The problem I was talking about is where a schematic with 50 parts on it is a giant rats nest of stuff all on a single page that someone then tries to fit onto a single page/monitor, making extensive panning and zooming necessary to understand what's going on. Following the rules I outlined prevents that problem by making it so you can simply print out the images on a normal sized piece of paper and find everything completely legible (and similarly with most monitors).

The problem that you are talking about is one that can be solved with any moderately competent image viewer. Even firefox's automatic image scaling is perfectly capable of handling it, within reason. If that's all that this javascript tool accomplishes, then it is incredibly sad how slow it is. On my computer it hangs for a good 20 seconds when trying to zoom in or out, and god forbid I try to drag the image. Surely a better tool for dynamic image viewing already exists -- the one on amazon.com seems to work fine, after all, for the same task.

And yes, the PNG image is 2952x2202 pixels, but it's also 70kB... this isn't exactly an unwieldy image. And if it were a vector graphic or pdf, it'd probably be even smaller. I only chose not to do that because there is less support for it in browsers... like how I can't open PDFs inline in firefox very pleasantly.

If it had labels for logical blocks, I think that the schematic you linked to would be perfectly legible. A bit overkill, but at least it's not messy. What if there were two LEDs? Three? Four? There is definitely a limit where no matter how obvious the connection is, it increases legibility to use named nets.

Yeah, I definitely agree with this. This seems to be a solution to the problem "how can I make illegible and amateurish schematic drawings more readable without learning anything?"

Use a frame that limits the total schematic size to a standard paper size. Use named nets and labels on nets instead of actually connecting wires between parts (except for trivial connections like capacitors). Put lines in your schematic that separate logical blocks of your schematic. Label logical blocks with a title (AC Rectifier, Boost Converter, Control System, ADCs, Filters, etc). This makes it trivial for someone to look at your schematic and rapidly identify errors. It makes it simpler for *you* to rapidly identify errors!

Just follow these four simple rules and your schematics (pretty much regardless of software used to make them) will suddenly appear to be fairly professional (if not perfect). For examples, take a look at this. I'm not an EE by any means, but the more you separate functionality into logical blocks and limit your size with frames, the closer it looks to "professional".

http://saikoled.com/lightshield/
http://saikoled.com/lightbrick/
http://web.mit.edu/neltnerb/www/artwork/design.html

(for the last link, some fairly complex schematics are shown in the "New Schematics and Diagrams" section. The ones near the top are duplicates of what I published on the other website.)

You can either have academic labs researching things which are commercially interesting, and then give the professors working on it the perk of having the opportunity to commercialize it first (or at least royalties), or you can have academic labs researching things which the professor is academically interested in, and hope that it is commercially interesting. It is difficult to get both.

Either you get people complaining that publicly funded research isn't free to the public to use, or you get people complaining that stuff invented in academia has no practical application. And since there aren't any industrial research labs left, that means either no commercially interesting research, or encumbered research.

Not to mention that it would be *damn* hard to get professors to work for peanuts (seriously, I've seen what these people make for their qualifications) while training basically all high-skill future scientists, and under a contract where all work they do they can't even commercialize because some big company will snap it up underneath them.

No, I'm afraid that I have to disagree with your position. Yes, I have a bias because I am working very hard to commercialize technology that my lab invented, but I also think that is is more than fair to give the actual inventors first dibs on trying to commercialize something. I would have left academia in a hurry and just did all my work as a trade secret pretty quickly otherwise.

National labs of course are a totally different story. Usually their inventions are licensed under reasonable terms in only non-exclusive licenses. But those inventors are *working* for the government as opposed to just having a small fraction of their costs paid for by a government grant.

I dunno, they seem to manage fine with iOS and android. We're talking about netbooks, so the different form factor makes people intuitively not expect it to be *exactly* the same as what they've always used. And Unity is closer to looking like android/iOS than windows, which makes even more sense if the device is looking more like a phone than a desktop... although I definitely agree that not including Unity is an obvious choice. That stuff is just a disaster at present.

If memory serves, the giant flywheel that MIT uses to spark their fusion test reactor is rigged with explosive charges to blow it to pieces if it ever came loose. I believe the calculations show that without detonating it, it would likely continue *through* several buildings before landing in the Charles River... could have been an urban legend though.

I have heard from electric companies that they have absolutely no problem with this. People doing "energy arbitrage" are essentially helping the power companies even out the grid, which means said power company doesn't have to turn on the expensive natural gas generators as often (or purchase less natural gas power when they are on). You're just getting into the business of providing space and equipment to do grid leveling informally.

The efficiency of an electric motor can be in excess of 90%. Energy is transferred to a flywheel via electric motor, and extracted (mostly likely) through the same electric motor, so your maximum theoretical efficiency is going to be your motor efficiency squared. If they tried hard, probably something like (92%)^2 or something like 85% total storage efficiency.

This is of course assuming that mechanical losses are zero, but given the design they are very likely to be close to perfect. There will also of course be some energy lost indirectly in levitation/cooling/ohmic stuff outside of the flywheel.

I think the thing about this article that bugs me the most is they say that the flywheels can store 20MW. What on earth kind of way to measure an energy storage device is that? 20MW for 0.5 seconds? 20MW for three days? Embarrassing.

Hopefully someone who is on an academic IP address can explain why this is any different than the standard wet-embossing techniques that we've been using to do this kind of thing for the last decade and a half... those SEM images sure look awfully similar to the stuff I was doing back in 2001. Maybe they're just saying that they crush the porous substrate whereas with standard techniques we suck up solvents in substrate inks? That would be kind of neat, although it seems like it'd be limited in utility so I imagine it's more clever than that... do they crush some porous substrate and then manage to lift off the pattern or otherwise remove the crushed portion? Do they have a technique to deposit different substrates on the same device? Otherwise, it's not really going to be useful for most electronics right? I mean, making a pattern of n-type silicon isn't going to make a useful device unless you can deposit p-type and conductor on the same device and manage high degrees of alignment... maybe they mean that this can be used as memory? DIffraction gratings by themselves are rather boring...

A shame that the article doesn't say what the substrates actually are. I do like the photos of the little tubes, although without a scale bar I'm not sure what I'm looking at.

Wow, I'm glad that my TI-89 from literally 1998 still works perfectly... I use that thing *all* the time at work. I would be furious if I could no longer use the eigenvalue and eigenvector solving software. Did they cripple it in any other ways since then? As is with the stock OS I can solve fairly complex integrals without even simplifying them...

Indeed. I found the Unity interface to be "just another option", and one that makes a great deal of sense for computers with minimal screen real estate. However, I confess to switching back to normal gnome even on my netbook because of stability issues and graphical glitches. Very annoying ones, including stupid stuff like lost windows which I can no longer access despite them being visible and such. I would personally have thought it to be much more sane to make it a "special option" for one more release so that it gets troubleshot and tested by bleeding edge users before bothering to pretend it's actually ready for the mainstream.

Also, asking the bartender how to pick up the ladies is usually an entertaining conversation starter.