It looks like a hosts file is what you want then. If you really don't want to spend any time on it, just download the MVPS HOSTS file from http://www.mvps.org/winhelp2002/hosts.htm but that blocks pretty much everything. Otherwise just put the few in there that you want to block.
I prefer to support those sites that I like to visit since I'm not paying them. I'm not really going to pay much attention to their ads, but at least they'll make enough to cover the cost of serving the page to me. I just want block the egregious ones. Sounds like you're about the same,
Your wife is one of the relatively few highly dedicated teachers. Most teachers don't put in that much extra time simply because they don't have to, or don't know what to do with their time to become better teachers. One of the problems is that the teacher that clocks in when school starts and leaves as soon as it ends gets paid the same as your wife. Until teachers have the proper incentives to be good teachers and be truly professional about their jobs, the opinion of the profession and the pay will not improve.
The other problem is that almost universally, teachers get paid more only by having more experience. So your wife is probably a new teacher. An experienced teacher can easily make well over twice that much money and not put in that much time. So many teachers are over paid.
The answer is pay for performance and proper incentives, but that is extremely difficult to do well and not have it turn into a system that distorts learning rather than supports it or one where the administrator's or union president's friends are the only one's that get the bonuses. Because of these issues and the unions not wanting to give up power, pay for performance will likely not get implemented any time soon, even though there are answers for the issues.
Of course, I still don't understand why we require teachers to have a bachelors or masters degree to teach grade school
Depends on the subject of course, but you'd be surprised how much expertise you need in math for example to properly teach students the concepts behind the mathematics they are supposed to learn at the grade school level. The current standards call for them to learn number and operations (arithmetic etc), algebra, geometry, and data analysis and probability all with understanding. It doesn't take much mathematical knowledge to be able to do these things yourself, but current research is building that it takes a lot of knowledge to understand what a student is thinking and how best to help them learn a given concept. The knowledge of a subject that a teacher needs is in many ways deeper than that which a practitioner needs. In fact probably the biggest problem in math education today is that most elementary education teachers go into elementary ed partly because they hate math and their math skills are very poor. The bachelor's degree they get only gets them started on the mathematical knowledge they need to teach properly.
Risk is the price of freedom, and the sooner people learn this, the sooner we can move on to improving our civilization.
Taking on certain risks is indeed a price of freedom, but that doesn't mean one shouldn't reduce certain risks as much as possible and then accept the rest. Not reducing those risks that can be reduced without much negative impact on the desired outcome is simply irresponsible. However, the opposite extreme, never doing anything in order to minimize risk is indeed a problem. It's just not the whole problem. Choosing appropriately what risks to accept and which to avoid or reduce is the name of the game. Risk management is the formal study of it and for the most part people are bad at it. By natural human tendencies we underestimate familiar risks like car crashes and over estimate unfamiliar or extreme risks like dying in a plane crash.
In fact a good argument has been made that all progress of civilization boils down to reducing risk. A much lower percentage of the worlds population is at risk of starving to death or even not having enough food each day than perhaps any time in history. In the developed world few people spend much time worrying whether the lights will come on when they flip the switch and whether they will be able to get to work. This reduction of risk where the basics are continually taken care of to a greater degree so that less and less critical things can be focused on is essential to progress. In fact, it basically means more people can spend more time on arts, science, mathematics, or whatever else productive they want that is enabled by the reduced risk of having basic and even not so basic necessities like food and energy.
consumers won't be able to download the operating system
Yeah, the article and thus the summary say that, but it's just the result of really bad journalism. If you read the first FA it links directly to how to download the source. There's even a place to browse the git repository.
Then unless I've grabbed the wrong license file it appears to be under basically a BSDish license.
I should have specified very high resolutions like 2560x1600 and all the effects set on highest quality because thats where the most advanced cards are required and what I meant.
