Oh, I will concur that it's quite well known modern handwriting applications suck. That's why I respect Microsoft's defunct Transcriber product so much: it did something nobody else has ever bothered to do. Handwriting transcription is a thing that's done terribly; all modern examples are technologically inferior to one prior art. Dragon Naturally Speaking had the same thing going on: it was phenomenal (as much as speech recognition gets), and modern examples never picked it up--because nobody cares about speech recognition (I don't want people to hear my text message conversations, I don't want to sit in my apartment talking to myself, I can actually type faster...).
Swype has never, ever worked anything close to well for me. I hate it. Similarly, fat-fingering a 5mm wide on-screen button with my 20mm wide digit as a stylus.
Transcriber I could scribble fast. Not as fast as I can type on a Dvorak keyboard, but I sure as hell can't screw with a touch screen keyboard like a real, physical keyboard of any layout. The physical keyboard on my Cliq was better, and Transcriber is faster than that.
Transcriber didn't have an error rate measured in percent. It had an error rate measured in errors per week. MY handwriting is unrecognizable to humans, and after writing 20 pages of text I got one error. The damn thing let me scribble page after page of crap for days on a 3 inch screen with a 400MHz ARM Xscale, and it could understand my handwriting more efficiently than I can three days after writing it. I used it when I was doing a lot of writing and didn't own a laptop.
You can claim Swype is faster and more accurate; I can claim Windows has a much lower TCO than Linux. Transcriber wrote what I wrote, even if it was names of Pokemon or names of Seareach Giants or Unix commands, because it could read letters and symbols; Swype uses a dictionary and has to spell check itself. Transcriber relies on a set of 52 phoenetic characters, 10 numerals, and about 32 symbols that I already have memorized and practiced, and can handle congruent (rotation, translation), similar (expanding, shrinking), and incongruent (deforming, smashing, etc) transformations; Swype relies on learning a new muscle movement pattern for each word I want to write, and they all have to be congruent based on the dimensions of the screen and on the same origin.
Draw this exact shape of this exact size in this exact orientation and at this exact spot, and if you're a little off it'll use simplification to remove subtleties, and a dictionary to guess what you mean? Versus scribble something vaguely like the correct shape, and it'll do multiple forms of analysis to determine if it's topologically similar to a particular glyph, even if horribly deformed? Try to make your faerie stories a little more realistic.
Most people ignore the obvious anyway. CO2 and factory output making the world hotter is preposterous.
Think about the natural order. In the natural order, the sun shines on the earth. Trees capture sunlight, bind that energy into sugar. Much of the sunlight is reflected back to space, some is sunk into the ocean and eventually radiates out. Sugar is consumed by plants and animals (yes, plants respire at night for energy--they store sugar to feed themselves). Metabolism is combustion, burning sugar: oxygen combines with CH2O to form CO2 and H2O, releasing heat.
All of this impacts surface temperature. Heat comes from the sun, heat leaves the planet by reflection and radiation. Some heat is absorbed and radiated, keeping the planet at a stable base temperature based on absorption and radiation rate and on the average intensity of sunlight over some short time. Plants capturing sunlight remove heat actively by binding the energy from the sun into chemical compounds rather than absorbing it and radiating warmth. Animals and plants metabolizing such energy-storing chemicals release that heat.
Pave the world in blacktop.
More heat is absorbed into the ground, then radiated. Temperature locally increases.
Drive cars on the road
Have you ever been near a car engine? Riding behind a truck in the winter on my bicycle is awesome. It's so warm. The thing's a god damn oven. They need active cooling with fans and water pumps, plus the oil pan is a radiator and the oil is a cooling fluid. Plus the exhaust vents much of the hot air, although allowing it to expand lets it cool dramatically (it's compressed in the cylinder--initially it's at atmospheric pressure, then it's burned such that even without physical compression by the piston it's now a compressed gas; turbochargers utilize waste heat in that way, since the heat increases pressure and the pressure difference gives potential just like a voltage drop, and there's a turbine in between). Engines are fucking hot we are running thousands of furnaces all over the god damn planet all the time.
Planes. Trucks. Cars. Factories. Power plants. Solar power, seriously, you collect sunlight and turn it into electricity which is always 100% lost as heat (electric heating is 100% perfect efficiency 100% of the time--just if your heating element is low resistance you get more of that heat at the battery terminal than in your actual heater...). Solar energy causes global warming, we collect sunlight that would otherwise be (partially) reflected back to space and burn it.
The world is so hot because everything on this planet is on fire. The only things we do that get cold are heat pumps, sucking heat from a hot area to a more hot area without actually eliminating any of the heat, and that process requires energy input and releases additional heat. Do you know what would happen if we moved entirely to "clean energy" and stopped outputting "greenhouse gasses" and everything ran on nuclear and solar power and we used 10 times as much energy because it's "clean" and abundant? We'd burn the planet to a crisp. You know what would happen if we used wind and tidal energy instead? We'd slow the tides and mess with the weather (energy has to come from somewhere, and in this hypothetical we're pulling a LOT of energy out of the wind). Geothermal? There's no way we could freeze the inside of the planet--it's ridiculous, think about how much hot mass is inside versus just on the crust; we'd bake the planet until the surface flowed and the mountains sagged into pits of molten rock.
Notice Above I did say "used 10 times as much energy". If we make a car 10 times as efficient, that car uses 1/10 as much energy to get the same acceleration and maintain the same speed. If we make the car 10 times as efficient and then put 100 times as many on the road, we use 10 times as much energy. We can only use so much energy before we burn to death.
I for one support the continuous production of highly enriched nuclear material. Enriched nuclear material is not waste, and can be fed back into the reactor to make more energy. We should continue enriching the waste products until we've burned out most of it and have little waste left, though that might take 200,000 years or so. Considering the amount of nuclear material available, we may be able to add fresh material to the pile and have some 70% left over when it comes time for the sun to burn out.
Oh psh. Medicine is supposed to cure or lessen peoples' ailments, and medical research is supposed to further our ability to do just that. These aren't bankers giving loans at 30% interest to people who need money right now or they'll be out on the street, coming back to leech away their last life savings and take their homes in foreclosure. When medical researchers find someone that desperate, they have three possible outcomes: refuse experimental treatment and die now; accept experimental treatment and die now; accept experimental treatment and get miraculously healed for free.
Journey before destination, always. But when it comes to medical research, the service in trade is valuable experimental data. This doesn't cost a dying man anything: if the experiment ends in your death, what have you lost? Butchery it may be, but butchering the young and healthy would be functionally useless. We want to butcher the old and sick.
