Another vote here for Linode - I've been running web services on a node for almost 2 years now and all sorts of background stuff that uses the bandwidth that isn't needed by httpd. Currently have a murmur server running on it for game voip as well. I don't use PV-GRUB because the Archlinux-paravirt kernel that they officially support is exactly what I want.

Over the course of those 2 years I have only had one downtime, and that was because the machine I was on was scheduled for maintenance - they sent me 2 weeks notice and I was able to shut down and transfer the VPS over to the new physical machine when it was good for me any time in that window.

I will say that I've told friends about it and they scoff at the price, but I've yet to find a better price from a reliable host that gives you exactly what they specify with no added restrictions or hidden things. There are links in various places as you're maintaining the node from their web interface like "backup" and "add more resources" that bring you to the extras page where you can pay for them, but I'd expect nothing less and that's the grand total of "ads" that you see anywhere.

As a MechE who is currently working in Aerospace doing design, let me tell you that those of us who know how to properly CAD stuff up and indicate the important dimensions on a drawing hate the other guys as much as you do. I can't say how many times I've opened up a part file, gone to the sketch, and found that none of the lines were fully constrained, or the constraints were arbitrary and were only tangentially related to the driving dimensions. I used to go back to the original author and ask what was going on in their head, but found it to be easier to just silently redo constraints on the features that needed it, hopefully without moving any lines. The place I'm in now is full of people who have been using NX since it was new, and yet the "guru"s in house all say that sketches are bad and want us to use solid features instead - completely ignoring that it's so much harder to change parameters when a design needs to change, all because sketches used to suck (or so I hear) and they can't be arsed to learn how to use constraints correctly now.

The fun part comes when you have to mix units - two weeks ago I had to draft up a simple adapter plate that had 4 force transducers on it, which all happened to have metric bolt patterns. Trying to indicate that the distance to the center of each group of holes was the driving dimension is fun when you don't have a feature at the actual center, but at least you can dual dimension with the nice even number in mm under the ugly inch one. (disclaimer: I hate the english system. I have to use it because that's the policy when you're .gov).
Then there's the "here's a vaguely circular bolt pattern with 28 thru holes, and the only important thing is that they're symmetric about a center point, have a minimum radius, and line up so the bolts go into a 1" grid on some table somewhere", but that ends up needing 20 dimensions and all sorts of center lines. These are times when GD&T is just annoying and it would be a whole lot easier for me to put a note on there with the intention (though that's probably because I don't know enough yet to do it cleanly and correctly).

I like it when the machinists or someone else checking the drawing tells me what I did wrong so I can fix it and not have them need to yell at me again - I just wish more people I worked with had that attitude.

Mixed-mode communication completely breaks this:
When I get a really long email from a friend or family member asking a question that would take longer to write out than to explain over the phone, I'll wait until I'm free and then give them a call. I guess that means from an email perspective that I hate them and never reply.
Or what about various organization mailing lists where you reply to the sender with a new email instead of sending something to the whole list?

Of course this is all irrelevant because this study isn't really intended for emails, despite how they report it - it's for social networking sites with embedded messaging systems to be able to mine more data about you, so they can show you ads that your "closer" friends have clicked on in addition to matching with your profile items, so they can charge more for ads.

So it's like a library, but you pay for it?

I know, I know, libraries are paid for with taxes and their various fundraisers, but that's because they have physical buildings to maintain in addition to the books. Unless this service is pennies a week, it's gonna be a ripoff. Their distribution costs are negligible (text files of books are what, a few kB?), inventory control is practically free because there's no such thing as a lost book, and they can send out as many copies as necessary instead of waiting for one patron to return one after they're done. The only cost to them is the book rights in the first place.

I agree completely.

As a G+ user I'd be perfectly content with games being not even included in the first place, because that's not at all what I want to use it for. If I were a developer already invested in trying to make money off G+ games, then I'd have made a bad business decision. A huge share of users are there because it's not covered with games and bloat, it's only a simple way of connecting with people and sharing links and posts. I hope Google also realizes that and sticks to just making a little money off ads and whatever they sell right now, instead of trying to make even more with games and losing their user base.

Since iDevices (at least the iPods I've used) have the ability to charge from the "dumb" wall bricks with a USB port, why not bring a custom cable with you to unsecured locations that only connects to VCC and Gnd on the USB port, and has the appropriate resistor between the data lines to indicate a valid charging station? Voila, no risk of data going either way and you still get a charge from an unknown location.

Well, they could just move to the 2.5mm standard that's been around just as long, but that would make too much sense.