And um, you should really check your sources on the double precision performance of that card. Nvidia doesn't even list the numbers which leads me to believe it's not one of the cards that has hardware double precision support. And since the card doesn't get over a teraflop of single precision floating point there isn't a chance it's over it for double. I think you're thinking of the single precision number. Wikipedia lists the theoretical single precision peak for your card at 705 gigaflops and the 9800 GX2 at 1152 or so. Double is a more difficult challenge for graphics cards, and they've really been lagging at it.
If you read TFA you'll see that Crysis pretty much requires one of these very high end cards to run at high resolutions and better than 30 frames per second.
That's not really what interests me though. The fact that this card has nearly a teraflop of double precision floating point processing power is the most impressive for GPGPU applications. GPGPU applications typically don't care about the single precision number that the card companies like to point out. To figure out what the integer capabilities are you really have to look at the specific problem.
Too bad AMD isn't putting much focus on GPGPU and has only released a beta SDK that still is more difficult to use than CUDA. Nvidia's Tesla only has 78 gigaflops double precision, but it has 4GB of RAM on it. If AMD put the focus on that a bit, they could crush Nvidia.
How exactly could it be more practical? Currently to read a book a blind or visually impaired person that can't read needs a sighted person that can assist them or they need to manually scan a whole book or whatever pages they want. A flatbed scanner is usually slower than snapping a picture. This or a cellphone with similar software on it is a portable solution that makes the same thing possible. Yes they have to wait for the OCR process to know what the page number is or they have to OCR the whole book and search, but I'm not sure how that could be much better. This device at least makes going to the library or finding out what is in a given book or on a piece of paper a little easier and faster.
I guess I also can't see how the device itself isn't visually impaired friendly. I haven't used this particular device but the Nokia phones with the similar software announce the process to you and essentially have a screen reader running on them. When you press a button or navigate through the menus, you hear the options, making them rather accessible. If this device is really targeted to blind or the visually impaired it would have the same functionality.
I believe yes, the software is covered and is similarly expensive. I can't remember which phone it is for, but for one of the high end Nokia phones, a blind friend of mine was showing me some beta software he was testing. It not only did what this Intel device claims to do, but it would also correctly OCR pictures of signs. This is actually a quite a bit more difficult problem given 3D distortions and other imaging issues. It did cost about $1,000 though. One pretty cool thing for blind people was that it correctly recognized paper currency. It's one of those little things you and I don't think about, but blind people have an elaborate folding system for being able to recognize which note is which in their wallet. That feature wasn't really part of the OCR, but something that was built into the image processing. Sorry I don't have any direct links.
I still laugh at the for the blind part. Are the blind expected to know what page they're on with which to have read? "I really wonder what page 47 says".
What exactly makes you laugh about this? If you scan or take a picture of a page and OCR it, that typically includes the page number. If you scan and OCR a whole book, you can ctrl-f for the page number or skip to that page in the document if you have kept the pagination the same. Then you listen to what is on page 47 and no longer wonder.
Blind people are perfectly capable of flipping pages to scan and screen readers are pretty good at telling them what is going on. You should check out what they are able to do before you laugh.
I'm not sure you can legally implement that in GPL since it's a shampoo that I think either my wife or my girlfriend use, I can't remember which.
It seems you were thinking of the Pantone color matching system. Which I can't see how they can claim any intellectual property protection on, but such as it is.
I agree this is thinly veiled shilling. We're supposed to get excited that this phone has managed to catch up to the hardware that other phones already have? What about some innovation like fitting an underclocked OMAP4 core in the phone so that it draws less power and has better performance? OMAP4 has the CortexA9 vs the CortexA8 in the OMAP3 series.
Not only fair, but desperately needed. Not many people know that the NSF funded the creation of 9 math curricula (5 high school and 4 middle school), but part of the idiot requirements of the funding was that they had to be given over to a publisher to publish them. The project could not retain the works and release them for free. I see this ridiculous requirement as essentially stealing high quality math textbooks from the people that paid for them and keeping them away from the students that need them the most.
