My children had no idea how long a "yard" was until I described the length with my hands...
It's quite simple. A football field is 120 yards long, including the endzone. The offensive team has 4 downs to go 10 yards. It's simple when you put it in terms they know.
Of course, I suppose in Canada, the rules are a bit different -- there's that 1 yard neutral zone thing.
Alright now, lets first realize that there are TWO aspects of spam filters that must be evaluated, and thus a single metric cannot fully encapsulate information about those aspects. The measures to be concearned with are false positive ratio, and false negative ratio.
A false positive occurs when email that is not spam is marked as such. These are often much more costly, spam may be automatically discarded, or otherwise effective lost to an enduser (Think about what the signal to noise ratio is in your spam inbox (S/N).
A false negative occurs when spam is not marked as such.
Any evaluation of an anti-spam technology should include these two measures. For example; in my current configuration, SpamAssassin has a false negative rate of about 25%, so it is approximately 75% correct in identifying spam. The false positive rate is 1/1898 (and the one legitimate email that it classified as spam was pretty worthless), which may be viewed as a sucess rate of 100% in identifying non-spam as such...
As a side note, I could factor in the total volume of mail in the corpus that I'm considering (~18000 messages), and use that to weight and combine the two numbers, and say that my spam filter is in effect 97.37% accurate, but as we've just established, that doesn't tell the real story.
Well, there's skill in how you deal with the randomness, and knowing probability will help you out in the long run. No poker player will argue that its bad to be able to quickly calculate whether pot odds are in your favor if you stay in a hand. Sure, being a math whiz is not going to guarantee that you take town Johnny Chan in a heads up no-limit game, but among a big table playing a structured game, a knowledgable player's expected value will be positive.
Perhaps the hundreds of thousands of dollars of equipment available for the taking in the the AI Lab and LCS have something to do with it, no?
You're exactly right. AI/LCS previously inhabited a building in tech square (NE-43), which had locks to get onto each floor, and into every office. (Granted, many people had copies of these keys, but there was still some degree of security). All the computers and lab equipment were locked inside of offices or research space. In the stata center, there are 1000 faculty, staff and gradstudents. ~350 of them have offices, and the rest are in "open plan" seating. Open plan, a scheme designed by Gehry (the architect), was meant to inspire colaboration. In practice, it means that there are many desks sitting in the middle of wide open spaces, most equiped with unsecured flat panel displays. Factor in personal belongings and lab equipment that tends to gather around a workspace, and there's a lot of money and research sitting behind these RFID locked doors.
Last time I checked, linux doesn't fall under any of these: Operating Systems: Microsoft Windows 98SE, Windows ME, Windows 2000, Windows XP.
The Linux player is for embedded systems - similar those alluded to in the interview. This
press release specifically mentions Set-top boxes and other OEM solutions. Once again, these are not products that a consumer can buy and run on their home linux system.
Because as many posters have pointed out, the requirements for a secure ATM and a secure e-voting machine are completely different.
An ATM verifies identity, authorizes an action, and keeps an audit of actions and identity.
An e-voting machine must confirm an identity, reliably record a vote, and make sure that the entity casting that vote can not vote again. It must then assure that the vote remains intact, and is included in the final tally, WITHOUT associating it with the casting entity.
Instead of being so worried about it, why not simply close the loop with the voter to make fraud detection easy?
Because there are several conflicting goals with e-voting systems, and this is against one of them. An e-vote should be anonymous, and user verifiable initially (as in your proposed system). However, it should not be possible for the user to be able to decrypt their vote after it's cast. To do so allows one to reliably buy votes, which is considered an undesirable feature.
If you think the outside is ugly, you should see the inside. There are exposed, rough finish concrete support beams cutting through the middle rooms (those gray circles on the
floor plan), and all the exposed wood on cabinets and shelves is low grade plywood. Other walls are bereft of shelves (Many researchers here keep their own copies of journals), because it was cheaper to leave the studs out of non-structural walls.
I'm not even getting into the lack of straight lines and right angles (very disorienting). But other than that, and the fact that it's years behind schedule and massively over budget, it's great.
First off, anyone who has read his article realizes Bray isn't entirely qualified to understand what he's covering.
In this case, he's covering the innovation of someone who is fairly smart.
For a more indepth look at these systems, check out:
http://theory.lcs.mit.edu/~rivest/MicaliRivest-Mic ropaymentsRevisited.ps Rivest's paper on micropayments.
As for other points raised here: The idea behind many small transactions being lumped together into a larger one makes this a feasable system to use when the cost per transaction (to credit the merchant) matters.
Consumers will be shielded from the statistics - the idea is their bank / the micropayment provider makes up the differences, absorbing the swings themselves (but having them dampered by the large sample size).