My dad's 30-year-old stereo is what I'm listening to right now, ever since he replaced it and I "inherited" it, complete with nice speakers with real wooden cases from the early 80s. So no, his 30-year-old stereo sounds exactly like mine!

I also bet that if I bought a new one today that wasn't all the way in professional grade it wouldn't be still working in 30 years, but this old silver-faced Kenwood receiver/head unit (KR-710) that's older that I am has my mixer board plugged into one of the tape inputs and still works like a champ. Well, except the radio tuner carriage that doesn't like to move all the way to the right, but that doesn't matter. And sometimes I have to jiggle the volume knob so that the right channel comes back on. And the EQ knobs don't really do much except make fuzz when you turn them. But just using it as an amp while playing FLAC or CDs on the computer and controlling the volume on my mixer board, then it's rock solid and sounds great.

The older units may not have remotes, fancy digital displays, or let you play your surround sound DVDs through them with optical TOS-Link cables, but for an amp that sits in the corner that you pipe your audio to with RCA jacks, there's not much to complain about because they do the important part right.

I know this article is about the desktop APUs, but as I've been running the C-50 Ontario on my netbook (Acer AO522-BZ897) for a few months now, I think I can share some real-world experience.

Overall: It's a dual-core netbook, and still gets 6 hours battery life if I'm writing code with the brightness down, a little less if I'm listening to music. It may be slower on the individual cores than a competitive Atom, but if your program is threaded it's great. I'm very happy with the performance. It replaced a Powerbook G4 (I know, different class altogether, but still), and in terms of % CPU used for common tasks it's far and away better. No more mp3/m4a decoder taking 10+% CPU for decent bitrate songs.
Real-world Performance: I can say that any downside I've seen is entirely due to bad software - I hear that in windows I could watch 720p on it, but right now with x64 linux and the beta multi-threaded flash player in the latest firefox or chromium I can't watch youtube videos at more than 480 before it starts to drop frames. Not a big deal for me though. Once the video drivers caught up with the chipset I can say that compositing and desktop effects work flawlessly, no lag whatsoever. I don't play games on it, being a netbook, except for the occasional flash thing (which sometimes lag, but again that's the flash plugin).
Hooking it up to a 1080p TV over HDMI and running at native resolution, playing standard definition (624x352) XviD files zoomed to fullscreen works flawlessly in VLC. 720p x264 almost works - it saturates a single core and drops a frame here and there, and will really hang if you try to bring up a semi-transparent control bar over the video, but again if the codec were multithreaded it would be perfectly fine. I suspect that VLC on linux isn't taking full advantage of the GPU here either, considering I'm running the open source radeon drivers and not the official binary. For those of you running with officially supported closed-source software, i.e. official drivers or windows, I suspect it might even play 1080p without a problem.
Let's remember that this APU is in the same power class as an Atom, and it's a netbook - impressive performance in my mind.

As my only direct comparison points are the G4 powerbook that it replaced and my Phenom X4 9950 desktop, it's (gasp) right in the middle, but comparing a netbook to a desktop built for CAD and gaming is stupid, and so is comparing it to something 5 years older.

My only gripe is that you can't set how much RAM the GPU side takes, so no matter what size stick I've got in there the system sees the total minus 256M. The upside is that you only need a total of one stick of RAM, but the downside is that when it comes with a 1GB stick, suddenly you're trying to run a windows 7 system (out of the box) with ~750MB, and that's asking for trouble. As I swapped out the HD and RAM before ever turning it on the first time and installed linux fresh, I can't say I've seen the slowdown, but it could be there.

I think they're talking about unintentional energy usage. I know the heater in my apartment uses a lot of power, and that my computer running 24/7 makes up a good chunk of my power bill as well. I know that my refrigerator uses power, I hear it running - same with the dehumidifier. This is more like leaving your phone charger plugged in while it's not in use and having it eat 2W constantly, or people who leave their powered computer speakers on all the time and don't realize it costs them a few $ a month - "But there's no sound coming out, so it isn't using power, right?" People would expect that when they're not watching TV, that all the components in the TV chain are also powered down.

I just hate with the set top box that if you are power conscious and have everything on a power strip that you flick off when you're not watching TV, that when you turn it back on the next day it takes 10 minutes to boot up and get the satellite signal again, and half the time it loses clock and personal settings (and the damn 500ms button lag on the interface and a line of advertising at the bottom of the program guide, but those are unrelated). I thought that with decades of desktop computers being turned off that cable companies would learn what a CMOS battery is...