Basically you look up the internal command names such as file-jpeg-save and what arguments it takes then program either a plugin for gimp or a non interactive script. You can do it in scheme which they refer to as script-fu, or you can write them in python and it's called python-fu. The former is lightly documented and the latter barely at all.
The only way I've found to look up all the command names for the python interface is to run gimp, then go to filters -> python-fu -> console then hit the browse button.
But yeah basically it's either learn to program in scheme or python at this point.
I can't believe no one has mentioned the OpenDisc project, which is the succesor to the OpenCD. It's a project designed to collect the highest quality open source software on a disc. There's even an education version.
Sketchup runs reasonably well in Wine though it's not perfectly easy to set up and it still works best with Nvidia cards apparently. That said, Google could do a proper port to Wine like they have in the past, or better yet a native port so that it runs without issues. But if you'd like to use it, go file some Wine bugs on anything that doesn't work for you.
I looked for the same thing recently without luck. Basically it's a DIY problem and you have to make your own. You can find instructions if you google. The issue is the parts really aren't that cheap on ebay for what I'd get out of it.
I was looking for one because I actually have the opposite problem from the poster. I'm pretty sure I had the exact same model Equity 1 (with a HD though), but maybe it was an Equity 1e or something like that. The problem is I don't have that or any other 5.25 floppy drives now, but I do have all the floppies and software we ever had for the thing. There are a few I wouldn't mind trying to see if the bit rot hasn't killed them and fire them up in dosbox or whatever.
For example we played Nethack 1.0 for hours and hours and I don't find the newer versions as fun, but I can't find the old versions on the net. Also we had a lot of family history data on floppies that we still have but the hard drive backups got lost.
Wikipedia's article about that mostly talks about the FCC restrictions, but I suppose you wouldn't have to worry about that in the developing world. In other words it looks perfectly feasible, but it's a regulatory issue.
So after reading both Wikipedia pages for Moblin and Ubuntu Netbook Remix I can't really tell what the difference is. Are they just two different attempts at the same thing? As I recall Netbook remix came first, then Moblin came out, but that may just be what I recall being anounced.
Either way it certainly is a big deal that Intel is supporting the development of a Microsoft competitor. Mobile computing is an area where Microsoft is very weak. The big problem with Moblin is that it is focused on Atom and the Atom requires an order of magnitude more power than current ARM solutions. It's no wonder that Intel based MID's haven't been able to come up with any decent battery life, the chip plus the chipset and graphics drawing multiple tens of watts is simply never going to be compelling for a mobile device. But when you can put in something that uses milliwatts, then it becomes possible. Of course, it's no surprise that Intel doesn't want to help the competition, but not that Moblin is at the Linux foundation, they really should seriously look at an ARM port.
No, the Slashdot headline is, as usual, misleading. The article didn't really help explain the distinction either. This breakthrough doesn't help anybody break otherwise secure, non homomorphic cryptosystems and suddenly make them insecure. What the researcher did was be the first to create a fully homomorphic cryptosystem that allows the types of things described in the article, while still keeping certain desired information secure. This Wikipedia article gives a much better description of the issue, and you don't even really have to understand abstract algebra to understand that section.
Dear god man, just hit bookmark all tabs, create a bookmark for each of them (in a couple clicks total) and give up. Then you'll still have a record for them if you want them, you'll be able to search them in fact, and then you'll be able to go back to starting with a few tabs or what you actually read.
Instead of replying to myself, I thought I'd add it here. Here's a Linuxdevices article on Intel's upcoming lower power atoms. This is reducing the ridiculous power draw of the chipsets by combining the package into two chips. Quoting
"Even more important, the Pine Trail platform will have a seven-Watt TDP and require an average of just two Watts"
That's after the improvements on an upcoming chip release. The article goes on the say the setup will cost more for Intel to produce. Good luck to them though, I'm still rooting for the race to the greatest performance out of milliwatts.
