I lost track of the comment, but someone said they had seen a one-legged robot hopping and balancing.
In the late 1980s, Marc Raibert's Leg Lab had robots with one, two, and four legs which hopped, ran, and did somersaults. One of the things I worked on there was the idea that if you build the robot correctly, you need less control... actually, inspired by weebles wobbling (someone else's post about "prior art") it turns out that you don't need any computer to make a one-legged robot hop stably. I even whacked it with a stick, and it didn't fall over.
But this, this is a different thing. This robot as reacting in a way which is remarkably like what I would expect of a human. I liked the point where it widened its stance - that is impressive, because it had to decide that it was a good idea, and then do it despite the force being applied.
I have RSI. As a graduate student working on my Master's thesis, the pain was so bad I would go home after typing half a day. I was scared and a little depressed. Many of my friends at MIT had similar problems.
What I found is that there is no generic solution which works for everyone. I have a collection of things people have found useful.
0. Listen to your hands. When it hurts, stop. I made a habit of paying attention to feelings from my hands, and there's a funny feeling I get right before they start hurting. If you've got RSI, you have probably trained yourself to ignore this warning.
1. Keyboards. I use a dathand and love it. It is expensive, but almost immediately I could type all day. Other people I know prefer Kinesis (I still don't understand where they put their thumbs while typing...) or the Microsoft Natural, both of which are substantially lower cost.
2. Gently stretch your hands and wrists before typing. You're about to use them for hours; if it were for a sport you would stretch first.
3. HandEze gloves are very much like the ones described in this article, without the padding. They don't seem like they should work, but they helped me a lot. I used them for months until I no longer needed them. They're really cheap, it is worth the money to see if they help. If the seams bother you, turn them inside out, and if you don't like the color you can always dump them into some dye.
4. Posture. Occasionally I have a relapse. Most of the time I realize that it's because I've stopped sitting correctly.
5. Exercise. Occasionally I spend a weekend running around in the woods with a bunch of friends, and I noticed that every time my hands stopped hurting. Now I bicycle to work when the weather permits. Bicycling and going to the gym do not work as well for me as running around in the woods, so there is something else there - I'll be examining another poster's comment about insulin, which I had not heard before.
6. Ibuprofin. When my hands hurt at night (hasn't happened for a long time now) I take Ibuprofin before going to bed to reduce swelling and let the tendons heal. Don't do this when you're going to actually use your hands, it masks the warning signals that you are doing something bad.
7. Tension. Because I'd been through so much pain at keyboards, when I sat down in front of a keyboard my hands immediately tensed up in anticipation of pain. This did not help. For me, the Datahand was a different enough keyboard to break that cycle.
8. Arm rests. I took the arm rests off my chair. They were compressing my ulnar nerve (outside of the elbow) and causing numbness and pain in my pinky fingers.
9. Breaks. Every half hour, take a five-minute break. One of the reasons I listen to CDs in the office (on headphones) is that when a CD ends it forms a good pausing point at which you can ask "did I take breaks during the last hour?".
RSI takes years to develop. It is taking even more years to go away. But, for me, it is going away.
My opinion on wrist splints: use them at night if you wake up with hand/arm discomfort. Do not use them during the day, because they can let your muscles atrophy.
I have no relation to any of the companies which make the products I mention here, except as a user.
When I was in the MIT Leg Lab, I did some rendering work for our animations of legged robots. I recall reading a paper on NTSC (I think it was titled "NTSC: Nice Technology, Super Color") which pointed out how to avoid some of the common sources of off-color pixels.
The biggest is that if the colors and intensities change rapidly as you move across the screen, the decoders have difficulty pulling the proper hue/intensity/value out of the signal. Not a problem for real life, but extremely common for computer generated images. Visually, it means that when you render a white square on a dark background the left and right edges of the square aren't quite the right color. For the Leg Lab, I made a filter which clamped down on such changes and rendered side-by-side with and without the filter. The difference was visible.
Does anyone here know if the makers of video cards and console games have started prefiltering their outputs in anticipation of how they're going to be transmitted and displayed?
