> And... do the robots run Linux?
As a matter of fact, in the robotics world the use of linux is quite large. I don't have any exact numbers but a guestimate would be around 50% linux based and the rest split between windows and embedded operating systems like VxWorks - depending a bit on how you define a "robot" and how you define "linux". Many robots used for the applications mentioned above come from eg. irobot .
Oh, and yes, I am a roboticist.
One further question to ask, given that many investigators in copyright infringement cases uses specialized programs (I vaguely remember a slashdot story on this long time ago), is: How do we know what the programs realy are doing, that there is no bug in the accidentally giving the wrong IP address or possible exploit (by RIAA or anyone else on the P2P network) which gives the wrong IP address. Without seeing the sourcecode for the programs we cannot know that their workins are correct, and even if we see the sourcecode (made available to a large team of experts or OpenSource) how can we rely on it if there are any security problems in it (eg. possible buffer overflows etc.). I think this line of arguments are very important since it asks the very question of if we can trust machine generated evidence without complete transparancy in the process.
The important distinguision to make when comparing the benefits of going massivly paralel processing is that it is possible to solve NEW problems in realtime with these processors. Eg, we don't need to run Word 100 times faster, however we can get eg. games and scientific simulations (two sides to the same coin) that uses detailed physics engines and realtime raytracing. Raytracing can be almost naivly paralellised with up to as many processors as screen pixels. I remember using a computer with 65536 processors called the maspar which was built in the early 90's. Our main use for this computer was for image processing which also could easily be parallelized. It just took a bit of a shift of perspective to learn how to program it since it was SIMD (Single Instruction Multiple Data) but boy where it fast for it's time.
Physics is a bit more difficult but there are tehniques too for paralellization utilizing the fact that object interactions form islands of connected parts. Eg, when simulating your hair in a realistic way don't test for interactions with the objects in a distant part of the game. Physics engine are just starting to become used for these purposes but can easily require how much CPU power you want for it. Simulating eg. the clothes in the game characters or dynamic subdivision of parts as they break or bend due to forces (do you want realistic dents in your car after hitting that pedestrian?). These would both require an order of magnitude more CPU power than what we can do in realtime physics today.
So, to make a short summary. Yes, we can always achieve new tricks with even more computing power. Give me a cluster of a million processors and i would still complain that it's too slow for what i want to do.
The biggest problem with this proposal is that is costs a *lot* of money to make patents and even more money to take patents violators to court. Sure, some of this money *might* be raised by donations but it kind of removes the whole free-as-in-beer advantage of opensource (leaving the speech option for all us die hard OSS people...).
It's a nice suggestion and all but i have a hard time seeing it work. Perhaps in a few limited occassions like for some kernel stuff where the patents could be used for cross licencing with some big companies.
Nonetheless, the disadvantage would be that the reputation and credibility of the non-software-patent and OSS communities would be damaged.
It does not sound like you have had very much hands on experience with sonars within robotics. The current state of commercially available sonars (in the air) provides a single range measurement for a cone (usually ca. 30 degrees, but sometimes much smaller). This range measurement is *very* unreliable since it only gives the distance to one point (usually the closest) within this cone and only under the rights circumstances (depending on the material, the angle towards sonar etc.). The biggest problem with these sensors is the low angular resolution and unreliability. Nonetheless the state of the art in, for instance, map building manages to construct some surprisingly accurate maps, to navigate in indoor environments etc.
To say that the data isn't _used_ well enough is not an accurate description of the problem. The problem rather lies in the sensors and in the signal processing (computing eg. range data).
Oh, and also: if you are interested in robotics make sure to take a look at Player/Stage http://playerstage.sourceforge.net/ which is an opensource implementation of drivers for various commercial robots as well as some controll functionalities, simulators etc.
I believe Google's standpoint is that there is an implicit permission to copy the webpage since it is published on the internet (everone viewing the webpage makes atleast one copy of it when viewing it and possibly one for their cache). Even though not a foolproof argument I think Google has enough money / lawyers to make an interesting case of it.
Quite true, EU is not a democracy just as little as it is a country. Talking about EU as a homogenous state (yes, like USA!) is just plain wrong. EU is just a fancy collaboration between states just like the trade agreements between Mexico, USA and Canada, or
perhaps like the UN, NATO etc.
So why don't we talk about the african union not beeing a democracy? Or the WTO, or NATO, or UN or....