Gtk does for the most part comply with the Gnome Accessibility API, it just takes some extra tagging and some other extras listed there, certainly not entire re-writes in most cases. Have a look at Glade. While you have to look a little bit in the interface, the accessibility stuff is right there, and basically amounts to making sure everything is labeled for screen readers, etc. I realize Glade may not be what you want to work in, then just go with GTK+. Ok, that has drawbacks sure, but the accessibility part is mostly solved.
Gnome has a decent accessibility framework and apps. Orca is the screen reader and it's pretty decent, but it crashes too much. When it runs, it pretty much works as well or better than JAWS or WindoEyes (the other option on Windows) from what little I know, but doesn't sound quite as good. Sun contributed a ton of the work for bringing Gnome accessibility up to the level it is at, I doubt it would be anywhere near that good without them. I don't think KDE really has anything remotely at the same level.
There is no decent OCR software yet which is necessary for certain visual disabilities, but ocropus is progressing. There is definitely no decent speech recognition as far as I know.
So yes, there is a lot of work to do, part of the problem is that accessibility is a relatively niche problem, and open source really shines on apps and projects that have wide appeal and thus a larger chance of attracting many eyes to look at the code and to help improve it. I'd really love to have an computer that could entirely be controlled without a screen using the screen reader software, but Orca is not yet quite up to the challenge. But at least it doesn't cost $1,000 like JAWS and WindowEyes and it's getting there.
But the submitter's point is good, there does need to be more contribution to this type of thing because these technologies are incredibly liberating for people with disabilities. Even though Orca isn't perfect, you can sit a blind person down at a Linux computer with up to date Gnome setup with accessibility and show them how to use it and they can navigate web pages and interact with many apps that follow the Gnome accessibility API This just happens to be the area I know most about, the submitter, clearly is coming in from another area, but the benefits are similar.
Slashdot Mirror
It looks like a hosts file is what you want then. If you really don't want to spend any time on it, just download the MVPS HOSTS file from http://www.mvps.org/winhelp2002/hosts.htm but that blocks pretty much everything. Otherwise just put the few in there that you want to block.
I prefer to support those sites that I like to visit since I'm not paying them. I'm not really going to pay much attention to their ads, but at least they'll make enough to cover the cost of serving the page to me. I just want block the egregious ones. Sounds like you're about the same,
Your wife is one of the relatively few highly dedicated teachers. Most teachers don't put in that much extra time simply because they don't have to, or don't know what to do with their time to become better teachers. One of the problems is that the teacher that clocks in when school starts and leaves as soon as it ends gets paid the same as your wife. Until teachers have the proper incentives to be good teachers and be truly professional about their jobs, the opinion of the profession and the pay will not improve.
The other problem is that almost universally, teachers get paid more only by having more experience. So your wife is probably a new teacher. An experienced teacher can easily make well over twice that much money and not put in that much time. So many teachers are over paid.
The answer is pay for performance and proper incentives, but that is extremely difficult to do well and not have it turn into a system that distorts learning rather than supports it or one where the administrator's or union president's friends are the only one's that get the bonuses. Because of these issues and the unions not wanting to give up power, pay for performance will likely not get implemented any time soon, even though there are answers for the issues.
Of course, I still don't understand why we require teachers to have a bachelors or masters degree to teach grade school
Depends on the subject of course, but you'd be surprised how much expertise you need in math for example to properly teach students the concepts behind the mathematics they are supposed to learn at the grade school level. The current standards call for them to learn number and operations (arithmetic etc), algebra, geometry, and data analysis and probability all with understanding. It doesn't take much mathematical knowledge to be able to do these things yourself, but current research is building that it takes a lot of knowledge to understand what a student is thinking and how best to help them learn a given concept. The knowledge of a subject that a teacher needs is in many ways deeper than that which a practitioner needs. In fact probably the biggest problem in math education today is that most elementary education teachers go into elementary ed partly because they hate math and their math skills are very poor. The bachelor's degree they get only gets them started on the mathematical knowledge they need to teach properly.
