This looks like a cargo cult to me.

Not "cargo-cult science" as denounced by Feynman where some nutjobs tried to make telepathy or astrology look scientific. This is real, serious work, with a sound methodological grounding.

I'm afraid that they are actually building a "cargo-cult" brain, because they actually are at the wrong spot of the aforementioned graph too. Think of AI in the 70's. AFAIK, we know far too little and the field is still too fragmented to build a reasonable model of the meso- and macroscopic structure-function of the brain (I'm actually unable to tell where one start and the other ends, assuming the distinction between both concepts isn't a false dichotomy).

You cannot emulate a brain out of the box. You have to simluate the whole embryologic developmental process (which is half directed (mostly at the population scale), half random (at the microscale), but it's actually more complicated) before you can start to educate it, assuming you built something functional. Once again, we know, at the same time, a lot and close to nothing about these development.

Brain simulation could be, oh the irony, a good target for genetic programming/engineering :-).

I think most engineers recognize that the brain has a lot of unknowns.

When you say "educate" what exactly do you mean? We can train neural networks now to, say, recognize handwriting or other behavior. I think we are just after a system that can learn anything in general, say take a chess program and then give it checkers. Or Magic: The Gathering.

Also, the field seriously needs people like you. It seems like there are at lot of engineers who are into the brain and want to build stronger AI but not many biologists who want to do the same.

To understand the basic concepts behind these properties is essential for two reasons: The life-science point of view and the information-technology point of view.

    - The first point of view has potential medical applications to cure brain and mind related diseases or even the longer-term goals to work towards neural prosthetic devices and artificial sensory organs.
    - The second point of view could lead to new computing devices radically different from contemporary IT technology. Such devices could provide support for complex decision making processes like the one we are currently used to obtain only from human beings.

To improve the performance of such information processing systems, several approaches can be followed depending of the prior knowledge available. Indeed, depending on additional labels (class or continuous value) which can be used (or available) on none of the patterns, on a subset of the patterns or on all of them, unsupervised or supervised learning can be sequentially performed. When there is no prior knowledge on the problem to be solved, knowledge extraction may use an unsupervised neural network as a front-end for forecasting applications or extracting rules. Because of its synthesis capabilities, an unsupervised neural network can be used both for limiting the computation complexity and for extracting the most significant knowledge. Moreover, knowledge extraction is facilitated as soon as multi-viewpoint unsupervised neural network model is used. This kind of methods also allows using in a second step additional information when it is available for optimizing a forecasting problem. However, for a forecasting problem where all patterns are labelled, classical networks using supervised learning can be successfully improved by finding the minimal architecture using pruning algorithms. The pruning methods consist in removing, during learning, the connections or neurons, or both, that have the least influence on the system's performance. Reducing the complexity of the networks prevents overtraining and allows easier implementation and knowledge extraction (variable selection, rule extraction). In any case, combining several models into a committee helps to improve the quality of the knowledge extracted or the forecasting and the proposed methods must be efficient for typical real-world in our domain, dealing with large amount of noisy and temporal data. Both topics are recently developed in the project.

I actually think that, in OpenCog, for example, reasoning or other systems could take advantage of a SNN to provide rule selection or knowledge extraction, etc.

So the point is that even though they might be completely ignoring most higher-level necessary aspects of human cognition, a more comprehensive system can still benefit from incorporating SNNs.

I just thought that they were going to have some serious problems with yield at that kind of scale, but then it struck me that that's kind of the beauty of it: it doesn't really matter if a couple of neurons don't work - if it's anything at all like a brain, it'll be remarkably failure-tolerant.

(based on my admittedly limited knowledge of brains, simulated brains and semiconductor engineering)

I really recommend this book - The Man Who Mistook His Wife For A Hat by Oliver Sacks ( http://www.amazon.com/Man-Who-Mistook-His-Wife/dp/0684853949 ) to change opinions of brains and people.