Two-Sided Marketing to Incent Internetworking Networks of Things

Two-Sided Marketing to Incent Internetworking Networks of Things – HKUST IAS 3. 3. 2015 Jon Crowcroft, http: //www. cl. cam. ac. uk/~jac 22 http: //nymote. org

History of Mobile Cloud n Smart Phone disintermediates Cellular n n But wait … we missed something Internet was meant to democratize n n Cloud/Web disintermediates telco Cloud/Web end run ISPs Flatter, P 2 P, Decentralized Devices (in pocket&home) powerful enough Uplinks nearly fast enough now How to reboot?

Old school capitalism & socialism n See Future Perfect: The Case For Progress In A Networked Age , by Steven Johnson n § http: //www. amazon. com/dp/1594488207 There’s a good case made for new ways of doing business § § Not a Legrand Star or a Hayek Hub, but A Baran Web….

So what’s this to do with startups? n Need the right eco-system n n n No “one size fits all n n n Not just government initiatives Not just a technology innovation Cellular different from internet Biotech different from materials Are there special network features n For Internet, Cloud and Io. T?

Taking sides…Big v. Small n Alec Broers wants Big Companies n n n Andy Hopper likes what we do n n n Like Intel, IBM, etc ex IBM, VC Cambridge, now chair of house of lords ctee on S&T Head of dept CL 10 startups, 2 IPOd…. Who do you believe? n Time for science (experiment w/ biz model)

Cambridge has been doing new business cases for some time n ARM n n n Real. VNC n n n Mostly fabless chip designer More chips than intel out there… Open source desktop virtualisation Millions of users… Ubisense n n Actually, fairly traditional h/w -> IPO Used by BMW on production line.

More radical n Xen n n Open Source - the Cloud (Amazon EC 2) Engaged with ALL of industry (not just 1 preferred initial customer) Intel AND AMD Microsoft and Sun/Oracle and HP … Acquired by Citrix for 500 M USD. n n n Pretty much standard value in expertise 50 kernel hackers with Ph. Ds: ) Big 2 nd customer was Tescos… … …

More radical still n Raspberry Pi n n HAT -- Io. T n n n Charity, 0 staff, 0 VC, 0 offices 5, 000 units shipped in 3 years. Doubled application rate to University CS Building Open Buildings Internet of things - to be seen Nymote -- Cloud n n Mirage, Irmin Signpost … … …disintermediation Open Source, decentralised, appstore

HAT: Internet of Things n Thing 1 - we need an app store to network legacy (and new) appliances n n Thing 2 - we need a new app store for people to contribute apps to federate (and isolate) Systems of Things n n Commodity Appliance Thing Home Appliance Thing See Dr Seuss for more details: http: //www. youtube. com/watch? v=fj_xwg. LW_4 I

Internet of Things n Need a grass roots (cottage industry) n n Therefore need to give people tools Equivalent of knitting needles+wool n n n Hence need 3 D Sculpters & Printers n n n Or sewing machines and thread (better) Or even looms But also need to give them upload capability Moral equivalent of knitting pattern contributions made to magazines & books But 3 D CATs uploaded to websites 3 D HATs are easy (c. f. Fresnel project: ) Hence need 3 D scanners too

Render Unnecessary http: //fffff. at/free-universal-construction-kit/

Intellectual Property is Theft n n See aforesaid book for why Patents are not capitalism, n n n they are monopolistic Long patterns incentivize companies to hide background longer term knowledge 10 year pharma pattern is longer than the survival rate of someone’s creativity So blocks the sideways flow of innovation Especially in methodologies

Alternatives n n Xensource worked without IP protection because…. A big incumbent cannot catch up with 5, or even 50 smart people working on an exciting new idea n n n Throwing 500 people (at MSR) doesn’t work Old news (c. f. Mythical Man Month) So novel S/W doesn’t need any protection than raw speed of development

Some possible exceptions n Modest, but useful, genuine h/w innovations n n e. g. Dyson bagless vacuum cleaner Patent lifetime perhaps could be shorter Invention to market cycle*2 == 5 years? Peer Patent Review n Deals with scale out of n n Finding prior art Detecting obviousness << patent >> sufficiency

Employee Owned Business models n Good karma n n n Works outside of business too Incentive alignment without toxicity What if society is the employer? n n So government is the employee So how to use EOB in government?

Mezzanine Conclusions, Discussion n Today HAT, REMS, Ocaml Labs n n n http: //www. cl. cam. ac. uk/projects/ocamllabs/ Platform (free), toolchain (free), service (tbd) Tomorrow, n n http: //hubofallthings. wordpress. com/ http: //www. cl. cam. ac. uk/~pes 20/rems/ World Peace and an end to Hunger? Live long and prosper

DO$H – Decentralized Object Storage Help (need better name ) n n n How to incentive decentralized system? Lets think about monetizing … Not just token payment…

Background to P 2 P and Virtual Currencies n Peer-to-peer systems avoid infrastructure n n n Eschew centralised ownership/management Examples: Internet (originally), Usenet Mesh wireless nets Mobile Ad Hoc Nets (MANETs) Oppportunistic/mule nets are examples

P 2 P #2 n Mutual exchange of resources n n n Storage n n Bit. Torrent, Freenet, Eternity Computation n n Layered on another (infra or p 2 p) net Structured v. Unstructured, DHT, Key, Value stores, in Cloud (Cassandra etc) SETI@Home, Climate. Prediction. com Service (e. g. Presence) n Skype (originally)