Risk is the price of freedom, and the sooner people learn this, the sooner we can move on to improving our civilization.
Taking on certain risks is indeed a price of freedom, but that doesn't mean one shouldn't reduce certain risks as much as possible and then accept the rest. Not reducing those risks that can be reduced without much negative impact on the desired outcome is simply irresponsible. However, the opposite extreme, never doing anything in order to minimize risk is indeed a problem. It's just not the whole problem. Choosing appropriately what risks to accept and which to avoid or reduce is the name of the game. Risk management is the formal study of it and for the most part people are bad at it. By natural human tendencies we underestimate familiar risks like car crashes and over estimate unfamiliar or extreme risks like dying in a plane crash.
In fact a good argument has been made that all progress of civilization boils down to reducing risk. A much lower percentage of the worlds population is at risk of starving to death or even not having enough food each day than perhaps any time in history. In the developed world few people spend much time worrying whether the lights will come on when they flip the switch and whether they will be able to get to work. This reduction of risk where the basics are continually taken care of to a greater degree so that less and less critical things can be focused on is essential to progress. In fact, it basically means more people can spend more time on arts, science, mathematics, or whatever else productive they want that is enabled by the reduced risk of having basic and even not so basic necessities like food and energy.
consumers won't be able to download the operating system
Yeah, the article and thus the summary say that, but it's just the result of really bad journalism. If you read the first FA it links directly to how to download the source. There's even a place to browse the git repository.
Then unless I've grabbed the wrong license file it appears to be under basically a BSDish license.
I should have specified very high resolutions like 2560x1600 and all the effects set on highest quality because thats where the most advanced cards are required and what I meant.
And um, you should really check your sources on the double precision performance of that card. Nvidia doesn't even list the numbers which leads me to believe it's not one of the cards that has hardware double precision support. And since the card doesn't get over a teraflop of single precision floating point there isn't a chance it's over it for double. I think you're thinking of the single precision number. Wikipedia lists the theoretical single precision peak for your card at 705 gigaflops and the 9800 GX2 at 1152 or so. Double is a more difficult challenge for graphics cards, and they've really been lagging at it.
If you read TFA you'll see that Crysis pretty much requires one of these very high end cards to run at high resolutions and better than 30 frames per second. That's not really what interests me though. The fact that this card has nearly a teraflop of double precision floating point processing power is the most impressive for GPGPU applications. GPGPU applications typically don't care about the single precision number that the card companies like to point out. To figure out what the integer capabilities are you really have to look at the specific problem.
Too bad AMD isn't putting much focus on GPGPU and has only released a beta SDK that still is more difficult to use than CUDA. Nvidia's Tesla only has 78 gigaflops double precision, but it has 4GB of RAM on it. If AMD put the focus on that a bit, they could crush Nvidia.
How exactly could it be more practical? Currently to read a book a blind or visually impaired person that can't read needs a sighted person that can assist them or they need to manually scan a whole book or whatever pages they want. A flatbed scanner is usually slower than snapping a picture. This or a cellphone with similar software on it is a portable solution that makes the same thing possible. Yes they have to wait for the OCR process to know what the page number is or they have to OCR the whole book and search, but I'm not sure how that could be much better. This device at least makes going to the library or finding out what is in a given book or on a piece of paper a little easier and faster.
I guess I also can't see how the device itself isn't visually impaired friendly. I haven't used this particular device but the Nokia phones with the similar software announce the process to you and essentially have a screen reader running on them. When you press a button or navigate through the menus, you hear the options, making them rather accessible. If this device is really targeted to blind or the visually impaired it would have the same functionality.