(There is more you can do based on the fact that the human eye does a Fourier transform in time of what it sees - reverse the transform and you can make images which only make sense when played back at full speed and are hopeless blurs otherwise. That's how we managed to get the ostrich legs to look good. But I digress.)
When our remote died, they sent us a replacement for free - even though we were off warranty.
I'm more confident of ReplayTV's privacy policies than TiVo's.
Don't forget the cost of the service (realistically, you'll buy the lifetime version) when you do cost comparisons. The boxes lose a lot of functionality without the TV guide service.
If you really _do_ intend to hack the box, sure, get TiVo. But if you just want a box you can plug in and record things, I like Replay.
TiVo did (I don't know if it still does) send every single button push received from the remote back to the servers, time stamped.
I use a Replay. Right or wrong, I have more confidence that they'll keep my private information, well, private.
Big source files. No, bigger than that.
on
Pet Bugs?
·
· Score: 1
While I was a graduate student at MIT, I was using SD/Fast, a program which wrote C code for simulating system dynamics. The input and output formats of SD/Fast could have been more user friendly (I have no idea if they are now), but I had written software which made it relatively easy to create input files, even large ones, for SD/Fast and integrate the results into physically realistic simulations. You could describe the creature you wanted, and it would eventually write everything except the control system itself. Including making input to SD/Fast, running SD/Fast, and incorporating the results.
In one case, someone was (ab)using it to simulate a tree with 100+ degrees of freedom. The file generated by SD/Fast was multiple megabytes of undecipherable C source code. We couldn't even figure it out enough to break it into multiple files (and the process of creating it was thoroughly automated, we didn't really want to).
So we showed it to the Sun C compiler. It churned for about an hour... and ran out of some internal resource. I never found out what, nor did I find out how to resize it.
So we showed it to the IBM C compiler running on the machine across the room. Lo and behold, it compiled and ran, doing a good imitation of Zippy the Wonder Slug.
So we took the -g flag out. After all, its working, we don't need debugging information.
Fortunately I'd limited the core dump size. Otherwise, some of the users who shared that disk space would have been really annoyed when it segfaulted.
IBM wanted the program, so they could try and figure out why it was different with the debugging information included.... We advised them of the size of the source code and the number of libraries in the MIT AI lab it depended on, and they said "Uh... we'll send you this patch which might be it."
This patching process happened a few times. Eventually, the program worked and the poor undergraduate who was trying to do this finished his project.
As a closing note, soon after that SD/Fast sprouted options to break the output into multiple files.
Many people here have listed "no distractions" as their #1 requirement for zoning.
I agree, with the following two caveats: 1. I need a minor distraction every 45-60 minutes to remind me to give my hands a break. Otherwise, I'll get an RSI relapse. 2. Sometimes, I end up thinking very hard about a problem - so hard that I fail to see the obvious "almost as good" alternative. I am stuck down in the weeds, when I should be looking for a solution in the bigger picture. I do have the bigger picture in my head but I am concentrating so hard on a small part of it that hard I do not realize I should step back.
My solution? CD player and headphones. Every 45-60 minutes, you have to change the CD. Time for a hand break if you haven't taken one. If I am concentrating on something too hard, eventually the music will distract me enough to make me reevaluate the situation. I find music, used this way, a gentle enough distraction that I slip back into coding extremely quickly.
Of course, listening to music doesn't address problems related to co-workers bothering you... other people have good ideas there.
I also find a well-chosen set of headphones filters out a lot of the ambient noise, even if you aren't listening to music.
My impression was that if you had Onstar, and your car was stolen, you could call them and they'd kill the engine remotely.
But I could be wrong, I don't have the service. Would someone who has details care to comment?
Re: Ergonomics (offtopic RSI experience)
on
To The Pain
·
· Score: 1
... and nex time, I will use "preview".
Honest, the original message had carriage returns in it.
Re: Ergonomics (offtopic RSI experience)
on
To The Pain
·
· Score: 1
I agree that "today's computer interfaces are an ergonomic nightmare." However, to some extent they can be corrected now, without going through voice recognition. Change the keyboard to something designed for people, not something designed for a mechanical typewriter.
I gave myself a fairly bad case of RSI while working on my Master's thesis. Since changing to a Datahand keyboard, my hands have been slowly getting better rather than worse. Other people I know prefer Kinesis.
