The Future of High Performance Computing

Just FYI, anyone that’s interested in HPC, super-computing, advanced modelling & simulation, problems, prediction, cyber-security and any associated field might be interested in this. It’s on Thursday 23 February in London. Event link here.

Beginning of a new Current & Future uses of HPC map below….

Current & Future Applications of HPC

Modelling & Simulation
Preventing the invention of unnecessary
Prediction of technology breakthroughs
Modelling specific species against climate change
Dynamic longevity prediction
Predicting M&A activity/hostile takeovers
Lifelike recreation of dead actors in movies
Volcano modelling
Real time national mood modelling
Hyper-local personal weather forecasts
Complete human brain simulations
Prediction of social unrest using global social media feeds
Finding holes in existing research
Finding new knowledge in Big Data
Automation of scientific research
Radiation shield modelling
Molecular dynamics modelling
Space weather forecasting
Trawling scientific data to find genetically applicable treatments
Molecular dynamics forecasting
Automation of scientific research
Aesthetics prediction
Seismic mapping of planets
Hurricane forecasting
Modelling of tornado trajectory & speed
Galaxy simulations
Oil well forecasting
Movie special effects
Simulation of fluid dynamics
Virtual crash testing
Re-creation of the origin of the universe
Earthquake prediction
Population growth simulations
Climate change modelling
Aerodynamics design
Whole city simulations
Pollution forecasting
Radiation shield modelling
Molecular dynamics modelling
Modelling impacts of bio-diversity loss
Power grid simulation & testing
Modelling of organizational behaviour
Optimization of citywide traffic flows
Emergency room simulation
Major incident modelling & simulation
Space weather forecasting

Healthcare & Medicine
Dynamic real-time individual longevity forecasts
Mapping blood flow
Prediction of strokes, brain injury & vascular brain disease
Pandemic modelling
Unravelling protein folding
Curing Alzheimer’s disease
Virtual neural circuits
Bio-tech research for SMEs
Acceleration of drug discovery & testing
Decoding of genetic data
Whole body imaging at scale
Remote medical triage
Foreign aid & disaster relief allocation
Dynamic simulations of muscle & joint interactions
Bone implant modelling
Modelling of the nervous system
Longevity prediction at birth
Design of super efficient water filters

Pre-trade risk analysis
Bond pricing
Real-time hedging
Fraud detection
Self-writing financial reports
Automatic regulatory control & compliance
Pre and post-trade analysis
Dynamic allocation of government tax revenues
News prediction
Flash crash prediction
Optimisation of investment strategies
Automated hiring & firing of employees
Automated due diligence for M&A
Whole economy simulation

Software & data
Software that writes itself
Holographic data storage
Coding for ultra-low energy use
Data that generates its own models

Engineering, materials & manufacturing
Space station design
Space colony design
Design of new aeronautics materials
Zero gravity manufacturing & design
Predicting properties of undiscovered materials
Design of smart cities
Identification of redundant assets
Optimization of just in time manufacturing
Optimization of crowd-sourced delivery networks
Design of ‘impossible’ buildings & structures

Recording of every individual human conversation on earth
Modelling of factors likely to lead to a revolution
Deliberate cyber-facilitation of revolutions
Breaking 512-bit encryption ciphers
War forecasting algorhythms
Virtual nuclear weapon testing
Modelling behaviour of terrorist suspects
Crime prediction down to individual streets
Identification of terrorist suspects
Forecasting of geo-political upheavals
Hyper-realistic war gaming
Simulation of large scale cyber attacks
Missile trajectory simulation
Screening of data from multiple spectra & media in real time
Threat detection
Crisis management decision support

Note: This is just me going off on a bit of a jazz riff at the moment. All subject to change!

A scenario for the future of insurance

Here’s a true story. A few weeks ago I decided to take one of my old cars for a run. It’s a very old car and if it isn’t run regularly things start to go wrong with it. It was the first dry day in weeks, although there was a heavy frost. The run was fine, although it wasn’t long enough so I decided to extend it. Long story short, I hit some black ice on a bend. I wasn’t travelling fast – 20mph perhaps – but I ended up on the wrong side of the road in front of a van coming directly at me at a similar speed. We missed each other, but I ended up in a hedge and did quite a bit of damage to my car.

Here’s an alternative scenario. My insurance company is well aware of the weather in my area. In fact they’ve been alerted by three local drivers that they’ve hit trouble. So when I open my garage, or possibly before, I receive a text saying that there have been three accidents in the area in the last few hours and it’s recommended that I don’t take the car out until any ice has melted. Maybe I’d get a tiny discount for not driving my car on this particular day.