Actually there's ongoing research by the Malaysian goverment to do just this. I had a nice reference to a playboy magazine detailing this (my boss gave it to me as background material - I swear!) but seems to have lost it. You can probably find it if you search a bit for it.
Actually, I'm a non-us citizen, diabetic with an insulin pump and apprantly looks very suspcicous since I'm always picked out for controll (once even three times when transfer in seattle), but nontheless I've *never* had any troubles with security and my insulin pump. I guess it must depend on how you present it to them.
Before 9/11 the scanners normally never even picked up the insulin pump. After 9/11 they do pick it up but all the security personel I've met have understood what it is and respected that it can't be removed.
Funny, maybe I do have good karma after all =)
Well said, however one minor comment:
> - The patent covers something not entirely obvious to an experienced programmer (the "five minute test": given the problem, could an
experienced problem come up with a solution in less than five minutes?).
I would like to argue rather a "one-year" test. Considering that the total cost for getting a patent approved in many cases are equiveaent to a few years salary for a professional in the field (think good programmer, univeristy PhD etc) I would argue that the limit for the depth if invention should be that it would take a professional at least one year to come up with a solution to the problem.
/ M
Forgive if I'm ignorant... but what does the oil price have to do with spaceship{1,2,3}? I thought their fuel was based on hydrogenperoxide and even *if* oil is involved in the current manufacturing process it's not neccessarily so in the future.
This comment is especially striking considering that the real effect of the great wall was not realy to keep the raiding parties out (you would have needed an awfull lot of guards for that) but rather to make it harder for them to retreat after a succefull strike - thus removing the incentative for raiding parties. So, this truly could be considered a modern version of the great wall - making it harder for people to escape
Yes,using this method is a *little* bit secure. However, don't be tempted to believe it's secure unless you can guarantee that the computer you are sticking it into (insert joke about safe sex here) is guaranteed "safe".
Admittedly, since the system runs on a virtual machine loaded from the memory stick it's difficult for a process on the host machine to access it's data but it's not impossible. Unless you can actually boot completly (no windows bootstrapping) from the usb stick and know that there's nothing evil in the bios it's just no saef enough for me.
/ Mathias
So in essense this seem to support the Sappir-Worph hypothesis (http://venus.va.com.au/suggestion/sapir.html) that the language strongly affect our ability to think.
This makes one wonder if a another language would give us the ability to better reason about other things. Would we be smarter if we had a better language in which to think?
There is an artifical language called lojban (http://www.lojban.org/) based on predicate logic but which is meant to be used as other "real" languages (compare with eg. esperanto, interlingua and swahili). The question is, would native speakers of lojban be better a rational thought? As far as I know there are no native speakers of lojban but what would happend if I raised my (hypothethical) children to speak if from birth?
Mathias
hmm.. do you know for a fact that you use NAT more in US than europe? I would guess it's the opposite way around since getting enough (IPv4) ip-numbers is a real problem for many ISP's in europe simply because you americans hogged them all in the begining... I had to wait for a whole year until I got an IP-number =(
uhm.. something's not right. You said back in the Win 3.1 days and that you had just built a great Athlon 1500+ system... I though M$ stoped supporting win 3.1 *long* before the athlons came to be?
yup, that's why I wanted to rewrite the cgi scripts - shouldn't be too hard to allow sequential access, access by specific ID as well as a "tree"-based access. Eg. when viewing an id of length n there's n "next" buttons. The first skips to the next algorithm, the second to the algorithm 256 steps in front, the k'th skips 256^(k-1) algos. Then it's plausible that a random user will encounter any algorithm of a certain length =)
Well, pretty long considering it's down at the moment (the CGI scripts broke with another http setup).... the langugage was similar with brain-fuck with the added feature that all string are syntactically correct programs. So that'd make 256^22=2^176 algos.
I'm thinking of rewritting it with a bit more "serious" lock. Including a better language and shortcuts to specific algorithms. Consider eg. the scenario: someone comes along with a patent, I write an implementation of it, computes the id number XX (eg. 23BC-EF25-544B) of the implementation and says "well, I published that back in 2002, just look under algogorithm ID: XX".
