably at least that many are hiding behind the firewalls.
1.8 billion people connected globally (~20%); about 15-20% of mobile devices are internet-enabled.
- Asia – 765.4 million, 20.1% penetration
- Europe – 425.8 mllion, 53%
- North America – 239.6 million – 76.2%
IP addresses – IP v4 will exhaust mid-2011 – we will introduce IP v 6 (128 bit v 32 bit – 3x4x10 -38 v 4.3x 10 -9)
Managing power with smart grid; could help forestall need to expand power grid capacity.
- Mobiles aren’t really always connected so you can’t rely on persistent connection – systems need to be transaction-like (may be disconnected at any moment). The http environment has a nice match for stateless behavior (cookies)
- Wide range of bandwidth — flexibility: have more than one way for the mobile device to be connected to the network
- Mobiles are beginning to be capable of engaging in interactions in sensory environments (video, stills, audio, multi-touch displays) – can be used as sensors (environmental, chemical, biological) — the notion here is that everyone can become a reporter of information
- we are becoming a community of information producers as well as consumers
Beyond Text Search
- Image search – Google goggles let you take an image with your mobile and search our database for similar images — take picture of book and search, find who wrote, title, where to buy — or maybe even a bar code
- Speech recognition — hands-free mobiles can surf the net orally rather than typing things in
- Gestures (Patti Maes, TED) — you just have a camera and projector, it puts the computer in the same environment that you and I normally interact with
- Deep-linking/Semantic web — nascent
** “dark matter” only 5% of the matter of the universe is “us” and in the web there is a lot of structured information that is not visible because it’s not HTML or XML (databases) – we we clearly need to do some serious work to discover this structured information
** web “publication” — historically making info available has historically been done on paper but on the web we should think of structured data in the same way; it can be used, fused
** semantic printing — analog of print publication — print something that has semantic meaning that can be processed by a computer
- Domain Names are not permanent – can disappear – URIs are intended to be more persistent –
- What do identifiers identify? A computer somewhere (URLs)
- What is an object -is a document, a file, an executable?
- How do I interpret the object? How do I know it is OK (integrity,security)? Where did it come from (provenance)?
Authenticity and Security
- Digital signatures – what do they really mean? Can we trust certificates and certificate authorities?
- Passwords – best are one-time use – random number generation is the best
- The worst problem we have is that we make spectacular mistakes as people
- Naive browsers (software not people) — running with too many privileges — Chrome browser is an attempt to isolate processes from each other and from the OS
- BOTNETs are a symptom – browsers are primary vector
- Weak protection of personal data by business, government, users, software — I see why privacy gets invaded
- Videos, audio, pictures – can be taken by anyone – we have created an environment where we can share that info with anyone we want to
- Toll roads with RFID chips — but — if someone puts an RFID detector where you don’t know it is, you can get tracked — but at the end of the month, your bill tells you where you were when — “personal” information
New Technology & Applications
- Internet started 37 years ago
- Massive data correlation – Jeff Jonas, every datum is a query … this will let you know how this data point is related to other things
- compare to early days of the net (1983) — before the net, every network thought it was the only network in the world
- clouds: you can replicate them to make things resilient, share data, collaborative
- we have info in a cloud, how do we get it out – Google calls this data liberation
- telnet defined a network virtual terminal – if every computer could interpret this, it became lingua franca — in a network virtual cloud, you would exhibit yourself to the cloud in an agreed-upon way
- Short Circuit — INNNNNPUT!
- Monotonic trends toward more BW/Memory
- Woodhurst Sensor Net — capture heat/humidity data for manage — wine cellar — if the temp goes about 60 degrees I get an SMS on my mobile — a couple of thousand bottles of wine
Internet Research Problems
- Security at all levels
- Broadcast – we take broadcast channels and turn them into point-to-point links (wifi) — but we could share widely if multiple people want the info — think sending software updates — but protocols are needed
- Moore’s law is broken – heat/power — we don’t see increasing clock speeds anymore, we have multi-core chips — so your crappy software isn’t going to benefit from increasing processor capacity
- What does law enforcement do when someone in one country harms someone in another country?
- The problem with the WWW is that it’s a giant copying engine — browser makes an immediate copy
- The solution to the problem is not in my opinion to exercise huge amounts of DRM
- What ways are possible to compensate people for IP other than copying?
- Open source is not necessarily the “answer” to challenge of OS and application evolution — we have to invent schemes that provide incentive for people to keep that old software around — emulation in cloud?
- Cloud computing makes this an even larger challenge!
- Does v6 allow extra latitude – speak to bandwidth? v6 is not a tool for adding a lot of functionality – the strongest incentive is if we run out of address space the Net can’t grow (like selling telephone without having new phone numbers). There are features that make it more efficient (header has been redesigned). There isn’t much more in security. The only thing in the spec – if someone asks you to go secure you are supposed to comply.
- As the chief internet evangelist, I want the internet to keep growing
- Dangerous to have everything we do attached to one digital identity and need two-factor authentication
- Mobile devices : issues with bandwidth — Google is working on finding places to find gigabit broadband to figure out costs, etc — on the mobile side, the biggest challenge is available spectrum — the current allocation tactic is auctioning. It does a disservice to creativity and competition. I’d like to see more attention paid to license-free shared spectrum. Technologically, we can do much better at sharing broadband than ever before. Policy needs to change. As we move into that space, we can probably deliver a substantial amount of bandwidth. Digging trenches and pulling fiber is very expensive; radio-based transmission is much less expensive.
- What is the salient aspect of configuration that needs to be improved? it’s not easy to tell when a config is wrong, when it’s mis-configured. “This perimeter is out of range” — the thing that’s hard is when things are in the right ranges but the combo causes a problem. “This configure causes battery loss in the mobile” for example. We don’t have tools to figure out this context and relationships. If there were algebraic ways — this one is safe, this one is not — that would be an amazing outcome.