In the future insurers will have a far better understanding of risk, much of it in near real-time, because of the devices we constantly carry around with us – phones especially – and due to ubiquitous smart sensors. Eventually there will trillions of these tiny sensors reporting on just about everything all of the time. The data these devices capture will be used to predict behaviour, which will be used to cluster pools of customers and aggregate risk, but also to personalise policies to single individuals and companies. Eventually these sensors will be mandatory in all vehicles and it will be impossible to get insurance cover without them.

The nature of this data will allow insurance companies to vastly reduce risk by warning customers to avoid certain situations, again in real time. This could be purely punitive, but more likely insurance companies will ‘game’ their customers to nudge them in various virtuous directions. Thus insurance companies will move from risk recovery to risk avoidance. This will further blur the distinction between real life and virtual life and insurance companies will cover virtual assets, information and identity as much as they cover physical assets.

Digitalisation will allow new pricing models and payment options too. For example, travel or life insurance will mostly be bought by the day – or even by the second – and the cost would be dynamic, responding instantly to changing context and variables. If it looks as though a tourist is straying into a risky part of town they might receive a text telling them so. Or perhaps their insurance company will notice that they’re away from home and ask for an increased premium or suggest that since they aren’t driving the family car for a while the reduced risk be transferred into cash-back or would result in a discounted travel policy.

If a customer is skiing and the weather looks nasty it would be possible to buy cover on the spot on a ski lift using a phone, but also to link to other skiers on the mountain to assess the risk locally and possibly cover it via the crowd. This might be ‘sold’ to users on the basis that it’s a little like online gambling.

An individual on the lift might also receive a text from Google saying that their latest weather data, together with known data about the individual’s left knee, would suggest that additional medical cover would be sensible if the individual is not wearing augmented reality ski goggles that display hidden hazards. Or maybe the text comes from the travel company, the ski-maker, the ski boot maker or the ski clothing company, all of which are connected to the internet. All companies, regardless of what they make, are now in the information business and offer added-value services direct to their customers.

Similarly, cars, even before they’re autonomous, will collect data on not only real-time driving conditions, but on the behaviour of the driver and other drivers in the vicinity. The telemetry and data analysis used by F1 teams now will eventually be available to everyone.

If a car noticed that a driver was driving erratically it could ask other connected devices for an explanation. The drivers bed might report that the driver had very little sleep the previous night so the car would automatically adjust its safety controls as a result. Insurance costs might be increased until the driver had a good nights sleep.

Of course we shouldn’t forget pets. These will be fitted with collars or embedded sensors that track physical activity and perhaps link to known food purchasing or consumption habits. This will allow for personalisation and the identification of overweight animals and owners.

Homes will be wired and intelligent too, with buildings automatically reporting on their condition and that of any significant object and appliance within. For example, inadequate heating would impact the cost of cover as this may in turn affect frozen pipe risks. Medical insurance would be much the same – constant real-time data reporting on the condition of insured individuals, perhaps with updates based upon daily exercise, food intake, pill consumption and any recent medical interventions. This would be augmented with genetic information about each individual. Any deviation from an agreed policy condition (a sneaky cigarette or too many jam donuts) would void cover, although good behaviour would open up a series of added value benefits and services – the use of certain hard to see NHS medical professionals or access to low-risk robotic surgeons. Expect Apple, Google and Vodafone to all be active in this area.

Most customer contact and pricing will be through mobile devices and this will itself see a high degree of automation with renewals simply requiring customers to press ‘9’ if they would like to renew a policy. Robotic insurance advisors and salespeople will also become commonplace.

How would all this be possible? Beyond the ubiquitous digital connection of individuals and objects, one very big change will be the disappearance of cash. All purchasing will be digital and will therefore record what is being bought, by whom, where and when. Once such modelling becoming precise it will be possible to offer customers cover across all risks with payment that’s constantly changing, much as a domestic utility bill is related to how much of a particular resource is used.

However, the ownership of all this data, much of it reporting on previously unseen, unobservable or private behaviours, will be extremely valuable and this is where the potential scenario breakers come in.

Firstly, whose data is this anyway? If the data is valuable individuals and institutions may demand full or partial payment beyond the payment in kind currently afforded by low-level personalisation.

Secondly, privacy. Will individuals and institutions be happy to let others see or share the data relating to their behaviour, especially when it becomes far more apparent how this data is being collected and how it’s being used and in some cases sold?

Third, security. Perhaps on-going problems relating to data hacking, identity theft or government surveillance will result in a significant move away from smart sensors and big data.