I believe yes, the software is covered and is similarly expensive. I can't remember which phone it is for, but for one of the high end Nokia phones, a blind friend of mine was showing me some beta software he was testing. It not only did what this Intel device claims to do, but it would also correctly OCR pictures of signs. This is actually a quite a bit more difficult problem given 3D distortions and other imaging issues. It did cost about $1,000 though. One pretty cool thing for blind people was that it correctly recognized paper currency. It's one of those little things you and I don't think about, but blind people have an elaborate folding system for being able to recognize which note is which in their wallet. That feature wasn't really part of the OCR, but something that was built into the image processing. Sorry I don't have any direct links.
I still laugh at the for the blind part. Are the blind expected to know what page they're on with which to have read?
"I really wonder what page 47 says".
What exactly makes you laugh about this? If you scan or take a picture of a page and OCR it, that typically includes the page number. If you scan and OCR a whole book, you can ctrl-f for the page number or skip to that page in the document if you have kept the pagination the same. Then you listen to what is on page 47 and no longer wonder. Blind people are perfectly capable of flipping pages to scan and screen readers are pretty good at telling them what is going on. You should check out what they are able to do before you laugh.
I'm not sure you can legally implement that in GPL since it's a shampoo that I think either my wife or my girlfriend use, I can't remember which.
It seems you were thinking of the Pantone color matching system. Which I can't see how they can claim any intellectual property protection on, but such as it is.
I agree this is thinly veiled shilling. We're supposed to get excited that this phone has managed to catch up to the hardware that other phones already have? What about some innovation like fitting an underclocked OMAP4 core in the phone so that it draws less power and has better performance? OMAP4 has the CortexA9 vs the CortexA8 in the OMAP3 series.
Not only fair, but desperately needed. Not many people know that the NSF funded the creation of 9 math curricula (5 high school and 4 middle school), but part of the idiot requirements of the funding was that they had to be given over to a publisher to publish them. The project could not retain the works and release them for free. I see this ridiculous requirement as essentially stealing high quality math textbooks from the people that paid for them and keeping them away from the students that need them the most.
More about all the high school texts themselves can be found at http://www.ithaca.edu/compass/
Basically you look up the internal command names such as file-jpeg-save and what arguments it takes then program either a plugin for gimp or a non interactive script. You can do it in scheme which they refer to as script-fu, or you can write them in python and it's called python-fu. The former is lightly documented and the latter barely at all. The only way I've found to look up all the command names for the python interface is to run gimp, then go to filters -> python-fu -> console then hit the browse button. But yeah basically it's either learn to program in scheme or python at this point.
I can't believe no one has mentioned the OpenDisc project, which is the succesor to the OpenCD. It's a project designed to collect the highest quality open source software on a disc. There's even an education version.
Sketchup runs reasonably well in Wine though it's not perfectly easy to set up and it still works best with Nvidia cards apparently. That said, Google could do a proper port to Wine like they have in the past, or better yet a native port so that it runs without issues. But if you'd like to use it, go file some Wine bugs on anything that doesn't work for you.
I looked for the same thing recently without luck. Basically it's a DIY problem and you have to make your own. You can find instructions if you google. The issue is the parts really aren't that cheap on ebay for what I'd get out of it.
I was looking for one because I actually have the opposite problem from the poster. I'm pretty sure I had the exact same model Equity 1 (with a HD though), but maybe it was an Equity 1e or something like that. The problem is I don't have that or any other 5.25 floppy drives now, but I do have all the floppies and software we ever had for the thing. There are a few I wouldn't mind trying to see if the bit rot hasn't killed them and fire them up in dosbox or whatever.
For example we played Nethack 1.0 for hours and hours and I don't find the newer versions as fun, but I can't find the old versions on the net. Also we had a lot of family history data on floppies that we still have but the hard drive backups got lost.
Wikipedia's article about that mostly talks about the FCC restrictions, but I suppose you wouldn't have to worry about that in the developing world. In other words it looks perfectly feasible, but it's a regulatory issue.