Actually it's not perfectly reasonable for the user to install a kernel module... not without telling the package manager. Installing a kernel module is perfectly doable, and in DEB and RPM it's done by telling the package manager there's a module to install and letting it put the module somewhere. You can also have a pre-install script (that runs before anything's even installed, and even if the install fails and the packager rolls back changes) or post-install script that spits a kernel module into the appropriate directory, unbeknownst to the package manager. It could even straight out modify the kernel in/boot.
I've actually considered writing a secure package manager. From my point of view, you need to ban running any application as root. The whole thing should run the installation scripts as a non-privileged user with a fakeroot. Track user changes, additions, setuid, etc. pre-install and post-install through bash scripts (maybe with a modified bash), and an 'environment builder' that lets it bring in non-read files for modification (hence a bash interpreter with a sandboxer or modification to prevent writing to the system--can run the thing as root and prevent it from writing to something the packager can read).
That way you could stop and tell the user, "It wants to pull these configs in so it can modify them. They're usually marked sensitive." You can stop and tell the user, "Installation will add system services, kernel modules, and these setuid binaries." You can control the whole installation. World-writable directories need protection though, which is hard; we actually need a system function to supply world-writable protection, and POSIX doesn't supply one (but we're dealing with a Linux package manager, we can always use process namespaces or something... Linux also lets you ban syscalls, so you can block networking. We can add these facilities to Minix if we want to use the packager there).
A couple bumps. You can always detect escapes--or rather you can detect an escape condition. Everything's done in bash scripts? Hell we can use a modified bash or busybox, you can't break out of our policy with that. A preload sandbox or libc could even control bash, perl, awk, the like. You need to run some other binary, something included in the package? The system doesn't have a control to prevent it stomping on world-writable directories? Warn the user: we may lose control of the installation soon, it could stomp all over world-writable directories. It's possible to attach to it with ptrace (POSIX!) and intercept/stop all system calls, but that could get slow. Just watching (strace) isn't as slow, and can give you a report of possible nastiness (files read/altered, network connections made, the like). You can easily detect if the program stayed within its bounds too.
No, Samsung had the internal memo that people hate the stylus and we need to get rid of the stylus. It's what destroyed the viability of PDAs and nerfed what became smartphones.
http://www.youtube.com/watch?v=NuwrWHKraQ8 Microsoft's "Transcriber" handwriting recognition on Windows CE/Pocket PC 6 on Compaq iPaq was better than this, but this is good. I used to scribble a whole cellphone-sized 3 inch screen with chicken scratch, and it could tell what it was. The AI involved is sort of a voting system, a kind of computerized democracy... OCR works by looking at shape, which is great, but with handwriting recognition you can also look at vectors.
Consider a squashed down 'e' or horrifically deformed 'a', while similar (practically rotated), are totally different. The 'e' is one long stroke, right, then up and around counter clockwise--even if it's a mess, jagged, squashed, tilted some, etc. The 'a' is two strokes. The print 'a' that most people draw is almost an 'o' with a vertical stroke, often a counter-clockwise circle with a sudden inflection and clockwise stroke at the end (the clockwise stroke is usually straight--up along the right side, stop and come down). We can also guess orientation, distance between letters, etc if you write more than one letter at once, or just based on screen orientation.
Take all of this, measure how strongly it follows, then combine all that data--vote on it. The shape recognizer feels strongly that it's 'e' or 'a', and the orientation recognizer believes that it's oriented such that it's more likely 'e'. Given that, the shape recognizer feels more strongly that it's 'e' if the orientation recognizer's assumption about which way is up is correct. Meanwhile the vector analyzer thinks there's no way in hell this could be 'a', maybe the more normal stylized 'a', it could be 'o' but it circles too far around and misses overlapping (i.e. the indications that the center stroke in the 'e' isn't intended to be the bottom of the 'o'). 'e' or 'o' but definitely not 'a'. Definitely 'a' or 'e'. Definitely 'e' in the more likely orientation. I think we can discard 'a', and only the senator from Georgia thinks it may be 'o', so we're going to settle on 'e' here.
Well there's waivers and we can write the laws to protect researchers who follow the proper procedure and provide the proper disclosures.
What's wrong with taking advantage of the desperate for medical research? Imagine living with cancer, in pain every day, as your body rots. The doctor says in 4 months your vital organs will fail, maybe sooner, maybe a month or two later. Now this guy shows up with a treatment that may clear the cancer, but could also destroy your nervous system or cause your vital organs to fail outright. It's a good try, at least.
What's the value of 4 months of a dying man's life? The primary value of a life is wages, and in this state you're not working so you're worthless. The primary philosophical value of a life is the struggle to live, which in this case is completely pointless because you're going to die. That an experimental medical procedure might kill you is plainly of no concern: it won't cost you anything, it won't cease your non-existent income, and it's the last possibility you really have for actually living. You might live, or you might just live longer--which both satisfy the survival drive of a dwindling life. It might kill you, but at least somebody tried.
Desperation. It comes when you face death. The cancer is going to kill me in 40 years? Shit I'll be dead in 40 years anyway, who cares? I don't need cancer treatment.
A lot of ethics are imaginary, but on the other hand some ethics are quite arbitrary as a point of necessity. For one thing, you can't save a starving man by feeding him--he's starving for a reason, and if he gets food today that just means tomorrow he will need more food. On the other, there's something to be said for your humanity if you're willing to give it a try against sense. In the same way, we try to avoid blatantly becoming tyrants and monsters to "save" people, because then they get to live in a world full of tyrants and monsters who care nothing about numbers.
On the other hand, if the whole world is going to die, maybe it's better that a tyrant save it... painfully. Try reading Brandon Sanderson's Mistborn series; it's not a straight shot, this happens partially in a few incarnations.
More subtly, we have the natural progression of the original point. We can't let people get taken advantage of, but then we write all these ethics rules and it gets to the point that 10,000 people die painfully a month from something we have 50 cures for but need 10 years to get a tentative approval for small scale clinical testing on. That's 1,200,000 people dead before we start testing? And then more time to review those tests... and if it doesn't work out, any adjustments take years to get approval because we're not going to keep sending people to the slaughter (you said this was gonna work last time, what's different now? Yes but how is that DIFFERENT?).