As an example. This is the algorithm with ID 2321-3112-3481-2352-8123-7183-2161-8123-1
<define name="ciruei"> <lambda var0="baruas"> <if> baruas <mult> baruas <call> ciruei <dec> baruas </dec> </call> </mult> 1 </if> </lambda> </define> Bonus karma to the one who first guesses what the algo does... (just kidding) Please note the importance of inventing a *new* language here so that we don't accidentaly break any existing copyrights =( A shorter and more "serious" representation of the id will also be reimplemented... if I find the time to do it
As a patent examiner I assume that you can reasonably well defend the legislative system you are representing and would therefore like to hear your oppinions on the following reasoning:
After talking to a patent layer briefly I was told that publications on the net (websites) count as prior art in patent applications. Regardless if I could actually prove that anyone had read the website or not. Shortly after this I made a (dynamically generated) website containing *all* algorithms which could be expressed with program smaller than 4 MB in a turing equivalent language of my invention. They way this worked was that enumerated all such programs and after viewing the algorithm you just had to click the "next" button to get to the next algorithm until you have seen all algorithms (no doubt a *very* time consuming activity). Formally I had thus given a description by example of in principle every humanly concievable algorithm. Thus, do we have a case of prior art for all yet unpatented algorithm?
Even if you dont think the above reasoning would hold in court it would be nice to hear any thoughts about since it realy highlights that there's a fundamental difference between an algorithms (which, just as mathematics, can be *discovered* - not invented) and "normal inventions".
Please note that most of the limitations in the above reasoning can be removed with just a little bit more of effort (currently the website is down but I'll put a new version up if there's any interest in it). The underlying problem here is that the set of all algorithms is a quite small enumerable set as compared to the set of all inventions which is realy a set of untangible
ideas, not mathematically enumerable.
Well, the label only says it doesn't work on PC's and Mac's. There's nothing on the label about not working on for instance a Sparc station. Of course I'm bright enough to realize that it probably doesn't work on my 'puter at work but the intended audience (Celine Dion listeners) might not be.
Can this treaty realy be compared to the DMCA. As I see it the worst part about the DMCA - restricting the freedom of speech by outlawing the *construction* of circumvention devices (read programs) - is not present in this treaty. The clost they get to this seem to be Article 18 and 19 which I interpred solely as forbing the *use* of said devices/programs. Which of course is bad that too... but not as bad as the DMCA.
Has anyone thought about Wine and the Digital Millenium Copyright Act? Consider for instance the scenario that Windows Media Player has built in "technology" to prevent screen dumps from being taken during playback and that Wine is mature enough to run Windows Media Player. Then Wine could be though of as a circumvention device (!) since I can take an X-windows screenshot while viewing copyrighted movies through Windows Media Player and Wine....
I'm sure there's *lots* of possible scenarios in which Wine can be described as a circumvention device - especially if we let MS do the talking. I wouldn't dare live in the US if I where a Wine developer. Hell, I might not dare live in the US anyway!
Slashdot Mirror
> And... do the robots run Linux?
As a matter of fact, in the robotics world the use of linux is quite large. I don't have any exact numbers but a guestimate would be around 50% linux based and the rest split between windows and embedded operating systems like VxWorks - depending a bit on how you define a "robot" and how you define "linux". Many robots used for the applications mentioned above come from eg. irobot .
Oh, and yes, I am a roboticist.
One further question to ask, given that many investigators in copyright infringement cases uses specialized programs (I vaguely remember a slashdot story on this long time ago), is: How do we know what the programs realy are doing, that there is no bug in the accidentally giving the wrong IP address or possible exploit (by RIAA or anyone else on the P2P network) which gives the wrong IP address. Without seeing the sourcecode for the programs we cannot know that their workins are correct, and even if we see the sourcecode (made available to a large team of experts or OpenSource) how can we rely on it if there are any security problems in it (eg. possible buffer overflows etc.). I think this line of arguments are very important since it asks the very question of if we can trust machine generated evidence without complete transparancy in the process.
The important distinguision to make when comparing the benefits of going massivly paralel processing is that it is possible to solve NEW problems in realtime with these processors. Eg, we don't need to run Word 100 times faster, however we can get eg. games and scientific simulations (two sides to the same coin) that uses detailed physics engines and realtime raytracing. Raytracing can be almost naivly paralellised with up to as many processors as screen pixels. I remember using a computer with 65536 processors called the maspar which was built in the early 90's. Our main use for this computer was for image processing which also could easily be parallelized. It just took a bit of a shift of perspective to learn how to program it since it was SIMD (Single Instruction Multiple Data) but boy where it fast for it's time.