If you have mathmatical questions as to this scheme, read the paper - it's very complete (if a bit dense)
Umm, $8.75 is nowhere near competitive with what local companies pay - the cost of living in and around Boston is very high. What the media lab offer(ed||s) students is:
A chance to do actual research. Even if a student starts off as a code monkey, they will often use it as a springboard toward a thesis.
Flexibility. Working part time at the local convenience store won't let you set your own hours
A dot com environment with an academic pace. Sure, there are some late nights, but there's less danger of having a bad quarter and losing funding for your project.
The encryption itself isn't quite a problem - you can use faster secret key algorithms for the actual transfer (randomly generate secret key and encrypt that using public key)
The difficult part is more how you do distributed file sharing well. The Napster model still provides a central point to DoS / Sue, and gnapster doesn't scale... The freenet project (
http://freenet.sourceforge.net/
)might interest you, as it's trying to do distributed sharing with privacy / anonymity.
the thing i wonder though, is couldn't you still sniff the packet just like you would a password, except rather than "mypassword6" you get "23857723984...653765846358" or something?
Well, that's the point of SSH, you can't sniff a password.
You can spend a lot of time studying secure interaction, but in a nutshell, here's how it works with (I'm not intimantely familiar with SSH, so I'm making some assumptions):
You have your account on localhost and remotehost (To use the same terminology as the article). REMOTEHOST generates a public key, which is widely published. You get a copy of this key on localhost (SSH does this automagically, but not technically securely the first time you connect).
When you SSH from localhost, it asks for remotehosts public key. SSH compares it to the cached, verified (hopefully this has been done in some degree) key.
Information is then encrypted before being sent over the link. (Most protocols use public key methods for negotiation, and then a faster algorithm such as RC4 for data transfer).
Your password is never sent in the clear, and therefore can't be sniffed. The only way to obtain it is to compromise one of the hosts.
tell me, are the dead tree news outlets saying the same
Um, last time I checked, the New York Times (linked to from the main article) and Boston globe (one of these links you're criticizing) are real newspapers- They just also happen to have a copy of the content online.
Students may be able to cheat on exams.
Upkeep may be hellish
Many of MIT's courses are already online in some form or another. Many of these classes distribute old tests (
http://6004.lcs.mit.edu/ is one example that has quizzes from the last 3 terms - other home pages are linked from the course catalog.
MIT can lose students since they could go to other universities and still learn at their level
MIT is more than just course materials. It is being surrounded by some of the smartest people in a given field. Where else can you take a class on accoustics taught by Bose? Or take a required, introductory biology class taught by one of the principle investigators in the human genome project. Or on a smaller level, email a professor with a question at 2 am, and get a response in a few minutes.
I'm in a similar situation as you (College student), but have had no problems finding unix internships. There are a LOT of companies out there looking for those skills.
However, you can't expect them to find you. Go on hotjobs, or monster (or insert job site of choice). Post your resume. But SEARCH for jobs you'd be interested in. Then submit your resume anyway. Mention that you're looking for an internship in your cover letter, and ask if they'd be interested. Some will turn you down. But a surprising number wont.
Too bad they never did the same for Winamp -- it would force AOL to release the modified source to Winamp (assuming the GPL is upheld)
No it wouldn't - by purchasing Nullsoft AOL became the owner, and could re-issue the program under a different license. (Now if Nullsoft had released the program under GPL, and assigned the rights to FSF....)
Personally, the most annoying sound from computers is the high pitched whine that monitors sometime make. If you can hear those frequencies, it is much more annoying than the hum of cooling fans, or eavn a noisy drive.
FWIW, my solution at work to noise (no monitors with bad flybacks, thank $deity) are the Sony MDR-V600s
-JB
Problems with link? Try This one...
on
@Home UDP Lifted
·
· Score: 1
I had some problems following the link from the story... This one seems to work however...
IMHO MIT Tech Review is always a bit optimistic with time frames. Breakthrough's are always portrayed as just around the corner. I think that this view comes from working close to researchers on a regular basis - they tend to have a different mindset than the general populace.
So one of the reasons given for waiting so long between movies was for technology to catch up to the director's vision. Yet now that there is a technology capable of delivering his grand works to the home populace, he ignores it. hrm.
The Miami Herald article has a Lockheed slant to it, i.e. "No-one told us about that canyon". However, the official site says that they can't spot any evidence of the lander's parachute. Both parties want to save face by blaming the others....
Remember, The truth is out there (specifically, it's sitting on the surface of Mars somewhere).
Any language can be written incomprehensibly (C even has the obfuscated C contest to prove it), PERL just makes it easier. However, having worked in a company that made extensive use of PERL, I can say that maintence of that code was just as difficult (no more, no less) than maintenence of other code.