I don't see a severe ethical problem when we have people suffering and dying in great quantity and we take a few of the desperate ones that won't finish out the year and give them a candidate treatment that probably won't kill them and might cure them, or at least lessen the disease so they live longer and with less pain. If they die, who cares? They were gonna die soon anyway, and we can move on and figure out what went wrong. If we come out ahead, we come out ahead by millions of lives saved; and if we come out behind, in six months we won't be behind anymore because all those people who died in the attempt would have been dead by now anyway.
Not only that, but we warned them and they said try it anyway--they get a chance at life and even if it's a complete and utter failure they become part of the cure, which when you're dead doesn't really matter but when you're facing possible death it's comforting to know that if you don't come out of it alive, you'll still come out of it a hero. You'll find plenty of people who'll take either outcome because that kind of thing means something to them. Sure, being arbitrarily executed for some greater cause at 20 when you have a whole full life ahead of you sucks, regardless of whatever it achieves; but if you've got no way to live for more than a couple months and suddenly we can make your life meaningful on the scale of thousands of lives, that's a hell of a lot more than you could achieve just waiting to die. To some people, it's like giving them their life back--either way they cheat death and accomplish something they weren't given time for.
The loss is small, the gain is great, and the people are strictly volunteers. They stand to gain much, they stand to lose little, and some may find a deeper meaning in being part of medical research if things don't turn out well. I don't see the ethical problem here.
How so? Apple's design for the iPhone circa half a decade ago came at a time when phones had slide-out keyboards and buttons. My Cliq has volume up/down, power, camera, and on the front at the bottom a menu/home/back set. Newer Android phones are all suddenly super thin, no physical keyboard, menu-home-back is pretty integral to the Andorid interface though. Some have eschewed camera, most use volume down to enter the bootloader and boot a recovery system.
Everyone's trying to eliminate physical buttons just like they eliminated the stylus after Samsung declared the stylus must die--eliminating the stylus crippled smartphones and PDAs (back in 1999, you could get a Compaq iPaq with Windows on it with Pocket Word and PocketPC 6, and the handwriting recognition could take my unreadable scribble and interpret it as text proper -- it seemed to be able to read topology, rather than just shape). I used the PDA as a pretty serious professional portable word processor and it was fantastic. Do you see AbiWord on Android with handwriting recognition and a stylus? Just as Samsung led the demise of the stylus, Apple is leading the demise of buttons... starting with physical keyboards.
That's the problem here, really. Unregulated market: lots of nasty shit that's either expensive and ineffective (read: scam) or outright dangerous (read: scam that can kill you) diluting the market, so finding life-saving treatment is nigh on impossible. Regulated market: Twenty years to get life-saving treatments to market, prove it looks safe, find out forty years later it's a terrible idea.
At some point we need to stop being so damn protective of human life and let a few things slip by. Seriously, it shouldn't take lots of experimental tests, animal tests, petri dish pre-clinicals on human tissue, then clinical trials that are still ridiculously hard to get approved, then another decade of bullshit before we can get things to the market. Clinical trial: this is risky as hell, the treatment may be ineffective or dangerous, we've somewhat-kinda tried it on mice and on human tissue, aiming for an exclusive experimental group of volunteers. Experimental stage: Worked out in the clinical trials and after a year or three nobody is showing signs of terrible side-effects, still experimental but we can advise you try this with reasonable certainty that you won't die. Established: It's established, essentially a meaningless upgrade.
Oh, people died? The treatment has problems. Yeah my cancer treatment killed 20 people but let's be realistic: they all had like 4 months to live anyway, no big loss. On the fourth or fifth try I'll get it working right, and then hundreds of thousands of lives will be saved; that's worth a handful of old dudes that were gonna die horribly anyway. Besides, I told them if it didn't work they might die horribly a few months faster; the ones that stayed around flatly did not care, so no sweat off my ass.
RPM and DEB packages run arbitrary Bash scripts as pre-install/post-install during package installation, with full rights to alter the entire system. Gentoo ebuilds use a sandbox prelinked object that prevents writes into the system--it overlays the sandbox through libc function calls, writing new files to a separate directory tree and reading them from the real filesystem if they don't exist in that tree--but you can easily escape this by making direct syscalls.
Making a secure package manager is hard. When you install anything, it gets free reign of your system as root. From there you could even insert kernel modules if you wanted.
You could also consider, that basic software features like an OS, a web browser etc are something that everyone requires these days and should very much be free.
Everybody needs food. Food should be free. We shouldn't pay farmers.
No one can, will, or should stop Steam from coming to Linux. It will never be put in the repos of mainstream distros, and should not be
Pay games are in Ubuntu's repo, like Braid.
The FSF has this ideal where they convince everyone that everything should be free, and then every commercial company stops charging and for some reason continues developing. They don't understand a division between software coming from people who make software for free and software coming from people who make software for money, and assume that people who make software for money can be convinced to make software for free and operate at the same capacity. This is the ideal they push, and the ideal threatened by Steam on Linux.
What they fail to understand is that proprietary software is the enabled of free software. It's commerce: it makes money and makes the platform viable. Linux--and by extension Unix-alikes (BSD, Minix, etc wouldn't be a far jump)--have a vibrant supporting community and a host of free development tools enabling anyone and everyone to get into the game. Windows development is hard--most of the tools are expensive, unless you want to go with MinGW. There's plenty of free software on Linux and plenty of draw to Linux for free software developers--a free development platform is part of that, community (in the sense of everyone else making free stuff for Linux) is another part. Closed, proprietary software won't evaporate that; it'll enable more users to switch to a Linux desktop, thus driving more free software adoption and making free software development more prevalent by exposure of untapped potentials (i.e. people who would become developers if it ever occurred to them).
You claim that the term "Useful" has a particular meaning to a document written over 200 years ago. This is ludicrous. While we're on the topic, let's have a flame war about whether the "right of the people" to "keep and bear arms" refers to individuals for self-defense or a militia for military defense; and if "people" refers to individuals or to "The People" as a whole and thus means that they can keep arms for an organized militia and bear arms against foreign invaders, but not carry for self-defense. No, seriously, this is a common argument between gun nuts and gun control nuts. At one point lawyers were trying to define what a fucking comma was supposed to mean (somebody should have used a semi-colon).
"Useful" does in fact mean that it has utility. If the support of our culture through the creative arts improves society, then it has utility and is thus one of the useful fucking arts.
As a matter of satire, I've previously thoroughly debunked the idea of driving being useful to society. People have thoroughly bunked the idea that exercise is good for you. Hell, I've thoroughly debunked the idea that diets will make you healthy--in a serious fashion, because they're not (you realize dinner and supper are two different things, right? There's two meals in the end of the day...). Economists live and die by flawed theories, like the Keynesian Economics that everybody uses to run their countries into the ground.