Physics is a bit more difficult but there are tehniques too for paralellization utilizing the fact that object interactions form islands of connected parts. Eg, when simulating your hair in a realistic way don't test for interactions with the objects in a distant part of the game. Physics engine are just starting to become used for these purposes but can easily require how much CPU power you want for it. Simulating eg. the clothes in the game characters or dynamic subdivision of parts as they break or bend due to forces (do you want realistic dents in your car after hitting that pedestrian?). These would both require an order of magnitude more CPU power than what we can do in realtime physics today.
So, to make a short summary. Yes, we can always achieve new tricks with even more computing power. Give me a cluster of a million processors and i would still complain that it's too slow for what i want to do.
The biggest problem with this proposal is that is costs a *lot* of money to make patents and even more money to take patents violators to court. Sure, some of this money *might* be raised by donations but it kind of removes the whole free-as-in-beer advantage of opensource (leaving the speech option for all us die hard OSS people...).
It's a nice suggestion and all but i have a hard time seeing it work. Perhaps in a few limited occassions like for some kernel stuff where the patents could be used for cross licencing with some big companies.
Nonetheless, the disadvantage would be that the reputation and credibility of the non-software-patent and OSS communities would be damaged.
It does not sound like you have had very much hands on experience with sonars within robotics. The current state of commercially available sonars (in the air) provides a single range measurement for a cone (usually ca. 30 degrees, but sometimes much smaller). This range measurement is *very* unreliable since it only gives the distance to one point (usually the closest) within this cone and only under the rights circumstances (depending on the material, the angle towards sonar etc.). The biggest problem with these sensors is the low angular resolution and unreliability. Nonetheless the state of the art in, for instance, map building manages to construct some surprisingly accurate maps, to navigate in indoor environments etc. To say that the data isn't _used_ well enough is not an accurate description of the problem. The problem rather lies in the sensors and in the signal processing (computing eg. range data). Oh, and also: if you are interested in robotics make sure to take a look at Player/Stage http://playerstage.sourceforge.net/ which is an opensource implementation of drivers for various commercial robots as well as some controll functionalities, simulators etc.
I believe Google's standpoint is that there is an implicit permission to copy the webpage since it is published on the internet (everone viewing the webpage makes atleast one copy of it when viewing it and possibly one for their cache). Even though not a foolproof argument I think Google has enough money / lawyers to make an interesting case of it.
So why don't we talk about the african union not beeing a democracy? Or the WTO, or NATO, or UN or ....
Actually there's ongoing research by the Malaysian goverment to do just this. I had a nice reference to a playboy magazine detailing this (my boss gave it to me as background material - I swear!) but seems to have lost it. You can probably find it if you search a bit for it.
Actually, I'm a non-us citizen, diabetic with an insulin pump and apprantly looks very suspcicous since I'm always picked out for controll (once even three times when transfer in seattle), but nontheless I've *never* had any troubles with security and my insulin pump. I guess it must depend on how you present it to them.
Before 9/11 the scanners normally never even picked up the insulin pump. After 9/11 they do pick it up but all the security personel I've met have understood what it is and respected that it can't be removed.
Funny, maybe I do have good karma after all =)
Well said, however one minor comment:
> - The patent covers something not entirely obvious to an experienced programmer (the "five minute test": given the problem, could an experienced problem come up with a solution in less than five minutes?).
I would like to argue rather a "one-year" test. Considering that the total cost for getting a patent approved in many cases are equiveaent to a few years salary for a professional in the field (think good programmer, univeristy PhD etc) I would argue that the limit for the depth if invention should be that it would take a professional at least one year to come up with a solution to the problem.
/ M
Forgive if I'm ignorant... but what does the oil price have to do with spaceship{1,2,3}? I thought their fuel was based on hydrogenperoxide and even *if* oil is involved in the current manufacturing process it's not neccessarily so in the future.
This comment is especially striking considering that the real effect of the great wall was not realy to keep the raiding parties out (you would have needed an awfull lot of guards for that) but rather to make it harder for them to retreat after a succefull strike - thus removing the incentative for raiding parties. So, this truly could be considered a modern version of the great wall - making it harder for people to escape
Admittedly, since the system runs on a virtual machine loaded from the memory stick it's difficult for a process on the host machine to access it's data but it's not impossible. Unless you can actually boot completly (no windows bootstrapping) from the usb stick and know that there's nothing evil in the bios it's just no saef enough for me. / Mathias
So in essense this seem to support the Sappir-Worph hypothesis (http://venus.va.com.au/suggestion/sapir.html) that the language strongly affect our ability to think.