Software maintence all comes down to design standards and code review. In a good system there are agreed upon coding standards that all developers are familiar with, and code gets a peer revier before going into production. This review stage tends to lead to confusing bits of code getting commented heavily. As always, YMMV
Slashdot Mirror
My children had no idea how long a "yard" was until I described the length with my hands...
It's quite simple. A football field is 120 yards long, including the endzone. The offensive team has 4 downs to go 10 yards. It's simple when you put it in terms they know.
Of course, I suppose in Canada, the rules are a bit different -- there's that 1 yard neutral zone thing.
Alright now, lets first realize that there are TWO aspects of spam filters that must be evaluated, and thus a single metric cannot fully encapsulate information about those aspects. The measures to be concearned with are false positive ratio, and false negative ratio.
A false positive occurs when email that is not spam is marked as such. These are often much more costly, spam may be automatically discarded, or otherwise effective lost to an enduser (Think about what the signal to noise ratio is in your spam inbox (S/N).
A false negative occurs when spam is not marked as such.
Any evaluation of an anti-spam technology should include these two measures. For example; in my current configuration, SpamAssassin has a false negative rate of about 25%, so it is approximately 75% correct in identifying spam. The false positive rate is 1/1898 (and the one legitimate email that it classified as spam was pretty worthless), which may be viewed as a sucess rate of 100% in identifying non-spam as such...
As a side note, I could factor in the total volume of mail in the corpus that I'm considering (~18000 messages), and use that to weight and combine the two numbers, and say that my spam filter is in effect 97.37% accurate, but as we've just established, that doesn't tell the real story.
Well, there's skill in how you deal with the randomness, and knowing probability will help you out in the long run. No poker player will argue that its bad to be able to quickly calculate whether pot odds are in your favor if you stay in a hand. Sure, being a math whiz is not going to guarantee that you take town Johnny Chan in a heads up no-limit game, but among a big table playing a structured game, a knowledgable player's expected value will be positive.
Perhaps the hundreds of thousands of dollars of equipment available for the taking in the the AI Lab and LCS have something to do with it, no?
You're exactly right. AI/LCS previously inhabited a building in tech square (NE-43), which had locks to get onto each floor, and into every office. (Granted, many people had copies of these keys, but there was still some degree of security). All the computers and lab equipment were locked inside of offices or research space. In the stata center, there are 1000 faculty, staff and gradstudents. ~350 of them have offices, and the rest are in "open plan" seating. Open plan, a scheme designed by Gehry (the architect), was meant to inspire colaboration. In practice, it means that there are many desks sitting in the middle of wide open spaces, most equiped with unsecured flat panel displays. Factor in personal belongings and lab equipment that tends to gather around a workspace, and there's a lot of money and research sitting behind these RFID locked doors.
Did you dig beyond the press release and look at the system requirements for this product?
Last time I checked, linux doesn't fall under any of these: Operating Systems: Microsoft Windows 98SE, Windows ME, Windows 2000, Windows XP.
The Linux player is for embedded systems - similar those alluded to in the interview. This press release specifically mentions Set-top boxes and other OEM solutions. Once again, these are not products that a consumer can buy and run on their home linux system.
This Tech interview was published on April 16th, 2004, and took place on April 15th, 2004.
The Turbo Linux news was announced (via the press release) on April 27th, 2004.
Thus, at the time of the interview, the 'good zealot rant' was correct about there not being a licensed player for linux.
Man I hate it when a good troll is spoiled by facts.
Because as many posters have pointed out, the requirements for a secure ATM and a secure e-voting machine are completely different.
An ATM verifies identity, authorizes an action, and keeps an audit of actions and identity.
An e-voting machine must confirm an identity, reliably record a vote, and make sure that the entity casting that vote can not vote again. It must then assure that the vote remains intact, and is included in the final tally, WITHOUT associating it with the casting entity.
Instead of being so worried about it, why not simply close the loop with the voter to make fraud detection easy?
Because there are several conflicting goals with e-voting systems, and this is against one of them. An e-vote should be anonymous, and user verifiable initially (as in your proposed system). However, it should not be possible for the user to be able to decrypt their vote after it's cast. To do so allows one to reliably buy votes, which is considered an undesirable feature.
If you think the outside is ugly, you should see the inside. There are exposed, rough finish concrete support beams cutting through the middle rooms (those gray circles on the floor plan), and all the exposed wood on cabinets and shelves is low grade plywood. Other walls are bereft of shelves (Many researchers here keep their own copies of journals), because it was cheaper to leave the studs out of non-structural walls.
I'm not even getting into the lack of straight lines and right angles (very disorienting). But other than that, and the fact that it's years behind schedule and massively over budget, it's great.
-JWB
In this case, he's covering the innovation of someone who is fairly smart.
