* Posts by John

6 posts • joined 15 Nov 2007

Europe's Tesla will be first with full performance


@ Transmission?!

No, no. You don't understand how steering works in a real car. It does NOT work by the car tyres being turned to lie on the tangent of the circle of turning.

In a turning car, the tyres are always slipping. The slip-angle determines the sideways force of the road onto the car, and that is what makes the car go round the circle.

Putting more drive on one side than the other will TURN THE CAR, but leave it going in a straight line. It's called an oversteer skid. The way that the car nose points through a corner has very little in principle to do with the way its CofG actually moves. Similarly, the restoring-force on the steering-wheel (aka "feel")has no relation to the amount of grip available.

Its pretty much the whole art of car design to make all these effects feel to the driver as if they are the same, so that the car becomes driveable. Colin Chapman's book is an excellent introduction.


Open Wi-Fi network wraps Mumbai man in bomb blast probe



Presumptions? "Obviously you're american, and clearly have never been on either side of the table"

English, currently hiring, have found two decent engineers so far, looking for two more. Require good wireless 3G/4G Layer 1 algorithms experience, and a brain. Know any?


UK gov announces Road Pricing 2.0 - Managed Motorway


How to solve traffic congestion

Solving traffic congestion is very hard. The solution is to constantly reduce speed limits on cars. We know this, because governments tell us so, and governments are the only organisations allowed to actually build this infrastructure.

Let's think about the following problem: "Maximise the throughput of IP packet-traffic through a network of interconnects, with cross-connections through IP routers".

Funded by a government think-tank, my solutions are:

1) Enforce a policy on data-pipes that adds a fixed latency per router hop. That is, impose a maximum speed-limit on the packets.

2) Increase the per-packet-mile charge until people stop sending this traffic - it only harms the environment, and clearly has no economic benefit.

3) Under no circumstances build out any capacity.

4) Take the income from the network, and invest it in building a slower, non-DARPA-networked, non-scaleable solution like snail-mail [OK, railways, get with the programme....]

Funded by companies with an interest in economic success, my solutions are:

1) Build out more capacity.

2) Make smarter routers that route packets in the cheapest direction, using RIP

3) Weighted Fair Queueing algorithms. Proved mathematically to optimise throughput.

What's especially interesting, as a network design problem, is how obviously mad the idea is, that imposing a maximum speed-limit on packets would increase the net throughput. Every network engineer knows that the cause of congestion is OTHER PACKETS. The solution is to get them out of the network ASAP, not delay them.


God makes you stupid, researchers claim


Is this a record?

A bit off topic here, but I'd like to register what I'd like to believe is a world record - the maximum difference between two genuine consecutive IQ test scores. 180 versus 90.

Yes, I was genuinely trying on both of them, and no distractions.

Yes, 90 was the second score.

Yes, I concluded that IQ tests were b**ks and never wasted my time again.

A high IQ score denotes an ability to fit in with the crowd, by thinking the same way as the question-setter. For obvious reasons, that correlates with success in modern society, and hence easy-to-measure achievement metrics. But intelligence? I think not.


Pork and politics energise the biofuel delusion


Re: an inconvenient truth

To Jim -

No, this is not a self-fulfilling prophecy. I am not saying "Don't research renewables". I am saying "Don't build out renewables until they become cost-effective without subsidy."

By the way, the Oozlum bird is not involved. Under current technologies, it is not logically possible for renewables to get >50%. Given that renewables consume more than they produce, something has to produce the net energy consumption. Renewables will reach a maximum of about 30% when the entire world economy is harnessed to burning fuel to support the renewables bandwagon. At that point, everyone will be starving and cold, as the resource usage of everything else in the world drops to zero.

Subsidising renewables in the belief that it is eco-friendly, is actually eco-vandalism. Furthermore, the process of build-out, in the current crisis situation where the amount of oil left is small, is extremely dangerous. I actually believe that a sustained research program can solve the technical problems.

But at the moment, vanity is causing us to waste our resources in order to look as if we are doing something. It is now entirely likely that the cupboard will be bare when it comes to actually building the solution. We will be like Easter Islanders who've finally figured out how to build a boat, but have used all the trees to make paper for discussion documents about whether foreign travel is morally right. Dead. Which is what actually happened to them, pretty much.

The point about coal vs. oil is a good example though. There is nothing in principle to stop us converting coal kWh to a fuel to power cars. Again, in practice, why don't we? Because it costs. Why? Because the practical technology means the overall energy conversion efficiency to a portable power source is low. Cf "what is the cost per kWh when I buy a battery". Hundreds of dollars !

For extra credit, repeat the above exercise for a rechargeable battery, including the production cost of the battery divided by the number of recharges. And then consider why only specialised devices come with rechargeables.

What is wrong with the sunflower oil idea? Nothing, so long as the sums add up. In fact, I absolutely encourage people to make this a business. It would be fantastic if someone could prove this works, by the simple expedient of selling sunflower oil, for fuel, for a profit. That's all it takes to make it a goer. Just don't ask for subsidy.


Merkel-Sarko confab could yield Galileo rescue plan


The Americans are smart

The real problem is that the Americans have played a blinder, politically, for the last twenty-odd years. Firstly, they demand frequency & signal-design co-ordination for interworking. Disagree, and they will ensure that your system won't work. They require the "selective availability" for military purposes - so if you design something that is too accurate, they will refuse to let you deploy it, on grounds of National Security.

Fine. Now they've got you to deploy something which is negligibly better than theirs. For interworking, they can now spend the next 20 years, blowing hot & cold with the technical specs, so you can never QUITE get to the point of deploying.

Then they will always undercut your paid-for service. Galileo COULD deploy with paid-for accuracy, for a cost of billions. As soon as it's up there, the U.S. flip the accuracy-switch, make their system temporarily more accurate, and Galileo uncompetitive. Goes out of business, accuracy-switch is flipped back, go to square 1.