Since they are based on faulty concepts that don't actually work
There's a fallacy where you just make an assertion and hope people believe you, you know that?
Also, the 'useful arts' in the context of the US Constitution actually refers to technology, not what we commonly call 'art.' A minor point, but it suggests an unfamiliarity on your part with copyright law's history, and that your support for copyright is likely based upon it being familiar to you and you having never put critical thought into it
Thank you. I've put in critical thought now, and I realize that 'art' is not useful. Literature is a waste of time and I have implemented a policy by which school teachers who purport to have students read in class will be executed, and we'll start rounding up and burning all these worthless tomes so as not to corrupt future generations.
I used Unity in 12.04 and when I hit Maximize or dragged a window to the left, the side bar vanished.
The argument placed was that nobody will use Steam on Linux because nobody can figure out how to use Linux and Windows is familiar. Thus the argument that Windows X to Windows Y is similarly unfamiliar as Windows X to Linux is relevant.
What the fuck does Active Directory and Exchange have to do with playing video games? Do you get paid to sit at work playing Portal? We're talking about home users.
nvidia and AMD will be run aground by Intel HD 3000 on the i5 and i7. On-chip video is getting up there... adding an extra graphics card has become somewhat silly. It's actually confusing the hell out of me, because they packed all the shit that gets ridiculously hot and needs its own active cooling--CPU, north bridge (yes that lone heat sink on your motherboard somewhere random is usually this), graphics--into one space. I was pretty sure this was never going to work without boiling liquid nitrogen. Somehow it does.
Having had to adjust to Vista and Windows 7, I don't feel too bold in saying that switching from Windows to Linux with Gnome 2 or Gnome 3 wouldn't be a stretch for anyone. Windows typically frustrates me, the new Office Ribbon whatever crap is HORRIBLE, etc.
Really, Windows to Gnome 2 isn't a big deal. With Ubuntu or Fedora, pretty much no problems: everything hardware works out of the box or it will never work. More software works out of the box (more file formats work immediately on Linux than Windows, more stuff is installed, etc). On Ubuntu, you can pull up the Ubuntu Software Center and type in vague things like "Games" or "video editor" and it shows you everything ranked by popularity, and you hit Install and it tells you when it's installed (no questions, it just does it, no installers and next next next and do you want this on C: like in Windows).
When it comes to going from XP to Vista or Win7, versus from XP to Ubuntu, I'd say going to Gnome 2 will leave little shock. Windows: Start menu. Gnome 2: Applications, right at the top. And on top of that, the menu is organized better, broken down by type (Office, Internet, Games, etc). Gnome 3 or Unity is going to be more iffy; I dare say Gnome 3 fairs better, but as maligned as Unity is (it really is stupid) it's not a far cry off in this case. Gnome 3 you'll eventually accidentally figure out you can tap the top left corner (which is labeled ACTIVITIES anyway, and you can click in that area for the same effect).
As for a direct comparison between Gnome 3 Gnome-Shell and Ubuntu Unity, the problem with Ubiquity lies in the applications bar on the left vanishing when something overlaps it. Then you have to somehow get into the expanded view or make it pop back up (I haven't figured out how to do the latter). The search box I guess comes up with alt+f2? On Gnome 3 there's a search box right there when you pop up the Activities view, and it takes over the screen if you start to use it.
Gnome 3 is very adaptive to what the user is doing: if you see something and start to use it, it presents you with better context. The expanded Activities view has all your running windows on your desktop, and also on the right you can shift virtual desktops, and you have applications launchers on the left, notifications from applications along the bottom, an "Applications" button to switch to showing you available apps, and a search box in the top right. If you hit the Applications button, it shows you all applications and a list of categories. If you start using the search box, it replaces whatever view you're in with results of all matching applications.
Unity just assumes that a well-designed UI is magically intuitive, and then assumes that they've designed a well-designed UI. It starts working out more once you're used to it, though I eventually gave up before getting too comfortable. Unity's biggest failing just might be not advertising any obvious way to get into the Activities view, which leaves the user kind of floundering around trying to switch windows (no taskbar) or find apps that aren't in the default sidebar, not to mention deal with the sidebar vanishing (it won't come back if you push the mouse on the side of the screen--which would cause its own problems too, but less so than the wtf of just vanishing hard).
All the floundering around with Unity is about how I feel with the transition from XP to Vista or 7. I know how to get to my apps (hit the start menu), everything else in the desktop is alien and has changed a lot. All the configuration settings moved around. I imagine the effect is the same from Win 7 to Gnome Shell... hell, from Gnome 2 to Gnome Shell I was a little uncomfortable, not as bad as Unity but I felt it. Still, I don't think the transition is as terrible as most people want to believe. If I had to make a statement on it, I'd be inclined to say Unity will send people running and Gnome Shell will prove alluring, just because every victory over the initial alienness
This is one strategy. Another is to publish games with a LiveCD option, by which they burn a Fedora or Ubuntu ISO from the game and boot. Fedora or Ubuntu because Debian and CentOS are often behind, and developers will want the latest stuff because hype etc.
It's actually fully possible to boot from an image, too, in which case they could output a $HOME/Valve/Games directory filled with ISOs and put a rudimentary mid-boot-loader in/boot. The mid-boot loader would use syslinux memdisk to load a 64MB hard drive image into RAM and boot from it (you can add a grub entry to do this, yes). That in turn would mount/home or / and scan everything (either under/home/*/Valve/ISO or under/*/Valve/ISO) for games. The user picks a game from the list, the ISO gets mounted, and kexec is used to boot its kernel and begin the process of loading the LiveCD.
From there, a configuration file is loaded based on kernel command line parameters, which points to a directory (like/home/_Valve/) containing all persistent storage (save games, network settings, the like). Hell if you want to get fancy, we can load/etc/passwd and/etc/group from / proper and merge in all UIDs between 500 and 10000 and store saves in the user's $HOME proper, with proper permissions for the user, even make the user log into the system to play. In either case, permanent system settings and game saves are easily accessible. The system could even easily kexec back out into the original loader (or back to the bootloader).
Slashdot Mirror
Oh, I will concur that it's quite well known modern handwriting applications suck. That's why I respect Microsoft's defunct Transcriber product so much: it did something nobody else has ever bothered to do. Handwriting transcription is a thing that's done terribly; all modern examples are technologically inferior to one prior art. Dragon Naturally Speaking had the same thing going on: it was phenomenal (as much as speech recognition gets), and modern examples never picked it up--because nobody cares about speech recognition (I don't want people to hear my text message conversations, I don't want to sit in my apartment talking to myself, I can actually type faster...).