Even more offtopic:
If you have RSI, the most important things I figured out are:
1. Pay attention to your hands. There's a feeling of discomfort before your hands start to hurt. As you get used to being in pain, you get used to ignoring this discomfort... bad idea. An extra hour of typing is not worth several days of relapse.
2. Sit properly. Move your keyboard, chair, monitor, and desk as necessary. If your upper arms are long compared to your torso, like mine, you may want your keyboard in your lap.
3. If your hands tense up when you start using a keyboard, it may be in anticipation of pain. If so, try switching to a differently arranged keyboard (such as a Kinesis or Datahand, for example) to remove that association.
4. Be aware that extended periods of pain can cause depression. RSIs take years to develop, and for me it has taken years to go away.
5. Go see a doctor. Get your RSI documented, so that later if it has gotten worse you have a visit on record saying it has been an ongoing condition. Read books on RSI, in case your doctor is not familiar enough with RSI to provide useful advice.
6. Pain is a warning that you are doing damage. Ibuprofin reduces swelling, which is good, but also keeps you from getting the feedback that you are damaging your hands. I don't take Ibuprofin unless I am not going to be typing for a while.
7. Every half hour/45 minutes, get up and walk around for a couple minutes. Personally, I listen to CDs as I work - when the CD ends, if I haven't taken a break I should.
8. Examine the other activities of your life. I discovered that I hold paperback books in a way which puts a lot of strain on my thumb and pinky, and changed that.
9. Aerobic exercise seems to help me, and is a good idea even if you don't have RSI.
Caveat: I am not a medical doctor. I am simply stating the steps I took to go from stopping after a few hours because of pain, to consistently working 8+ hour days writing software.
I mention Datahand and Kinesis keyboards because they are the ergonomic keyboards I see most often. My only relation to Datahand is that of an extremely satisfied customer. I don't have a relationship with Kinesis.
Slashdot Mirror
I lost track of the comment, but someone said they had seen a one-legged robot hopping and balancing.
In the late 1980s, Marc Raibert's Leg Lab had robots with one, two, and four legs which hopped, ran, and did somersaults. One of the things I worked on there was the idea that if you build the robot correctly, you need less control... actually, inspired by weebles wobbling (someone else's post about "prior art") it turns out that you don't need any computer to make a one-legged robot hop stably. I even whacked it with a stick, and it didn't fall over.
But this, this is a different thing. This robot as reacting in a way which is remarkably like what I would expect of a human. I liked the point where it widened its stance - that is impressive, because it had to decide that it was a good idea, and then do it despite the force being applied.
Caveat: I am not a doctor.
I have RSI. As a graduate student working on my Master's thesis, the pain was so bad I would go home after typing half a day. I was scared and a little depressed. Many of my friends at MIT had similar problems.
What I found is that there is no generic solution which works for everyone. I have a collection of things people have found useful.
0. Listen to your hands. When it hurts, stop. I made a habit of paying attention to feelings from my hands, and there's a funny feeling I get right before they start hurting. If you've got RSI, you have probably trained yourself to ignore this warning.
1. Keyboards. I use a dathand and love it. It is expensive, but almost immediately I could type all day. Other people I know prefer Kinesis (I still don't understand where they put their thumbs while typing...) or the Microsoft Natural, both of which are substantially lower cost.
2. Gently stretch your hands and wrists before typing. You're about to use them for hours; if it were for a sport you would stretch first.
3. HandEze gloves are very much like the ones described in this article, without the padding. They don't seem like they should work, but they helped me a lot. I used them for months until I no longer needed them. They're really cheap, it is worth the money to see if they help. If the seams bother you, turn them inside out, and if you don't like the color you can always dump them into some dye.
4. Posture. Occasionally I have a relapse. Most of the time I realize that it's because I've stopped sitting correctly.
5. Exercise. Occasionally I spend a weekend running around in the woods with a bunch of friends, and I noticed that every time my hands stopped hurting. Now I bicycle to work when the weather permits. Bicycling and going to the gym do not work as well for me as running around in the woods, so there is something else there - I'll be examining another poster's comment about insulin, which I had not heard before.