Virtual Currencies n No currency - direct barter n n Air Miles Subway tokens Exchanges – credit cards, paypal New: : Bit. Coin, Ripple, Properties n Decentralised Mint n n Decentralised Verification n n no 0 wning/coercion mutual benefit in verifying == p 2 p store/check Non-inflationary (Bit. Coin)

Mint/Verify Bit. Coin - crypto n Basic trick is “proof of work” n n P 2 P verification entails keeping history n n Transaction chain->not strictly anonymous Various possible problems like hoarding n n Mitigates both forgery&double spending Plus finite total (eventual) number of Bit. Coins -> possible loss impact “Alien Technology”

P 2 p incentives and currencies n We’ve been trying to get incentive alignment in p 2 p n n n Bittorrent uses tit-for-tat tokens Did same in our work on mesh wifi With deflationary currency to deal with n n n People leaving with money in their pocket People joining/starters… Then along comes bitcoin (and ripple) n But….

Objections to Bit. Coin 1. 2. n n Proof of work is a waste of energy Not anonymous (at least not as much as some people think) We don’t like 1, but we don’t mind 2. So……

DO$H: an anonymous idea (by me) n To fix some elements of bitcoin n n And p 2 p storage systems Specifically, n n n Want to fund the Personal Cloud vision with backup/resilience other goodness properties

Personal cloud n Everyone keeps their personal data n n n But want backup (or nearby copies) n n In their pocket In their home hub, or car or bike Could crypt and put in cloud Or crypt and put in friends/family Or even (many) arbitrary other users stores How to pay for cloud storage/cpu?

Cloud provisioning: service offerings n Already work on tailoring services to particular constraints n n Already offer services based on user/tenant locale n n Not only for performance, but also security, rights management, etc. (e. g. i. Player) Providers already manage their infrastructure n n Differential privacy: tailor query results to not reveal too much private information Customising service and content for regional concerns Thus, already the capability to tailor services for particular regional and/or jurisdictional concerns 26 26

Cloud provisioning: Unikernels n Cloud exists to leverage shared infrastructure n Isolation is important: n n VMs – Separate for tenants, complete OS, managed by hypervisor Containers – shared OS, isolated users n Deployment heavy, isolation overheads, … n Future? Unikernels: n n library OS, build/compile a VM with only that required Hypervisor managed, removes user-space isolation concerns 27 27

Cloud provisioning: Unikernels (2) n Very small (1 MByte), lightweight easily deployed VMs: n Easily moved around the infrastructure n n Deploy in locales/jurisdictions when/where relevant Facilitates customised services n n n Specific unikernels for particular services Encapsulating specific jurisdictional requirements? Transparency: Natural audit trail n “Pulls” that what is required to build, on demand 28 28

Mirage OS Cloud xen 4. 4 arm/intel

Some other Mirage/Cloud components n Irmin storage n n Git API (also others, Imap etc) Immutable (merkel trees) Append only Signposts Connectivity Management n n n Including key distribution And identity management All Open Source …and available now

Runs on one of these… n n Hardware prototype: Cubieboard 2 n ARM Cortex A 20, costs £ 49. Software for private service domains: n Xen 4. 4 released with ARM support. n Mirage/ARM works in userspace, and almost kernel mode (a few weeks).

Monetizing Personal Data n n n Don’t “put all our eggs in one basket” s/facebook/personal cloud/ Monetize data case by case basis n n n Idea also from HAT project Have relationship with many vendors of goods and services – loyalty cards etc They don’t have usage data – we do n in personal cloud – home power, fridge, fitbit, scales, washing machines etc

Model is they pay us n n n Supermarket/pharmacy pay us for data In DOSH Generate DOSH coins by crypting our data – i. e. useful So we get adverts, but they don’t leak data we also get DOSH, and could give BACK in exchange for no ads (just storage)

DOSH is quite like Bit. Coin n But not deflationary – but bounded by people and goods used n n so can’t hyperinflate due to sustainability of world Keep Bit. Coin model of auditable verification chains n so can discourage use for Silk Road #3

Two more ideas n Could source randomness for hashes DOSH protocol from non-co-ercable places 1. 2. Sustainable energy sources (solar/wind) contain natural randomness, which can be observed/recorded but aint easy to fake/force Could also use 3 D printers DRM random source (and so mark real world goods 3 D printed with Bit. Coin verification hash

Have some pieces n http: //perscon. net/ http: //nymote. org/software/irminsule/ Need to do open source DOSH And find some seed users… n You didn’t hear this from me n n n

Who Am I?

By Dr Parry, with apologies to Dr Seuss and the image copyright holders The D. E. in a HAT An Alternative presentation

A home tech platform What ever is that? The Hub of All Things Which we know as the H. A. T.

How to will it work? Well let me tell you, We collect data using things Named Thing 1 And Thing 2

Thing 1 is a smart thing Io. T ready you know We just need to name him And link him And then let him go

Thing 2 makes objects smarter With just a little finance Milk bottles, cups and kettle They Io. T enhance

The things collect data Which algorithms analyse Contextual archetypes appear Right in front of your eyes

We learn from the home And from each other too We learn all about the things And what they tend to do

SMEs play at the HAT-Fest With the data compiled Creating novel offerings Which are deployed in the wild

The Hub of All Things We commend it to you We think it is the future We hope you do too