So after reading both Wikipedia pages for Moblin and Ubuntu Netbook Remix I can't really tell what the difference is. Are they just two different attempts at the same thing? As I recall Netbook remix came first, then Moblin came out, but that may just be what I recall being anounced.
Either way it certainly is a big deal that Intel is supporting the development of a Microsoft competitor. Mobile computing is an area where Microsoft is very weak. The big problem with Moblin is that it is focused on Atom and the Atom requires an order of magnitude more power than current ARM solutions. It's no wonder that Intel based MID's haven't been able to come up with any decent battery life, the chip plus the chipset and graphics drawing multiple tens of watts is simply never going to be compelling for a mobile device. But when you can put in something that uses milliwatts, then it becomes possible. Of course, it's no surprise that Intel doesn't want to help the competition, but not that Moblin is at the Linux foundation, they really should seriously look at an ARM port.
No, the Slashdot headline is, as usual, misleading. The article didn't really help explain the distinction either. This breakthrough doesn't help anybody break otherwise secure, non homomorphic cryptosystems and suddenly make them insecure. What the researcher did was be the first to create a fully homomorphic cryptosystem that allows the types of things described in the article, while still keeping certain desired information secure. This Wikipedia article gives a much better description of the issue, and you don't even really have to understand abstract algebra to understand that section.
Dear god man, just hit bookmark all tabs, create a bookmark for each of them (in a couple clicks total) and give up. Then you'll still have a record for them if you want them, you'll be able to search them in fact, and then you'll be able to go back to starting with a few tabs or what you actually read.
bah!! Don't put the email address up there, post a link to the instructions for how to set up a proxy There's also good material in this Wired article that auric_dude linked below and got that link from about how to get the proxies to who needs them.
"Even more important, the Pine Trail platform will have a seven-Watt TDP and require an average of just two Watts"
That's after the improvements on an upcoming chip release. The article goes on the say the setup will cost more for Intel to produce. Good luck to them though, I'm still rooting for the race to the greatest performance out of milliwatts.
Gtk does for the most part comply with the Gnome Accessibility API, it just takes some extra tagging and some other extras listed there, certainly not entire re-writes in most cases. Have a look at Glade. While you have to look a little bit in the interface, the accessibility stuff is right there, and basically amounts to making sure everything is labeled for screen readers, etc. I realize Glade may not be what you want to work in, then just go with GTK+. Ok, that has drawbacks sure, but the accessibility part is mostly solved.
Gnome has a decent accessibility framework and apps. Orca is the screen reader and it's pretty decent, but it crashes too much. When it runs, it pretty much works as well or better than JAWS or WindoEyes (the other option on Windows) from what little I know, but doesn't sound quite as good. Sun contributed a ton of the work for bringing Gnome accessibility up to the level it is at, I doubt it would be anywhere near that good without them. I don't think KDE really has anything remotely at the same level.
There is no decent OCR software yet which is necessary for certain visual disabilities, but ocropus is progressing. There is definitely no decent speech recognition as far as I know.
So yes, there is a lot of work to do, part of the problem is that accessibility is a relatively niche problem, and open source really shines on apps and projects that have wide appeal and thus a larger chance of attracting many eyes to look at the code and to help improve it. I'd really love to have an computer that could entirely be controlled without a screen using the screen reader software, but Orca is not yet quite up to the challenge. But at least it doesn't cost $1,000 like JAWS and WindowEyes and it's getting there.
But the submitter's point is good, there does need to be more contribution to this type of thing because these technologies are incredibly liberating for people with disabilities. Even though Orca isn't perfect, you can sit a blind person down at a Linux computer with up to date Gnome setup with accessibility and show them how to use it and they can navigate web pages and interact with many apps that follow the Gnome accessibility API This just happens to be the area I know most about, the submitter, clearly is coming in from another area, but the benefits are similar.