6. Ibuprofin. When my hands hurt at night (hasn't happened for a long time now) I take Ibuprofin before going to bed to reduce swelling and let the tendons heal. Don't do this when you're going to actually use your hands, it masks the warning signals that you are doing something bad.
7. Tension. Because I'd been through so much pain at keyboards, when I sat down in front of a keyboard my hands immediately tensed up in anticipation of pain. This did not help. For me, the Datahand was a different enough keyboard to break that cycle.
8. Arm rests. I took the arm rests off my chair. They were compressing my ulnar nerve (outside of the elbow) and causing numbness and pain in my pinky fingers.
9. Breaks. Every half hour, take a five-minute break. One of the reasons I listen to CDs in the office (on headphones) is that when a CD ends it forms a good pausing point at which you can ask "did I take breaks during the last hour?".
RSI takes years to develop. It is taking even more years to go away. But, for me, it is going away.
My opinion on wrist splints: use them at night if you wake up with hand/arm discomfort. Do not use them during the day, because they can let your muscles atrophy.
I have no relation to any of the companies which make the products I mention here, except as a user.
Are you sure you have no problem with selling compiled demographics?
Say, selling average income by zip code?
What about places which are sparsely populated, and zip+4 is enough to identify a single household, identifying it as a theft target?
What happens when there's another way to slice the information you didn't consider, which _does_ make your house identifiable?
When I was in the MIT Leg Lab, I did some rendering work for our animations of legged robots. I recall reading a paper on NTSC (I think it was titled "NTSC: Nice Technology, Super Color") which pointed out how to avoid some of the common sources of off-color pixels.
The biggest is that if the colors and intensities change rapidly as you move across the screen, the decoders have difficulty pulling the proper hue/intensity/value out of the signal. Not a problem for real life, but extremely common for computer generated images. Visually, it means that when you render a white square on a dark background the left and right edges of the square aren't quite the right color. For the Leg Lab, I made a filter which clamped down on such changes and rendered side-by-side with and without the filter. The difference was visible.
Does anyone here know if the makers of video cards and console games have started prefiltering their outputs in anticipation of how they're going to be transmitted and displayed?
(There is more you can do based on the fact that the human eye does a Fourier transform in time of what it sees - reverse the transform and you can make images which only make sense when played back at full speed and are hopeless blurs otherwise. That's how we managed to get the ostrich legs to look good. But I digress.)
Personally, I prefer ReplayTV.
When our remote died, they sent us a replacement for free - even though we were off warranty.
I'm more confident of ReplayTV's privacy policies than TiVo's.
Don't forget the cost of the service (realistically, you'll buy the lifetime version) when you do cost comparisons. The boxes lose a lot of functionality without the TV guide service.
If you really _do_ intend to hack the box, sure, get TiVo. But if you just want a box you can plug in and record things, I like Replay.
TiVo did (I don't know if it still does) send every single button push received from the remote back to the servers, time stamped.
I use a Replay. Right or wrong, I have more confidence that they'll keep my private information, well, private.
While I was a graduate student at MIT, I was using SD/Fast, a program which wrote C code for simulating system dynamics. The input and output formats of SD/Fast could have been more user friendly (I have no idea if they are now), but I had written software which made it relatively easy to create input files, even large ones, for SD/Fast and integrate the results into physically realistic simulations. You could describe the creature you wanted, and it would eventually write everything except the control system itself. Including making input to SD/Fast, running SD/Fast, and incorporating the results.
In one case, someone was (ab)using it to simulate a tree with 100+ degrees of freedom. The file generated by SD/Fast was multiple megabytes of undecipherable C source code. We couldn't even figure it out enough to break it into multiple files (and the process of creating it was thoroughly automated, we didn't really want to).
So we showed it to the Sun C compiler. It churned for about an hour... and ran out of some internal resource. I never found out what, nor did I find out how to resize it.
So we showed it to the IBM C compiler running on the machine across the room. Lo and behold, it compiled and ran, doing a good imitation of Zippy the Wonder Slug.
So we took the -g flag out. After all, its working, we don't need debugging information.
Fortunately I'd limited the core dump size. Otherwise, some of the users who shared that disk space would have been really annoyed when it segfaulted.
