Other than the occasional time-wasting, EST-inspired motivational training snow-jobs, we rarely brought anybody in-house to do any real teaching. Most of that went on at conferences or among ourselves.
The problem, of course, is that few actual voters are likely to subscribe. I don't think this would really end up "leveling the playing field" (which, BTW, I'm all for).
Its a unix security model -- just don't put the user accounts into groups that might be able to do anything unpleasant or unwanted. This is pretty standard stuff, no?
I got my first domain (waste.com, since sold) well before there were any associated charges with even having a domain at all. When NSOL took over and began to charge, they didn't bother notifying me -- I only found out about the charges while listening to CNET radio, which had one of NSOLs reps on complaining about unpaid domains, and threatening to delist domains whose accounts were in arrears. Sheesh.
So from this analysis, if making spam illegal is a desirable goal - and it seems to be from the cheers here whenever charges are pressed against spammers - then I think it's difficult to simultaneously rationalize and argument against companies' attempts to control linkage to their sites.
I don't know that the issue is simply control of linkage. I think it's the method employed that concerns most people -- legal threats. As I'm sure will be mentioned ad-nauseum here, there are a variety of technical (and probably lest costly) means to prevent unauthorized linking.
Have you verified that there are no locations _within_ the buildings (attics perhaps, close to a window) that would be suitable for antenna mounting?
If not, could you set up two wireless access nodes _outside_ the building (again in some discreet location), and then run cable from each into the respective buildings?
I've never had an overwhelming urge to overclock a CPU, so I'm obviously pretty clueless, but wouldn't a peltier junction cooler be easier?
I realize that the hot side of a peltier device puts out quite a bit of heat which has to be gotten rid of -- is this why people end up resorting to water cooling?
If understand his algorithm correctly, it should be possible to defeat his filtering methods by appending a list of low-value words to the end of a spam message.
While he advocates generating probablility tables from an individual user's corpus of messages, I would imagine that most users will have many low-spam-probablity words in common.
Even easier, since he assigns a low score to unknown words, appending a sequence of random sets of letters to the end of the message would have much the same effect.
Checking for phrases (rather than words) can mitigate this a bit, but all in all, this still looks like a stopgap measure.
If I phone someone in Nigeria and suggest a money-laundering fraud then it is obvious to all that I am breaking the law in two countries, not in 'phonespace'. Nobody has ever suggested that the content of the telephone network -all those voice calls -should be somehow privileged and treated as outside the normal world.
No one AFAIK has suggested (credidibly) that the same offence is not punishible if conducted over the internet. Where is the problem, exactly?
in the UK we're perfectly happy to prosecute someone for war crimes committed fifty years ago in another country, so why are there problems if the crime involved the Internet? Under English law a sex tourist can be prosecuted here even if he has sex with a child in Thailand: surely prosecuting someone for promoting racial hatred on a US-hosted website can't be that different?
I must have missed a news story or two -- but if a British citizen violates a crime using US servers, are British authorities powerless to prosecute? Really? If so -- and I doubt this is the case -- shouldn't such obviously idiotic juristictional issues be fixed legislatively?
The result is that cyberspace appears somehow to be divorced from the physical world - but this is just an artifact of our current technologies and not a fundamental principle.
It would only appear so to a complete moron, or perhaps a child. Who, exactly, is the target readership of this article? Reminds me a little of a segment of 20/20...
No, the EPO does not allow such blatant exploitation by twisted business. No... you have to be much more sneaky, describing your algorithm as a 'machine' that just happens to be implemented as software.
I worked for a US patent law office (way back in the 80s) and this used to be the policy in the United States. I'd be interesting in knowing what precipitated the change. Unless memory fails me, it would have been inconceivable for the USPTO to grant a patent for pure software, let alone a business model.
Ironically, I'm actually listed as an inventor on a pure software patent (sh*t, I know -- company legal manuever). Does make a nice resume decoration, tho.
IMO, email has been ruined not by SPAMmers, but by overzealous SPAM blockers and self-appointed SPAM wardens. These people like to think they are striking a blow for privacy, yet all they are really doing is making email unreliable
I get 70 emails a day that say otherwise. The problem IS caused by spammers. Spam filters have largely eliminated the problem for me (I use SpamAssasin). However, the false positives remain problematic. My solution has just been to use the filters to mark incoming email, which is then marked read and filed elsewhere; I can then whitelist the stuff I really needed.
Still means I have to peruse the spam periodically, but this only takes about 30 seconds. Obviously this doesn't solve the bandwidth issue, but this isn't an issue for me personally -- but then my mailserver doesn't serve many people.
If business are losing vital email, then the issue lies with a poorly managed IS department, not the maintainers of the BH lists. Ultimately, though, it's unsolicited email that is the root of the entire problem.
My point is merely that the problem need not be a software problem, but could be a mechanical or electrical shortcoming (see below for why I harp on this). Even laser-augmented sonar can perform imperfectly in the presence of obstacles that are themselves moving, coming unexpectedly into the robots "field of view". While you are correct that a designer should understand the mechanical limitations of a robot prior to writing its software, failure to do so is not a "software" problem. It is instead a much larger (and IMHO more serious) failure to properly design the machine as a whole -- but mistakes are to be expected in an experiment. (Not to mimimize your saftey concerns -- but Grace did apparently stop, presumably via bump sensor.)
As an aside, noise on the power supply can certainly be a problem on a battery-powered robot. EMI, poorly laid-out boards replete with ground loops, and noise from motors sharing a common power supply with the electronics, can all cause problems, though I seriously doubt this was an issue with Grace (at least I hope not!).
I only bitch about this, because I keep running across people that seem to approach robotics as purely a software problem (not that you're necessarily one of these people); but in a practical sense, it really isn't. It is instead a synthesis of (at the very least) three disciplines.