Newer reactor designs (so-called Generation IV designs, specifically the fast reactors in there) are being designed with the issue of waste in mind, and already offer significant improvements in both reduction of waste for the amount of power generated and reduction of the half-life of the spent fuel. One of the promising concepts is the ability to make a single initial fuel load that reacts over a long period and cannot be removed from the reactor, so all the elements that are useful for weapons are only a middle step in the reaction chain and cannot be extracted for military purposes even if you wanted to. This lets the fuel go much farther down the chain, producing less power as the reactor ages but ending up with waste with order[s] of magnitude shorter half-lives.

This is not saying that nuclear waste can ever be "safe", and it's anyone's guess as to current countries even lasting the centuries that the waste is still dangerous, but at least these can burn what is currently considered waste and reduce it even farther. Also keep in mind that nuclear power is less than 70 years old, plants last 30-odd years, and take 10 years to be built in the first place (nowadays). Likening the the release cycle to software, well, you could consider current reactor designs to be Windows 98. People are still learning, and it will only get better and safer.

I'd imagine it would be because the company that makes the machines that you're controlling only make drivers and control software for their own special computer systems that you have to buy from them. The advantage there is that if any part of the system goes wrong, from computer to end product, you have a single point of contact to get support from.
I think the mistake that many people on /. make is thinking that everyone is a 'computer guy', where in reality the people running these computers just know how to work the interface itself and not anything below the hood, so calling external support is very valuable.

And I bet a random *nix box that your friend in IT sent you wouldn't pass their ISO xxxxx certification.

As for internet connection, they must all be hooked up to an internal network at least for logging and alerting other systems in case of problems, but having that network also connected to the web is beyond me.

Instead of adding the rules manually, I've been using fail2ban for a few years now, and haven't had any problems. Well, except when I opened a shell to let my father upload pictures to my VPS, and he kept forgetting his password and getting locked out for a while.

It looks like it hasn't been updated in 2 years, but then again the iptables interface hasn't either, so no big deal.

Per radian!

Disclaimer: The viewpoint expressed below is from a graduate student studying mechanical engineering, in a Mechanical, Aerospace, & Nuclear department where many of the department seminars/colloquiums are on nuclear topics. As such it is biased towards scientific reasoning and facts that I have read in peer-reviewed journals or have seen in presentations by people with Ph.D.s on the topic, instead of media-led FUD.

While there are still significant issues to overcome when it comes to building new nuke plants, at this point the largest hurdle is waste and not reactor safety. Why? Because there have been 2 or more generations (depending on how you count them) of plant designs since the GE reactors at Fukushima Daiichi were built. Each with new safety features, and you'd better believe any future plant will learn from what has happened there over the past month. People seem to forget very soon the 3 Mile Island incident just outside Harrisburg PA, which was the worst nuclear disaster ever in the United States. How bad was it? Well, radiation levels around the reactor went up and and are still as high as ... the same as background levels anywhere else in the country, within a noise margin. That's because the plant was designed with safeguards, and as soon as something happened they went into effect. If you look for valid scientific sources about what is going on in Japan right now, it seems like the current knowledge is that the plants were either designed for a "worse case" scenario of a lower magnitude earthquake, or didn't anticipate the (inevitable) tsunami that happens when there's an earthquake on an island. What this means is that future plants will be built with an even larger safety margin and be more expensive, but that is the price to pay for clean, reliable energy.
There are already full plans for single fuel load reactors that don't ever require new rods and will run for their entire lifetime on an initial load, also reacting the fuel to a lower energy state than current designs, leaving the waste with a half-life an order of magnitude shorter than existing reactors. This also means lower temperatures in waste ponds (or no ponds at all until the plant is decommissioned). Unfortunately the time it takes to go from a proposal to a working new plant is on the order of 10 years, in the US at least. Maybe when all the current plants go down and aren't replaced, people will start to realize that they're just afraid of things they don't understand, and possibly learn to accept that "I don't understand it!" does not mean "Nobody understands it, and it's gonna kill me!"

On the flip side, it is also true that if the entire world were to switch to nuclear energy right now, there wouldn't be enough Uranium to go around, but it's still a darn sight better than the current state of petroleum based fuels. And many reports are saying that methane (natural gas) powered plants are a "sustainable" alternative, and current advances in filtration and sequestration make coal plants another short term option that are both more environmentally friendly than many existing gas fueled plants.

So for all the people who say "No nukes, NIMBY!", what DO you propose? If we don't build more of them, we're either going to harm the environment even more with coal/methane/petroleum plants, or run out of power. There isn't enough combined wind, solar, and hydro power available in the continental United States to power the current electricity usage. Period. Even if every square foot of land was covered in one or more of those collection methods. Either we build more nukes, build more smog plants, or drastically decrease usage, with the last option being the best but hilarious if you think it will happen without an even bigger disaster sparking it.