Some 3d printing services can print ceramics ;)

Services like that exist online, and they're excellent, albeit rather slow. I personally use iMaterialize because they have such a wide range of material options (everything from rubber to titanium) and finishes (for example, 4 different options for silver), but there's lots of others out there, and some are cheaper.

If you've ever played around with 3d modelling, I definitely recommend giving 3d printing a try, even if just a little test piece. :) Note that plastics are a lot cheaper than metals, although metals look the coolest.

And those nerdy kids will grow up playing around with and learning 3d modeling software to be able to make their toys.

This is a good thing.

What sort of 3d prints are you looking at?

Perhaps my expectations of 3d printers are too high because I buy from professional 3d printing services rather than using a low-end home 3d printer. They use high end products and sometimes do post-printing finishing work. But the quality of the stuff you can get is truly excellent, and out of a very wide range of materials.

Isn't that now the limiting factor?

So we have 3d printers in stores. Now we need all of the home devices that could potentially need spare parts printed to be available online, preferably in a unified database. You need manufacturer buy-in. Maybe some sort of certification mark that manufacturers can stick on their devices to show that printable replacement part models are freely available. I could use a new cheese compartment door in my fridge right now, for example. And I live in Iceland where shipping times are long and shipping costs / import duties high, so it'd make time and economic sense to print, too. But while having a 3d printer would be great, if the model isn't available, how does that help me?

Of course some companies, like iRobot, rely on profiting off of selling their spare parts.

It does seem rather weird to treat it as an intractable problem. Are we really talking about something that's AI-Complete here, like natural language understanding? Something not succeptible to a combination of chained rules, physics calculations, and statistical analysis? I seriously doubt it. So different machines can act differently due to wear, etc? Gee, people have never written programs to deal with that before, heavens no. So some things may require a decision from the operator, like whether to restart a defective piece or try to salvage it? Gee, I've never heard of a program asking the user a question during operation before! A piece of "printing" hardware experiencing a jam of some kind and needing manual intervention? Gee, nobody has ever experienced that one before!

I'm not saying that CNC machines and 3d printers are equivalent and that you can just swap a CNC machine in to the sort of role 3d printers are intended for. Of course the task of gouging out steel with power tools is a more intensive one than writing out plastic in layers with a slightly more advanced version of a hot glue gun. But we're not talking about creating superintelligent cyborgs here, we're talking about analyzing physical processes, including their various failure modes, and when a decision or action is required, presenting the user with the information needed to do that.

I'm sorry, I'm having trouble hearing you, your horse is too high.

Oh, and in #2, sound insulation would also be very important, both for the compressor (if compressed air is used, rather than bottled oxygen) and for the jet itself (which is basically like a tiny rocket engine). And I guess the filter isn't just about removing any incomplete combustion products from the exhaust, but also any dust or the like.

Even if it ultimately isn't suited for, say, a quiet home office, 3d printing isn't really an home office task, we're more talking about a "garage workshop" sort of thing. I'm just curious whether anyone has pursued such an approach, because at a glance it sure looks to have potential for making a very broadly capable product. I mean, such a system should even be capable of printing electronics, including resistors, capacitors, etc, maybe even some types of batteries (not anything requiring extreme precision, like a CPU, and li-ion batteries would be right out due to the thin, sensitive and rather complex membrane needed, there's no way you could just deposit that, but still..).

There's two types of processes that I'm surprised I've not seen more focus on.

1) Printing of, and then filling of molds, which can then be melted down and reused. That's how the higher-end 3d printed parts that you can buy online made, including almost all 3d-printed metal parts you get from online 3d printing services (the extra steps for metal being to coat the mold in a ceramic shell and melt away the mold). The only commercial 3d-printed metal that I'm aware of that doesn't work in this manner is iMaterialize's titanium, which uses laser sintering - and it has an out-of-this-world price tag.

It seems to me that if you used a mold, while in several ways it complicates the process (extra steps, preventing adherence to the molded object, etc), in others, it simplifies it. Your print heads don't need to handle a variety of materials or produce a pretty or durable product. They still need to be able to produce fine surface details but the ability to print thin structures loses importance. Once you've got a mold, you open up the floodgates to the sort of products you can fill it with, anything that will harden either through cooling or via chemical reaction, anything from thermoset plastics to candy.