Bumping into a judge indicates that the programmers in charge of GRACE failed basic obstacle avoidance...
If you do a lot of work robotics, you should be aware that failure to detect an obstacle seldom boils down to a failure of the software -- at least not as simple a failure as your example. Instead, such failures are often a function of the mechanical or electrical limitations of sensors, such as sonar failures on obstacles placed at a shallow angle. Unforseen sensor dead spots, slow sensor response or recovery time, even noisy supply lines are all major contributors to this kind of thing.
Slashdot Mirror
Seriously. Nobody is a bigger pain than a developer (I say this as a developer). If you do, count on considerable support costs up front.
But I don't work there any more.
My NEC V-25 is running just fine thanx. None of that heat-sunk glycol-cooled Pentium crap for me.
Maybe not ... but at least it was warm.
Will she break the 5-vote mark?
Great idea. I'm sure b4DD455 h4x0r will make a great president.
The problem, of course, is that few actual voters are likely to subscribe. I don't think this would really end up "leveling the playing field" (which, BTW, I'm all for).
... that the construction site is actually still up.
Its a unix security model -- just don't put the user accounts into groups that might be able to do anything unpleasant or unwanted. This is pretty standard stuff, no?
Thanx for giving me the benefit of the doubt. I love you, man.
So students get 100% off?
I got my first domain (waste.com, since sold) well before there were any associated charges with even having a domain at all. When NSOL took over and began to charge, they didn't bother notifying me -- I only found out about the charges while listening to CNET radio, which had one of NSOLs reps on complaining about unpaid domains, and threatening to delist domains whose accounts were in arrears. Sheesh.
I don't know that the issue is simply control of linkage. I think it's the method employed that concerns most people -- legal threats. As I'm sure will be mentioned ad-nauseum here, there are a variety of technical (and probably lest costly) means to prevent unauthorized linking.
If not, could you set up two wireless access nodes _outside_ the building (again in some discreet location), and then run cable from each into the respective buildings?
I realize that the hot side of a peltier device puts out quite a bit of heat which has to be gotten rid of -- is this why people end up resorting to water cooling?
...surely they'd be even more impressed with this . I mean, geez -- is the mowbot thing really news?
While he advocates generating probablility tables from an individual user's corpus of messages, I would imagine that most users will have many low-spam-probablity words in common.
Even easier, since he assigns a low score to unknown words, appending a sequence of random sets of letters to the end of the message would have much the same effect.
Checking for phrases (rather than words) can mitigate this a bit, but all in all, this still looks like a stopgap measure.
No one AFAIK has suggested (credidibly) that the same offence is not punishible if conducted over the internet. Where is the problem, exactly?
in the UK we're perfectly happy to prosecute someone for war crimes committed fifty years ago in another country, so why are there problems if the crime involved the Internet? Under English law a sex tourist can be prosecuted here even if he has sex with a child in Thailand: surely prosecuting someone for promoting racial hatred on a US-hosted website can't be that different?
I must have missed a news story or two -- but if a British citizen violates a crime using US servers, are British authorities powerless to prosecute? Really? If so -- and I doubt this is the case -- shouldn't such obviously idiotic juristictional issues be fixed legislatively?
The result is that cyberspace appears somehow to be divorced from the physical world - but this is just an artifact of our current technologies and not a fundamental principle.
It would only appear so to a complete moron, or perhaps a child. Who, exactly, is the target readership of this article? Reminds me a little of a segment of 20/20...
I worked for a US patent law office (way back in the 80s) and this used to be the policy in the United States. I'd be interesting in knowing what precipitated the change. Unless memory fails me, it would have been inconceivable for the USPTO to grant a patent for pure software, let alone a business model.
Ironically, I'm actually listed as an inventor on a pure software patent (sh*t, I know -- company legal manuever). Does make a nice resume decoration, tho.
I can think of lots of uses for severed heads:
Oh, but they'll want you, all right.
For spare parts.
I get 70 emails a day that say otherwise. The problem IS caused by spammers. Spam filters have largely eliminated the problem for me (I use SpamAssasin). However, the false positives remain problematic. My solution has just been to use the filters to mark incoming email, which is then marked read and filed elsewhere; I can then whitelist the stuff I really needed.
Still means I have to peruse the spam periodically, but this only takes about 30 seconds. Obviously this doesn't solve the bandwidth issue, but this isn't an issue for me personally -- but then my mailserver doesn't serve many people.
If business are losing vital email, then the issue lies with a poorly managed IS department, not the maintainers of the BH lists. Ultimately, though, it's unsolicited email that is the root of the entire problem.
Obviously lots of other areas of study are frequently drawn on, particularly subfields of biology (ethology, entomolgy, etc.)
Cheers -- m
As an aside, noise on the power supply can certainly be a problem on a battery-powered robot. EMI, poorly laid-out boards replete with ground loops, and noise from motors sharing a common power supply with the electronics, can all cause problems, though I seriously doubt this was an issue with Grace (at least I hope not!).
I only bitch about this, because I keep running across people that seem to approach robotics as purely a software problem (not that you're necessarily one of these people); but in a practical sense, it really isn't. It is instead a synthesis of (at the very least) three disciplines.
Ok -- I'm done ranting. Cheers.
If you do a lot of work robotics, you should be aware that failure to detect an obstacle seldom boils down to a failure of the software -- at least not as simple a failure as your example. Instead, such failures are often a function of the mechanical or electrical limitations of sensors, such as sonar failures on obstacles placed at a shallow angle. Unforseen sensor dead spots, slow sensor response or recovery time, even noisy supply lines are all major contributors to this kind of thing.