During construction or repair (for example of a bridge), when there are 2 lanes going each way with a median between them, they will sometimes merge down to 1 lane each way, then have a temporary road connecting them. The result is that the two lane separated highway becomes a 1 lane each way non-separated highway. From the point of view of GPS and maps, it looks like you have driven across the median and are now traveling in the wrong direction on the 2-lane highway.

I drew an ascii art diagram, but slashdot says I have too many 'junk' characters.

As a java developer, I have found that python libraries for machine learning, scientific computing, and numerical processing are much better. Numpy / scipy, scikit-learn, and the like are truly amazing, blazingly fast, and the ability to do terse matrix slicing and dicing is so much more efficient. And Jupyter makes it into a live web-based analysis and visualization tool. Almost like magic.

I still do most of my development in java, I understand the ecosystem better, can control the dependencies better (gradle FTW), can make a much larger application, better unit and integration testing, code coverage, can throw it into Tomcat and manage it, can do auth/auth in standard ways, etc. But if you are doing math, python beats its pants off.

The issue is that simply training a backprop system will only get you so far. It won't solve the general problem, and that is Hinton's point. The way that progress goes is that someone has a good idea, it percolates for a while, then people figure out how to use it, the field makes a jump, and then a huge number of grad students write their dissertations on the minor tweaking of the good idea. That's great. The problem is that we're nearing the end of the grad student tweaking of backprop; we will continue to extract every bit of information from the networks, apply them to lots of interesting areas (voice recognition, object recognition, video processing, stock markets, blah blah blah), but they don't push the field forward. We will need a new ideas.

You may be interested in OpenWorm. See: http://www.openworm.org/

They are working on simulating a worm. We can't replace individual neurons, but C. elegans is simple enough that we might be able to simulate it to the degree that we really understand it. An insect is way, way beyond what we can do now, and of course even simple vertebrates are a pipe dream. But, we're making progress. It's an open question of exactly which processes we need to simulate at what level.

As for replacing individual neurons, you'd have to know what they do in situ. Obviously, they receive signals, and they fire off other signals, but the strength of the connections change over time, the intercellular environment changes, the overall level of activity changes, they age, etc., so it's not just replacing a single neuron with a static piece of electronics; it would have to have both short and long term dynamics, and we would have to know what they are. And we don't yet.

That's not my experience at all. Most people go for the lower cost. Airlines have been forced into this for years. People could pay $10 more for a seat, but they go with Spirit because the price on the web site is slightly lower. The experience is horrible, people complain, and yet they still go for the lower cost when making their next purchasing decision. The same is true for consumer goods, where picking (for example) a toaster people pick the lower cost, even if it is only slightly less and with far lower quality.

One of the things that I learned upon talking to IBM developers is that they have to pay standard amounts to use IBM products. I work for a relatively small software company, but anything that we do is available to anybody else in the company. We have a lot of stuff that is narrowly focused for a customer and could not be re-used, but we have libraries, apps, frameworks, and various tools (for devops, for example) that we use freely. IBM doesn't do that; you want to use Watson? You have to pay for it, out of your budget; 'separate profit centers' gone crazy.

Mediocre movie. overly symbolic, annoying ending, but some good action. Danial Tosh is great, but hardly a great movie critic.

It would be interesting to see the distribution of trip times would be for students. That should be very easy to calculate.

Income is at (about) 40%, long term capital gains is %20, right? That's a huge break for people who have enough capital to live on the capital gains. Nobody thinks capital gains are not taxed at all, but that the relative rates favor the very wealthy.

You cannot know the cost when you still don't have any place to put the nuclear waste. It sure isn't going to be Nevada. Which means that you have temporary facilities all over the place. Solve that problem, and nuclear will make a lot more sense. It is a political problem, not a technical one, but it is still a problem.

Year over year, the price of batteries is decreasing, and the batteries themselves are lasting longer with higher density (both weight and size). fnj is just flat out wrong. The yearly change is not that dramatic, but several percent a year, and after a decade and a half you have doubled density and cut the cost in half.

That said, I think that the powerwall is not a good investment for most people.

I expect that we'll have a diversity of models, depending on the state. Having a utility to maintain the grid and have central, reliable, base load power is a good thing. The question is how to pay for it. If all the rich people go off the grid because it makes sense long term but poor people cannot because they can't afford the investment, then grid power cost will skyrocket. That will be bad for large numbers of people; rich people will be fine, they almost always are, but everybody else will be screwed.

They have been working on getting to 100%, their goal for 2045. I think that it will happen before that. Last year it was 26%. See this article. In agreement with much of the sentiment here, the incentives / disincentives of the utility commission have significant effect on the adoption rate of renewables. Things like, can you sell the power that you generate back to the grid and at what cost relative to getting centrally generated power, and when you can do that, and how costs are divided up between infrastructure and operating.

Yes, but it is also in society's best interest to have lower energy costs, distributed power generation, and lower emissions. It is in the utility's best interest to have large highly profitable utility regardless of society's best interest. The problem appears to be that in some cases the utility is getting what it wants at the expense of pretty much everybody else.

I took this, and it was good as far as it goes. Ng is an excellent teacher / explainer, so it's a good start to see if the field is for you.

However, it's really just a taste of what is in the field, and takes (obviously) a machine learning approach as opposed to an AI approach, and even then it focuses more on regression than I would have liked. It doesn't get into the math very much, and honestly, the math is vital to really understand what is going on.

Overall, I'd agree with you. But it will be really interesting to see different market segments. think about the city car, ubiquitous mid-sized family sedan or yuppie SUV. Those segments could easily become completely dominated by EVs and quickly become mostly self-driving. That will make it incredibly difficult to buy one of those cars that is _not_ EV.

No way that rural users, people that actually use their pickups for work or tow, people that drive significant distances, or other groups that have non-generic uses switch in this timeframe. I personally need to have a large, long-range vehicle often enough that I can't use the EV, but my family has two cars, so one will be EV and one will not. The disparity between city and non-city users will come to a head, because of the interactions between them will be initially messy. So, you could have a rule in NYC, DC, Boston, San Francisco that prohibits human driven vehicles, or non-specially tagged vehicles over a certain weight.