Swype has never, ever worked anything close to well for me. I hate it. Similarly, fat-fingering a 5mm wide on-screen button with my 20mm wide digit as a stylus.
Transcriber I could scribble fast. Not as fast as I can type on a Dvorak keyboard, but I sure as hell can't screw with a touch screen keyboard like a real, physical keyboard of any layout. The physical keyboard on my Cliq was better, and Transcriber is faster than that.
Transcriber didn't have an error rate measured in percent. It had an error rate measured in errors per week. MY handwriting is unrecognizable to humans, and after writing 20 pages of text I got one error. The damn thing let me scribble page after page of crap for days on a 3 inch screen with a 400MHz ARM Xscale, and it could understand my handwriting more efficiently than I can three days after writing it. I used it when I was doing a lot of writing and didn't own a laptop.
You can claim Swype is faster and more accurate; I can claim Windows has a much lower TCO than Linux. Transcriber wrote what I wrote, even if it was names of Pokemon or names of Seareach Giants or Unix commands, because it could read letters and symbols; Swype uses a dictionary and has to spell check itself. Transcriber relies on a set of 52 phoenetic characters, 10 numerals, and about 32 symbols that I already have memorized and practiced, and can handle congruent (rotation, translation), similar (expanding, shrinking), and incongruent (deforming, smashing, etc) transformations; Swype relies on learning a new muscle movement pattern for each word I want to write, and they all have to be congruent based on the dimensions of the screen and on the same origin.
Draw this exact shape of this exact size in this exact orientation and at this exact spot, and if you're a little off it'll use simplification to remove subtleties, and a dictionary to guess what you mean? Versus scribble something vaguely like the correct shape, and it'll do multiple forms of analysis to determine if it's topologically similar to a particular glyph, even if horribly deformed? Try to make your faerie stories a little more realistic.
Yeah and a study done in 2001 found out that airplane exhaust actually causes a global cooling effect. Next.
Most people ignore the obvious anyway. CO2 and factory output making the world hotter is preposterous.
Think about the natural order. In the natural order, the sun shines on the earth. Trees capture sunlight, bind that energy into sugar. Much of the sunlight is reflected back to space, some is sunk into the ocean and eventually radiates out. Sugar is consumed by plants and animals (yes, plants respire at night for energy--they store sugar to feed themselves). Metabolism is combustion, burning sugar: oxygen combines with CH2O to form CO2 and H2O, releasing heat.
All of this impacts surface temperature. Heat comes from the sun, heat leaves the planet by reflection and radiation. Some heat is absorbed and radiated, keeping the planet at a stable base temperature based on absorption and radiation rate and on the average intensity of sunlight over some short time. Plants capturing sunlight remove heat actively by binding the energy from the sun into chemical compounds rather than absorbing it and radiating warmth. Animals and plants metabolizing such energy-storing chemicals release that heat.
Pave the world in blacktop.
More heat is absorbed into the ground, then radiated. Temperature locally increases.
Drive cars on the road
Have you ever been near a car engine? Riding behind a truck in the winter on my bicycle is awesome. It's so warm. The thing's a god damn oven. They need active cooling with fans and water pumps, plus the oil pan is a radiator and the oil is a cooling fluid. Plus the exhaust vents much of the hot air, although allowing it to expand lets it cool dramatically (it's compressed in the cylinder--initially it's at atmospheric pressure, then it's burned such that even without physical compression by the piston it's now a compressed gas; turbochargers utilize waste heat in that way, since the heat increases pressure and the pressure difference gives potential just like a voltage drop, and there's a turbine in between). Engines are fucking hot we are running thousands of furnaces all over the god damn planet all the time.
Planes. Trucks. Cars. Factories. Power plants. Solar power, seriously, you collect sunlight and turn it into electricity which is always 100% lost as heat (electric heating is 100% perfect efficiency 100% of the time--just if your heating element is low resistance you get more of that heat at the battery terminal than in your actual heater...). Solar energy causes global warming, we collect sunlight that would otherwise be (partially) reflected back to space and burn it.
The world is so hot because everything on this planet is on fire. The only things we do that get cold are heat pumps, sucking heat from a hot area to a more hot area without actually eliminating any of the heat, and that process requires energy input and releases additional heat. Do you know what would happen if we moved entirely to "clean energy" and stopped outputting "greenhouse gasses" and everything ran on nuclear and solar power and we used 10 times as much energy because it's "clean" and abundant? We'd burn the planet to a crisp. You know what would happen if we used wind and tidal energy instead? We'd slow the tides and mess with the weather (energy has to come from somewhere, and in this hypothetical we're pulling a LOT of energy out of the wind). Geothermal? There's no way we could freeze the inside of the planet--it's ridiculous, think about how much hot mass is inside versus just on the crust; we'd bake the planet until the surface flowed and the mountains sagged into pits of molten rock.
Notice Above I did say "used 10 times as much energy". If we make a car 10 times as efficient, that car uses 1/10 as much energy to get the same acceleration and maintain the same speed. If we make the car 10 times as efficient and then put 100 times as many on the road, we use 10 times as much energy. We can only use so much energy before we burn to death.
I for one support the continuous production of highly enriched nuclear material. Enriched nuclear material is not waste, and can be fed back into the reactor to make more energy. We should continue enriching the waste products until we've burned out most of it and have little waste left, though that might take 200,000 years or so. Considering the amount of nuclear material available, we may be able to add fresh material to the pile and have some 70% left over when it comes time for the sun to burn out.
Oh psh. Medicine is supposed to cure or lessen peoples' ailments, and medical research is supposed to further our ability to do just that. These aren't bankers giving loans at 30% interest to people who need money right now or they'll be out on the street, coming back to leech away their last life savings and take their homes in foreclosure. When medical researchers find someone that desperate, they have three possible outcomes: refuse experimental treatment and die now; accept experimental treatment and die now; accept experimental treatment and get miraculously healed for free.
Journey before destination, always. But when it comes to medical research, the service in trade is valuable experimental data. This doesn't cost a dying man anything: if the experiment ends in your death, what have you lost? Butchery it may be, but butchering the young and healthy would be functionally useless. We want to butcher the old and sick.