This makes one wonder if a another language would give us the ability to better reason about other things. Would we be smarter if we had a better language in which to think?
There is an artifical language called lojban (http://www.lojban.org/) based on predicate logic but which is meant to be used as other "real" languages (compare with eg. esperanto, interlingua and swahili). The question is, would native speakers of lojban be better a rational thought? As far as I know there are no native speakers of lojban but what would happend if I raised my (hypothethical) children to speak if from birth?
Mathias
> Or sited a hell of a long way from any
> population centers...
Well, considering the risk for a Chernobyl I would think that the bush administration actually would support having the reactor in France?
hmm.. do you know for a fact that you use NAT more in US than europe? I would guess it's the opposite way around since getting enough (IPv4) ip-numbers is a real problem for many ISP's in europe simply because you americans hogged them all in the begining... I had to wait for a whole year until I got an IP-number =(
uhm.. something's not right. You said back in the Win 3.1 days and that you had just built a great Athlon 1500+ system... I though M$ stoped supporting win 3.1 *long* before the athlons came to be?
yeah, if you want the *peace* nobel price! For the real one you should contact the nobel commitie in Sweden...
yup, that's why I wanted to rewrite the cgi scripts - shouldn't be too hard to allow sequential access, access by specific ID as well as a "tree"-based access. Eg. when viewing an id of length n there's n "next" buttons. The first skips to the next algorithm, the second to the algorithm 256 steps in front, the k'th skips 256^(k-1) algos. Then it's plausible that a random user will encounter any algorithm of a certain length =)
I'm thinking of rewritting it with a bit more "serious" lock. Including a better language and shortcuts to specific algorithms. Consider eg. the scenario: someone comes along with a patent, I write an implementation of it, computes the id number XX (eg. 23BC-EF25-544B) of the implementation and says "well, I published that back in 2002, just look under algogorithm ID: XX".
As an example. This is the algorithm with ID 2321-3112-3481-2352-8123-7183-2161-8123-1
<define name="ciruei">
<lambda var0="baruas">
<if>
baruas
<mult>
baruas
<call>
ciruei
<dec>
baruas
</dec>
</call>
</mult>
1
</if>
</lambda>
</define>
Bonus karma to the one who first guesses what the algo does... (just kidding) Please note the importance of inventing a *new* language here so that we don't accidentaly break any existing copyrights =( A shorter and more "serious" representation of the id will also be reimplemented... if I find the time to do it
/ M
As a patent examiner I assume that you can reasonably well defend the legislative system you are representing and would therefore like to hear your oppinions on the following reasoning: After talking to a patent layer briefly I was told that publications on the net (websites) count as prior art in patent applications. Regardless if I could actually prove that anyone had read the website or not. Shortly after this I made a (dynamically generated) website containing *all* algorithms which could be expressed with program smaller than 4 MB in a turing equivalent language of my invention. They way this worked was that enumerated all such programs and after viewing the algorithm you just had to click the "next" button to get to the next algorithm until you have seen all algorithms (no doubt a *very* time consuming activity). Formally I had thus given a description by example of in principle every humanly concievable algorithm. Thus, do we have a case of prior art for all yet unpatented algorithm? Even if you dont think the above reasoning would hold in court it would be nice to hear any thoughts about since it realy highlights that there's a fundamental difference between an algorithms (which, just as mathematics, can be *discovered* - not invented) and "normal inventions". Please note that most of the limitations in the above reasoning can be removed with just a little bit more of effort (currently the website is down but I'll put a new version up if there's any interest in it). The underlying problem here is that the set of all algorithms is a quite small enumerable set as compared to the set of all inventions which is realy a set of untangible ideas, not mathematically enumerable.
Well, the label only says it doesn't work on PC's and Mac's. There's nothing on the label about not working on for instance a Sparc station. Of course I'm bright enough to realize that it probably doesn't work on my 'puter at work but the intended audience (Celine Dion listeners) might not be.
Can this treaty realy be compared to the DMCA. As I see it the worst part about the DMCA - restricting the freedom of speech by outlawing the *construction* of circumvention devices (read programs) - is not present in this treaty. The clost they get to this seem to be Article 18 and 19 which I interpred solely as forbing the *use* of said devices/programs. Which of course is bad that too... but not as bad as the DMCA.
I'm sure there's *lots* of possible scenarios in which Wine can be described as a circumvention device - especially if we let MS do the talking. I wouldn't dare live in the US if I where a Wine developer. Hell, I might not dare live in the US anyway!