Mapping Ignorance

From a followup post "Progress checking Zarankiewicz’s conjecture on the brick factory problem" [http://mappingignorance.org/2013/10/01/progress-checking-zarankiewiczs… "This means that N_0(5) \approx 6\cdot 10^{1228} , a really big number. The fastest computers today perform in the order of petaFLOPS (10^{15} operations per second), so even if they could check each case in a single operation, it would take in the order of 10^{1204} years to check if the conjecture holds for K_{5,n} with all n\leq N_0 ."

There's a pretty cool article about trying to understand electronics and fix a radio the way we try to understand biology. https://mappingignorance.org/2014/07/02/biologists-cant-understand-bio… So now instead of improving the tools of biology so we can actually understand it deeply - we increase complexity of IT so we have to rely on muddy, side-effecty tools of biology to try to infer some of the properties of the systems we made. That's depressing.

This reminds me, I was looking at the periodic table the other day and I noticed that gold actually has a pretty high electronegativity: almost as high as carbon! https://mappingignorance.org/2017/02/09/electronegative-gold/ Makes me wonder if there are a ton of unexplored properties and usecases for rare metals that we just haven't discovered yet.

I'm surprised there's no mention of the van der Walls attraction used by geckos feet. http://mappingignorance.org/2013/07/04/quantum-mechanics-in-biological…

Many of these approaches for cleaning up nuclear waste from legacy nuclear sites may also be useful for fusion reactors. I talked with one of the people working on building robots for Sellafield and one of their other projects was building a robot to refurbish parts for ITER. Nuclear fusion reactors like ITER will need to go down periodically for maintenance. The amount of time the reactor spends offline is lost money, so maintenance needs to be done as fast as possible. So one of the robots they were working on would go into a pipe, cut it into pieces, put in a new pipe and reweld it within 30 seconds. Another interesting challenge for robots in ITER is that some may be exposed to such high levels of radiation that the: "designers can forget all the advances in electronics we have witnessed since the 1960s: only a few discrete transistors can be employed, with integrated circuits being forbidden for their sensitivity to radiation."[0] [0]https://mappingignorance.org/2014/01/29/iter-engineering-challenges-of…

Because people are bloody minded animals. This is a survey: http://mappingignorance.org/2013/03/20/experiments-in-fairness/ The seminal experiment is Kahneman around 1986. I fail at finding a good treatise on it. But here's this: https://en.wikipedia.org/wiki/Prospect_theory And as applied directly to econ: http://www.econ.uzh.ch/dam/jcr:ffffffff-9758-127f-0000-00005e2c9579/Do… The banks are simply our agents in this. Roughly, other "innovations" have made debt more dangerous over time. My paltry understanding is that one of the uses of inflation is to erode the value of debt. We don't have meaningful inflation any more. And yes, I relaize how bizarre that sounds. But: http://www.interfluidity.com/v2/3212.html Something terrible has happened to the tradeoff between the balance sheet and cash flow. A really crummy analogy is position v. velocity on quantum realms; as competitiveness has escalated, we get to where the trade between them is more severe. Also IMO, much as wages have stagnated, the value of capital goods has also.

The winner got 6.7% in top-5 classification error. 2013 was 11%, 2012 was 15%. This may be better than exponential, since there is an upper limit of 0% error, and every additional percentage point is harder than the last. Exponential improvement in this area, and it's a pretty significant field that will affect everyone. The results will likely help improve neural networks in other areas as well, as NNs are extremely general purpose. Here is an example from a system from 2012 in the same competition to give an example of what this is: http://mappingignorance.org/fx/media/2013/04/Deep-learning-5.png A bunch of images were scraped from Flickr and then given a single label by a human. The computer gets 5 guesses to guess what label the human used. As you can see, even when the computer fails, it's guesses are usually semantically very good. This specific system got an error rate of 17%. You can explore the images of ImageNet here: http://image-net.org/explore. Here are some image recognition systems online you can play with: * Toronto Deep Learning Demo: http://deeplearning.cs.toronto.edu/ * clarifai: http://www.clarifai.com/ * AlchemyVision: http://www.alchemyapi.com/products/demo/alchemyvision/ Note these aren't related to the winner of this competition and probably are no where near as good. Just examples of this kind of system you can play with.