Actually it's not perfectly reasonable for the user to install a kernel module... not without telling the package manager. Installing a kernel module is perfectly doable, and in DEB and RPM it's done by telling the package manager there's a module to install and letting it put the module somewhere. You can also have a pre-install script (that runs before anything's even installed, and even if the install fails and the packager rolls back changes) or post-install script that spits a kernel module into the appropriate directory, unbeknownst to the package manager. It could even straight out modify the kernel in /boot.
I've actually considered writing a secure package manager. From my point of view, you need to ban running any application as root. The whole thing should run the installation scripts as a non-privileged user with a fakeroot. Track user changes, additions, setuid, etc. pre-install and post-install through bash scripts (maybe with a modified bash), and an 'environment builder' that lets it bring in non-read files for modification (hence a bash interpreter with a sandboxer or modification to prevent writing to the system--can run the thing as root and prevent it from writing to something the packager can read).
That way you could stop and tell the user, "It wants to pull these configs in so it can modify them. They're usually marked sensitive." You can stop and tell the user, "Installation will add system services, kernel modules, and these setuid binaries." You can control the whole installation. World-writable directories need protection though, which is hard; we actually need a system function to supply world-writable protection, and POSIX doesn't supply one (but we're dealing with a Linux package manager, we can always use process namespaces or something... Linux also lets you ban syscalls, so you can block networking. We can add these facilities to Minix if we want to use the packager there).
A couple bumps. You can always detect escapes--or rather you can detect an escape condition. Everything's done in bash scripts? Hell we can use a modified bash or busybox, you can't break out of our policy with that. A preload sandbox or libc could even control bash, perl, awk, the like. You need to run some other binary, something included in the package? The system doesn't have a control to prevent it stomping on world-writable directories? Warn the user: we may lose control of the installation soon, it could stomp all over world-writable directories. It's possible to attach to it with ptrace (POSIX!) and intercept/stop all system calls, but that could get slow. Just watching (strace) isn't as slow, and can give you a report of possible nastiness (files read/altered, network connections made, the like). You can easily detect if the program stayed within its bounds too.
No, Samsung had the internal memo that people hate the stylus and we need to get rid of the stylus. It's what destroyed the viability of PDAs and nerfed what became smartphones.
http://www.youtube.com/watch?v=NuwrWHKraQ8 Microsoft's "Transcriber" handwriting recognition on Windows CE/Pocket PC 6 on Compaq iPaq was better than this, but this is good. I used to scribble a whole cellphone-sized 3 inch screen with chicken scratch, and it could tell what it was. The AI involved is sort of a voting system, a kind of computerized democracy... OCR works by looking at shape, which is great, but with handwriting recognition you can also look at vectors.
Consider a squashed down 'e' or horrifically deformed 'a', while similar (practically rotated), are totally different. The 'e' is one long stroke, right, then up and around counter clockwise--even if it's a mess, jagged, squashed, tilted some, etc. The 'a' is two strokes. The print 'a' that most people draw is almost an 'o' with a vertical stroke, often a counter-clockwise circle with a sudden inflection and clockwise stroke at the end (the clockwise stroke is usually straight--up along the right side, stop and come down). We can also guess orientation, distance between letters, etc if you write more than one letter at once, or just based on screen orientation.
Take all of this, measure how strongly it follows, then combine all that data--vote on it. The shape recognizer feels strongly that it's 'e' or 'a', and the orientation recognizer believes that it's oriented such that it's more likely 'e'. Given that, the shape recognizer feels more strongly that it's 'e' if the orientation recognizer's assumption about which way is up is correct. Meanwhile the vector analyzer thinks there's no way in hell this could be 'a', maybe the more normal stylized 'a', it could be 'o' but it circles too far around and misses overlapping (i.e. the indications that the center stroke in the 'e' isn't intended to be the bottom of the 'o'). 'e' or 'o' but definitely not 'a'. Definitely 'a' or 'e'. Definitely 'e' in the more likely orientation. I think we can discard 'a', and only the senator from Georgia thinks it may be 'o', so we're going to settle on 'e' here.
It's the one thing Microsoft did that I still hold respect for. This is what I want to see on Android phones and tablets: http://s1.subirimagenes.com/imagen/414980transcriber.gif http://images.zatz.com/websites/pocketpclife/issues/issue200009/transcriber-a.gif
Well there's waivers and we can write the laws to protect researchers who follow the proper procedure and provide the proper disclosures.
What's wrong with taking advantage of the desperate for medical research? Imagine living with cancer, in pain every day, as your body rots. The doctor says in 4 months your vital organs will fail, maybe sooner, maybe a month or two later. Now this guy shows up with a treatment that may clear the cancer, but could also destroy your nervous system or cause your vital organs to fail outright. It's a good try, at least.
What's the value of 4 months of a dying man's life? The primary value of a life is wages, and in this state you're not working so you're worthless. The primary philosophical value of a life is the struggle to live, which in this case is completely pointless because you're going to die. That an experimental medical procedure might kill you is plainly of no concern: it won't cost you anything, it won't cease your non-existent income, and it's the last possibility you really have for actually living. You might live, or you might just live longer--which both satisfy the survival drive of a dwindling life. It might kill you, but at least somebody tried.
Desperation. It comes when you face death. The cancer is going to kill me in 40 years? Shit I'll be dead in 40 years anyway, who cares? I don't need cancer treatment.
A lot of ethics are imaginary, but on the other hand some ethics are quite arbitrary as a point of necessity. For one thing, you can't save a starving man by feeding him--he's starving for a reason, and if he gets food today that just means tomorrow he will need more food. On the other, there's something to be said for your humanity if you're willing to give it a try against sense. In the same way, we try to avoid blatantly becoming tyrants and monsters to "save" people, because then they get to live in a world full of tyrants and monsters who care nothing about numbers.
On the other hand, if the whole world is going to die, maybe it's better that a tyrant save it ... painfully. Try reading Brandon Sanderson's Mistborn series; it's not a straight shot, this happens partially in a few incarnations.
More subtly, we have the natural progression of the original point. We can't let people get taken advantage of, but then we write all these ethics rules and it gets to the point that 10,000 people die painfully a month from something we have 50 cures for but need 10 years to get a tentative approval for small scale clinical testing on. That's 1,200,000 people dead before we start testing? And then more time to review those tests... and if it doesn't work out, any adjustments take years to get approval because we're not going to keep sending people to the slaughter (you said this was gonna work last time, what's different now? Yes but how is that DIFFERENT?).