(note I'm not envisioning a little hobby home printer that fills molds with molten metal in your office, mind you... although I could envision a more garage-scale or small industrial scale version that could handle such a task)

2) I've never even heard of a 3d printer being based on thermal spraying. With thermal spraying, you can choose the balance of precision vs. flow rate via nozzle size. Your materials are virtually unlimited, pretty much anything you can turn into a powder. It could conceivably even let you work with metals in a home environment, if the rate was kept low enough that heat buildup wouldn't be a problem (and you'd want an air filter on the exhaust, even though it should be pretty clean). You can choose the temperature and velocity you're spraying at by varying the pressures of compressed air and combustible fuel fed into the chamber with the powder, so you can work both with heat-sensitive and heat-requisite materials, as well as materials that can't stand high velocity impacts and materials that require them. Such a system could likewise do more than just print - it could add and then sectively remove substrates, it could engrave, it could change surface textures by sandblasting/polishing with various materials, it could paint or apply high-performance coatings - pretty much anything you can envision from a device whose fundamental workings are "shoot grains of material of your choosing at a velocity of your choosing (1-1000+m/s) and temperature of your choosing (cold to thousands of degrees)".

In both cases #1 and #2, I'm genuinely curious as to why there's not been more work done with them. Or perhaps there has been work done with them that I'm unaware of? I'm asking as someone who makes and buys 3d printed items online but has never printed one herself.

The genes they identified were all proteins.

I'm not that much of an expert on microarrays, but I'm pretty sure most or all of the arrays they used predate the Encode project's results that made people re-evaluate the question of how much of the genome is really important. Here is a list of the arrays they used:

Illumina: HumanHap550, 318K, 350K, 610K, 660W Quad, HumanOmniExpressExome-8 v1.0, Human610 Quadv1, 370, 317, HumanOmniExpress-12v1 A

Affymetrix: GeneChip 6.0, 250K

This study was the keystone project of a consortium founded in early 2011. I think, given the size, it simply took this long to get the results. That, too, was a time before Encode publications had really started impacting the world. Whatever RNA genes they would have had at the time would be pathetic and paltry by comparison to what we consider worth studying now.

I've read the The Bell Curve, and I think it was a fair analysis for it's time, but--unfortunately for Murray--it was written right before the genetics revolution made all his speculation about race seem naive. The assumption at the time was that people of the same race were genetically similar; therefore, you could lump people of the same race together and make assumptions about their genes influencing their intelligence.

Then the Human Genome Project came along, followed by cheap genetic testing, and scientists like Craig Venter found that the genetic similarities between people of the same race are nothing compared to the genetic variations between any two humans.

In other words, The Bell Curve's conclusions were based entirely on phenotypic analysis, which was fair at the time, but the advent of genotypic analysis has rendered the book pretty much irrelevant.

I would add (6) Many states have regulations making it impossible to do what Musk is doing. I live in Republican-Controlled Virginia, where I can't buy solar panels from Musk's SolaryCity, which has a location 20 minutes away from me in Washington DC and more locations in Maryland, because my state has pretty much given Dominion Power a monopoly on supplying electricity here, giving them exclusive rights to net-metering--which they have made cost-prohibitive to implement, and the company has actually successfully sued organizations that install solar panels.

Helium exists in the atmosphere not because of the helium reserve, but because the planet constantly outgasses it. It's a product of the radioactive decay chains within the planet.

And if it costs $7 a liter, you better believe people will consume it a *lot* slower. Mainly recapture, but also less frivolous usage.

We know that the most important distinctions between humans and other animals are in RNA genes, that most of the genome is transcribed as RNA genes and that the brain modifies itself using them and that malfunctions in them cause disease. This study ignored RNA genes entirely, AFAICT. Its mindset is about ten years out of date and simply reaffirms what everyone already assumed: proteins aren't everything. Intelligence probably still has a significant genetic component, this study just looks in the wrong place. (Psst: SNP studies are snake oil in almost all unsolved diseases.)