China Tech

You will find a lot of it has changed now. I was looking at a factory which made polysilicon for solar cells. Ten years ago it was all workers in an industrial factory which looked a bit like an oil refinery, as in pipes everywhere. Now no one works in it at all apart from one inspector who does his rounds to see nothing has blown up. The rest of the staff sit behind monitors. They far prefer the work they do now because it is far more comfortable in nice air-conditioned office.

What happened in China was the government at some point got into the idea of being technocrats, so they initiated a huge investment programme to buy in as much new technology as possible, which has been the basis of their government investment. To start with they saw huge grown due to the low hanging fruit, as in processes that were ancient and could most benefit by replacement. This game is the law of diminishing returns, hence why GDP is only growing at about 4% now, where a while back it was 10+%. I believe the rural areas are not so advanced. It all depends on where you are.

The government's technology programme covers all of industry. It's to do with the way the macroeconomic system works over there. In any economy money gets created to keep the money supply at a level which reflects production. In China new money is injected via technology spending. The US did a similar thing, yet the money was directed to military technology, so that is why the military sector is so stinking rich. In the UK it is channelled through the City of London so it goes into the hands of the City slickers.

By the way, I almost got a job at one time to create some technology for a paper factory. The job was to measure the thickness and consistency of the paper as it flew off the rollers at a fast speed. My idea was to use NMR to figure out what was in the paper. They decided not to go for it in the end, but the other day I was looking at a chap who tried a similar thing himself and it seems that it may well have worked given good enough screening from interference.  It involved picking up a 148MHz signal. There is a similar technique called quadripole resonance. It works a bit like a mass spectrometer in giving data on chemical composition.

Are you aware the ideal material to make a speaker cone is a bamboo composite? Some speaker manufacturer in Taiwan patented it and of you look on youtube, their speakers are recommend everywhere. Not only are their drivers cheap, like around the 70 quid mark, but they are ultra compact and have an amazingly flat response. Using such speakers is said to perform to a standard ten times their price.

This bit of tech here in this link is a photonics chip for doing AI.

They call it 'All-Analogue Chip Combining Electronics and Light (ACCEL)' . The reason you would want computations to be done using analogue techniques is because AI is an analogue system. The basic computations in AI involve calculating weights by multiplying out a string of other weights, because essentially what you are computing is probabilities where p(a&b) = p(a)p(b) and of course the answer is analogue. What you want then is a physical effect that gives you a multiply and a physical effect that can store a level and then read that level. You also want to do that using photons. The reason you would want to use photons is because they go the speed of light. Lets look at the performance of a working AI chip using photons:

Do you recall the naysayers all nonchalantly telling us AI is super ungreen due to the vast energy it uses. Well it could be said the problem was we were doing it the wrong way all along. Digital computers are a poor match for AI computation. The answer is to fix the problem at the chip level. There is one catch though. To date they are unable to program this neural net. What they are doing is hardcoding the program onto the chip, so the chip can only run one program and to update it would mean changing the chip. That should not matter too much if say you want to use it in an AI camera. It only ever needs to run one program for its entire life. And besides this is early days. Perhaps they will go on to make some sort of FPGA type of architecture.

Here is a good one for today: water batteries for grid storage.

Energy density of water- 1200Wh (much higher than lithium) + very high current density 120mA/sq cm

Reversible multielectron transfer I−/IO3− cathode enabled by a hetero-halogen electrolyte for high-energy-density aqueous batteries

Micro bots