I don't see a severe ethical problem when we have people suffering and dying in great quantity and we take a few of the desperate ones that won't finish out the year and give them a candidate treatment that probably won't kill them and might cure them, or at least lessen the disease so they live longer and with less pain. If they die, who cares? They were gonna die soon anyway, and we can move on and figure out what went wrong. If we come out ahead, we come out ahead by millions of lives saved; and if we come out behind, in six months we won't be behind anymore because all those people who died in the attempt would have been dead by now anyway.
Not only that, but we warned them and they said try it anyway--they get a chance at life and even if it's a complete and utter failure they become part of the cure, which when you're dead doesn't really matter but when you're facing possible death it's comforting to know that if you don't come out of it alive, you'll still come out of it a hero. You'll find plenty of people who'll take either outcome because that kind of thing means something to them. Sure, being arbitrarily executed for some greater cause at 20 when you have a whole full life ahead of you sucks, regardless of whatever it achieves; but if you've got no way to live for more than a couple months and suddenly we can make your life meaningful on the scale of thousands of lives, that's a hell of a lot more than you could achieve just waiting to die. To some people, it's like giving them their life back--either way they cheat death and accomplish something they weren't given time for.
The loss is small, the gain is great, and the people are strictly volunteers. They stand to gain much, they stand to lose little, and some may find a deeper meaning in being part of medical research if things don't turn out well. I don't see the ethical problem here.
How so? Apple's design for the iPhone circa half a decade ago came at a time when phones had slide-out keyboards and buttons. My Cliq has volume up/down, power, camera, and on the front at the bottom a menu/home/back set. Newer Android phones are all suddenly super thin, no physical keyboard, menu-home-back is pretty integral to the Andorid interface though. Some have eschewed camera, most use volume down to enter the bootloader and boot a recovery system.
Everyone's trying to eliminate physical buttons just like they eliminated the stylus after Samsung declared the stylus must die--eliminating the stylus crippled smartphones and PDAs (back in 1999, you could get a Compaq iPaq with Windows on it with Pocket Word and PocketPC 6, and the handwriting recognition could take my unreadable scribble and interpret it as text proper -- it seemed to be able to read topology, rather than just shape). I used the PDA as a pretty serious professional portable word processor and it was fantastic. Do you see AbiWord on Android with handwriting recognition and a stylus? Just as Samsung led the demise of the stylus, Apple is leading the demise of buttons... starting with physical keyboards.
Read Warbreaker.
That's the problem here, really. Unregulated market: lots of nasty shit that's either expensive and ineffective (read: scam) or outright dangerous (read: scam that can kill you) diluting the market, so finding life-saving treatment is nigh on impossible. Regulated market: Twenty years to get life-saving treatments to market, prove it looks safe, find out forty years later it's a terrible idea.
At some point we need to stop being so damn protective of human life and let a few things slip by. Seriously, it shouldn't take lots of experimental tests, animal tests, petri dish pre-clinicals on human tissue, then clinical trials that are still ridiculously hard to get approved, then another decade of bullshit before we can get things to the market. Clinical trial: this is risky as hell, the treatment may be ineffective or dangerous, we've somewhat-kinda tried it on mice and on human tissue, aiming for an exclusive experimental group of volunteers. Experimental stage: Worked out in the clinical trials and after a year or three nobody is showing signs of terrible side-effects, still experimental but we can advise you try this with reasonable certainty that you won't die. Established: It's established, essentially a meaningless upgrade.
Oh, people died? The treatment has problems. Yeah my cancer treatment killed 20 people but let's be realistic: they all had like 4 months to live anyway, no big loss. On the fourth or fifth try I'll get it working right, and then hundreds of thousands of lives will be saved; that's worth a handful of old dudes that were gonna die horribly anyway. Besides, I told them if it didn't work they might die horribly a few months faster; the ones that stayed around flatly did not care, so no sweat off my ass.
RPM and DEB packages run arbitrary Bash scripts as pre-install/post-install during package installation, with full rights to alter the entire system. Gentoo ebuilds use a sandbox prelinked object that prevents writes into the system--it overlays the sandbox through libc function calls, writing new files to a separate directory tree and reading them from the real filesystem if they don't exist in that tree--but you can easily escape this by making direct syscalls.
Making a secure package manager is hard. When you install anything, it gets free reign of your system as root. From there you could even insert kernel modules if you wanted.
Wine doesn't run as root though (I tried, it actually screams and exits immediately). Wine has a mapping to $HOME that you need to remove though...
You could also consider, that basic software features like an OS, a web browser etc are something that everyone requires these days and should very much be free.
Everybody needs food. Food should be free. We shouldn't pay farmers.
No one can, will, or should stop Steam from coming to Linux. It will never be put in the repos of mainstream distros, and should not be
Pay games are in Ubuntu's repo, like Braid.
The FSF has this ideal where they convince everyone that everything should be free, and then every commercial company stops charging and for some reason continues developing. They don't understand a division between software coming from people who make software for free and software coming from people who make software for money, and assume that people who make software for money can be convinced to make software for free and operate at the same capacity. This is the ideal they push, and the ideal threatened by Steam on Linux.
What they fail to understand is that proprietary software is the enabled of free software. It's commerce: it makes money and makes the platform viable. Linux--and by extension Unix-alikes (BSD, Minix, etc wouldn't be a far jump)--have a vibrant supporting community and a host of free development tools enabling anyone and everyone to get into the game. Windows development is hard--most of the tools are expensive, unless you want to go with MinGW. There's plenty of free software on Linux and plenty of draw to Linux for free software developers--a free development platform is part of that, community (in the sense of everyone else making free stuff for Linux) is another part. Closed, proprietary software won't evaporate that; it'll enable more users to switch to a Linux desktop, thus driving more free software adoption and making free software development more prevalent by exposure of untapped potentials (i.e. people who would become developers if it ever occurred to them).
You claim that the term "Useful" has a particular meaning to a document written over 200 years ago. This is ludicrous. While we're on the topic, let's have a flame war about whether the "right of the people" to "keep and bear arms" refers to individuals for self-defense or a militia for military defense; and if "people" refers to individuals or to "The People" as a whole and thus means that they can keep arms for an organized militia and bear arms against foreign invaders, but not carry for self-defense. No, seriously, this is a common argument between gun nuts and gun control nuts. At one point lawyers were trying to define what a fucking comma was supposed to mean (somebody should have used a semi-colon).
"Useful" does in fact mean that it has utility. If the support of our culture through the creative arts improves society, then it has utility and is thus one of the useful fucking arts.