IBM wanted the program, so they could try and figure out why it was different with the debugging information included.... We advised them of the size of the source code and the number of libraries in the MIT AI lab it depended on, and they said "Uh... we'll send you this patch which might be it."
This patching process happened a few times. Eventually, the program worked and the poor undergraduate who was trying to do this finished his project.
As a closing note, soon after that SD/Fast sprouted options to break the output into multiple files.
You want me to watch a commercial? Then give me a commercial I want to watch. I promise, do that and I won't push the skip button on my Replay.
There have been a few commercials which have been entertaining enough that, after skipping over most of them, I went "Huh?" backed up and watched it.
Give me commercials worth watching and you won't have to resort to legalisms to get your revenue.
Many people here have listed "no distractions" as their #1 requirement for zoning.
I agree, with the following two caveats:
1. I need a minor distraction every 45-60 minutes to remind me to give my hands a break. Otherwise, I'll get an RSI relapse.
2. Sometimes, I end up thinking very hard about a problem - so hard that I fail to see the obvious "almost as good" alternative. I am stuck down in the weeds, when I should be looking for a solution in the bigger picture. I do have the bigger picture in my head but I am concentrating so hard on a small part of it that hard I do not realize I should step back.
My solution? CD player and headphones.
Every 45-60 minutes, you have to change the CD. Time for a hand break if you haven't taken one.
If I am concentrating on something too hard, eventually the music will distract me enough to make me reevaluate the situation. I find music, used this way, a gentle enough distraction that I slip back into coding extremely quickly.
Of course, listening to music doesn't address problems related to co-workers bothering you... other people have good ideas there.
I also find a well-chosen set of headphones filters out a lot of the ambient noise, even if you aren't listening to music.
My impression was that if you had Onstar, and your car was stolen, you could call them and they'd kill the engine remotely.
But I could be wrong, I don't have the service. Would someone who has details care to comment?
... and nex time, I will use "preview". Honest, the original message had carriage returns in it.
I agree that "today's computer interfaces are an ergonomic nightmare." However, to some extent they can be corrected now, without going through voice recognition. Change the keyboard to something designed for people, not something designed for a mechanical typewriter. I gave myself a fairly bad case of RSI while working on my Master's thesis. Since changing to a Datahand keyboard, my hands have been slowly getting better rather than worse. Other people I know prefer Kinesis. Even more offtopic: If you have RSI, the most important things I figured out are: 1. Pay attention to your hands. There's a feeling of discomfort before your hands start to hurt. As you get used to being in pain, you get used to ignoring this discomfort... bad idea. An extra hour of typing is not worth several days of relapse. 2. Sit properly. Move your keyboard, chair, monitor, and desk as necessary. If your upper arms are long compared to your torso, like mine, you may want your keyboard in your lap. 3. If your hands tense up when you start using a keyboard, it may be in anticipation of pain. If so, try switching to a differently arranged keyboard (such as a Kinesis or Datahand, for example) to remove that association. 4. Be aware that extended periods of pain can cause depression. RSIs take years to develop, and for me it has taken years to go away. 5. Go see a doctor. Get your RSI documented, so that later if it has gotten worse you have a visit on record saying it has been an ongoing condition. Read books on RSI, in case your doctor is not familiar enough with RSI to provide useful advice. 6. Pain is a warning that you are doing damage. Ibuprofin reduces swelling, which is good, but also keeps you from getting the feedback that you are damaging your hands. I don't take Ibuprofin unless I am not going to be typing for a while. 7. Every half hour/45 minutes, get up and walk around for a couple minutes. Personally, I listen to CDs as I work - when the CD ends, if I haven't taken a break I should. 8. Examine the other activities of your life. I discovered that I hold paperback books in a way which puts a lot of strain on my thumb and pinky, and changed that. 9. Aerobic exercise seems to help me, and is a good idea even if you don't have RSI. Caveat: I am not a medical doctor. I am simply stating the steps I took to go from stopping after a few hours because of pain, to consistently working 8+ hour days writing software. I mention Datahand and Kinesis keyboards because they are the ergonomic keyboards I see most often. My only relation to Datahand is that of an extremely satisfied customer. I don't have a relationship with Kinesis.