For a more indepth look at these systems, check out: http://theory.lcs.mit.edu/~rivest/MicaliRivest-Mi
As for other points raised here: The idea behind many small transactions being lumped together into a larger one makes this a feasable system to use when the cost per transaction (to credit the merchant) matters.
Consumers will be shielded from the statistics - the idea is their bank / the micropayment provider makes up the differences, absorbing the swings themselves (but having them dampered by the large sample size).
If you have mathmatical questions as to this scheme, read the paper - it's very complete (if a bit dense)
The difficult part is more how you do distributed file sharing well. The Napster model still provides a central point to DoS / Sue, and gnapster doesn't scale... The freenet project ( http://freenet.sourceforge.net/ )might interest you, as it's trying to do distributed sharing with privacy / anonymity.
Well, that's the point of SSH, you can't sniff a password.
You can spend a lot of time studying secure interaction, but in a nutshell, here's how it works with (I'm not intimantely familiar with SSH, so I'm making some assumptions):
You have your account on localhost and remotehost (To use the same terminology as the article).
REMOTEHOST generates a public key, which is widely published. You get a copy of this key on localhost (SSH does this automagically, but not technically securely the first time you connect).
When you SSH from localhost, it asks for remotehosts public key. SSH compares it to the cached, verified (hopefully this has been done in some degree) key.
Information is then encrypted before being sent over the link. (Most protocols use public key methods for negotiation, and then a faster algorithm such as RC4 for data transfer).
Your password is never sent in the clear, and therefore can't be sniffed. The only way to obtain it is to compromise one of the hosts.
Um, last time I checked, the New York Times (linked to from the main article) and Boston globe (one of these links you're criticizing) are real newspapers- They just also happen to have a copy of the content online.
Students may be able to cheat on exams.
Upkeep may be hellish
Many of MIT's courses are already online in some form or another. Many of these classes distribute old tests ( http://6004.lcs.mit.edu/ is one example that has quizzes from the last 3 terms - other home pages are linked from the course catalog.
MIT can lose students since they could go to other universities and still learn at their level
MIT is more than just course materials. It is being surrounded by some of the smartest people in a given field. Where else can you take a class on accoustics taught by Bose? Or take a required, introductory biology class taught by one of the principle investigators in the human genome project. Or on a smaller level, email a professor with a question at 2 am, and get a response in a few minutes.
I'm in a similar situation as you (College student), but have had no problems finding unix internships. There are a LOT of companies out there looking for those skills.
However, you can't expect them to find you. Go on hotjobs, or monster (or insert job site of choice). Post your resume. But SEARCH for jobs you'd be interested in. Then submit your resume anyway. Mention that you're looking for an internship in your cover letter, and ask if they'd be interested. Some will turn you down. But a surprising number wont.
No it wouldn't - by purchasing Nullsoft AOL became the owner, and could re-issue the program under a different license. (Now if Nullsoft had released the program under GPL, and assigned the rights to FSF....)
Personally, the most annoying sound from computers is the high pitched whine that monitors sometime make. If you can hear those frequencies, it is much more annoying than the hum of cooling fans, or eavn a noisy drive.
FWIW, my solution at work to noise (no monitors with bad flybacks, thank $deity) are the Sony MDR-V600s
-JB
http://news .cnet.com/news/0-1004-200-1526165.html?tag=st.cn.1 .lthdne
-JB
Some would argue that the discussion is about dorms being wired, not dorm life in general.
Others would argue that high speed networking and dorm life are so intertwined that they are the same thing.
Yet others would say that my first point is overly pedantic.
And still more would say that complaining about moderation in a thread about dorms is also offtopic
Now if only creative's drivers were SMP friendly...
However the opensound stuff is very nice, though
propriatary and closed.
-JB
IMHO MIT Tech Review is always a bit optimistic with time frames. Breakthrough's are always portrayed as just around the corner. I think that this view comes from working close to researchers on a regular basis - they tend to have a different mindset than the general populace.
-JB
So one of the reasons given for waiting so long between movies was for technology to catch up to the director's vision. Yet now that there is a technology capable of delivering his grand works to the home populace, he ignores it. hrm.
-JB
The Miami Herald article has a Lockheed slant to it, i.e. "No-one told us about that canyon". However, the official site says that they can't spot any evidence of the lander's parachute. Both parties want to save face by blaming the others....
Remember, The truth is out there (specifically, it's sitting on the surface of Mars somewhere).
-JB
Any language can be written incomprehensibly (C even has the obfuscated C contest to prove it), PERL just makes it easier. However, having worked in a company that made extensive use of PERL, I can say that maintence of that code was just as difficult (no more, no less) than maintenence of other code.
Software maintence all comes down to design standards and code review. In a good system there are agreed upon coding standards that all developers are familiar with, and code gets a peer revier before going into production. This review stage tends to lead to confusing bits of code getting commented heavily. As always, YMMV