As a matter of satire, I've previously thoroughly debunked the idea of driving being useful to society. People have thoroughly bunked the idea that exercise is good for you. Hell, I've thoroughly debunked the idea that diets will make you healthy--in a serious fashion, because they're not (you realize dinner and supper are two different things, right? There's two meals in the end of the day...). Economists live and die by flawed theories, like the Keynesian Economics that everybody uses to run their countries into the ground.
Since they are based on faulty concepts that don't actually work
There's a fallacy where you just make an assertion and hope people believe you, you know that?
Also, the 'useful arts' in the context of the US Constitution actually refers to technology, not what we commonly call 'art.' A minor point, but it suggests an unfamiliarity on your part with copyright law's history, and that your support for copyright is likely based upon it being familiar to you and you having never put critical thought into it
Thank you. I've put in critical thought now, and I realize that 'art' is not useful. Literature is a waste of time and I have implemented a policy by which school teachers who purport to have students read in class will be executed, and we'll start rounding up and burning all these worthless tomes so as not to corrupt future generations.
I used Unity in 12.04 and when I hit Maximize or dragged a window to the left, the side bar vanished.
The argument placed was that nobody will use Steam on Linux because nobody can figure out how to use Linux and Windows is familiar. Thus the argument that Windows X to Windows Y is similarly unfamiliar as Windows X to Linux is relevant.
What the fuck does Active Directory and Exchange have to do with playing video games? Do you get paid to sit at work playing Portal? We're talking about home users.
Therefor, obviously, content creators have no rights to the content they create. Thus ends the useful arts.
nvidia and AMD will be run aground by Intel HD 3000 on the i5 and i7. On-chip video is getting up there... adding an extra graphics card has become somewhat silly. It's actually confusing the hell out of me, because they packed all the shit that gets ridiculously hot and needs its own active cooling--CPU, north bridge (yes that lone heat sink on your motherboard somewhere random is usually this), graphics--into one space. I was pretty sure this was never going to work without boiling liquid nitrogen. Somehow it does.
Having had to adjust to Vista and Windows 7, I don't feel too bold in saying that switching from Windows to Linux with Gnome 2 or Gnome 3 wouldn't be a stretch for anyone. Windows typically frustrates me, the new Office Ribbon whatever crap is HORRIBLE, etc.
Really, Windows to Gnome 2 isn't a big deal. With Ubuntu or Fedora, pretty much no problems: everything hardware works out of the box or it will never work. More software works out of the box (more file formats work immediately on Linux than Windows, more stuff is installed, etc). On Ubuntu, you can pull up the Ubuntu Software Center and type in vague things like "Games" or "video editor" and it shows you everything ranked by popularity, and you hit Install and it tells you when it's installed (no questions, it just does it, no installers and next next next and do you want this on C: like in Windows).
When it comes to going from XP to Vista or Win7, versus from XP to Ubuntu, I'd say going to Gnome 2 will leave little shock. Windows: Start menu. Gnome 2: Applications, right at the top. And on top of that, the menu is organized better, broken down by type (Office, Internet, Games, etc). Gnome 3 or Unity is going to be more iffy; I dare say Gnome 3 fairs better, but as maligned as Unity is (it really is stupid) it's not a far cry off in this case. Gnome 3 you'll eventually accidentally figure out you can tap the top left corner (which is labeled ACTIVITIES anyway, and you can click in that area for the same effect).
As for a direct comparison between Gnome 3 Gnome-Shell and Ubuntu Unity, the problem with Ubiquity lies in the applications bar on the left vanishing when something overlaps it. Then you have to somehow get into the expanded view or make it pop back up (I haven't figured out how to do the latter). The search box I guess comes up with alt+f2? On Gnome 3 there's a search box right there when you pop up the Activities view, and it takes over the screen if you start to use it.
Gnome 3 is very adaptive to what the user is doing: if you see something and start to use it, it presents you with better context. The expanded Activities view has all your running windows on your desktop, and also on the right you can shift virtual desktops, and you have applications launchers on the left, notifications from applications along the bottom, an "Applications" button to switch to showing you available apps, and a search box in the top right. If you hit the Applications button, it shows you all applications and a list of categories. If you start using the search box, it replaces whatever view you're in with results of all matching applications.
Unity just assumes that a well-designed UI is magically intuitive, and then assumes that they've designed a well-designed UI. It starts working out more once you're used to it, though I eventually gave up before getting too comfortable. Unity's biggest failing just might be not advertising any obvious way to get into the Activities view, which leaves the user kind of floundering around trying to switch windows (no taskbar) or find apps that aren't in the default sidebar, not to mention deal with the sidebar vanishing (it won't come back if you push the mouse on the side of the screen--which would cause its own problems too, but less so than the wtf of just vanishing hard).
All the floundering around with Unity is about how I feel with the transition from XP to Vista or 7. I know how to get to my apps (hit the start menu), everything else in the desktop is alien and has changed a lot. All the configuration settings moved around. I imagine the effect is the same from Win 7 to Gnome Shell ... hell, from Gnome 2 to Gnome Shell I was a little uncomfortable, not as bad as Unity but I felt it. Still, I don't think the transition is as terrible as most people want to believe. If I had to make a statement on it, I'd be inclined to say Unity will send people running and Gnome Shell will prove alluring, just because every victory over the initial alienness
This is one strategy. Another is to publish games with a LiveCD option, by which they burn a Fedora or Ubuntu ISO from the game and boot. Fedora or Ubuntu because Debian and CentOS are often behind, and developers will want the latest stuff because hype etc.
It's actually fully possible to boot from an image, too, in which case they could output a $HOME/Valve/Games directory filled with ISOs and put a rudimentary mid-boot-loader in /boot. The mid-boot loader would use syslinux memdisk to load a 64MB hard drive image into RAM and boot from it (you can add a grub entry to do this, yes). That in turn would mount /home or / and scan everything (either under /home/*/Valve/ISO or under /*/Valve/ISO) for games. The user picks a game from the list, the ISO gets mounted, and kexec is used to boot its kernel and begin the process of loading the LiveCD.
From there, a configuration file is loaded based on kernel command line parameters, which points to a directory (like /home/_Valve/) containing all persistent storage (save games, network settings, the like). Hell if you want to get fancy, we can load /etc/passwd and /etc/group from / proper and merge in all UIDs between 500 and 10000 and store saves in the user's $HOME proper, with proper permissions for the user, even make the user log into the system to play. In either case, permanent system settings and game saves are easily accessible. The system could even easily kexec back out into the original loader (or back to the bootloader).