The earliest one I know of was in April 2000 on the 14-story science library at Brown University. Info & videos here: http://bastilleweb.techhouse.org/ I remember the press coverage at the time. Steve Wozniak flew in from CA to play.
Intelligence agencies have been doing this sort of thing for decades, giving slightly different versions of a sensitive document to suspected spies or places where possible spies might have access to it, with some subtle changes in the words, seeing which one gets leaked or appears elsewhere. Tom Clancy coined the term Canary trap for the technique. Patriot Games was published in 1987, but its real-world use for exposing information leaks most likely predates the novel.
But the classic Canary Trap requires someone to modify the document manually, which is hard to do on a large scale. Here it is being done automatically by an algorithm.
However, I am aware of published methods for this problem dating back to 2001 by Mikhail Atallah at Purdue. In fact Atallah received a patent for followup work in 2007, a year after the Amazon patent was filed.
Here are a few hundred papers on the subject, via Google Scholar. Some adjust whitespace, some modify images of the text, and some attempt fairly sophisticated syntactic analysis and restructuring of selected sentences.
I apologize that I haven't read the Amazon patent, or read the prior literature carefully, or gone to law school, so I can't comment on whether the patent seems valid or not.
That's true to some degree. But computers do slow down as they age. Components damaged by the constant heating cause more errors and therefore require retransmission or error correction, slowing things down.
My Dell desktop from 1999 has been running like the wind again since last week, when I reverted it to its 2002 state from backup tape. It goes superfast now that it's virus-free, off the network, and running old apps on Windows 98.
I was only trying to recover some old files before junking an unusable machine, but I may keep it around now as a non-networked machine for the kids.
The curriculum includes many technological courses (theoretical and applied crypto, network design, network protocols, red-teaming, etc.), but also some public policy courses. I'm guessing that their graduates will be prime candidates for these jobs.
I heartily second this. I've been using Thinkpads (T-series) for the past 10 years. The trackpoint is great -- you can navigate quickly and precisely while keeping your hands on the keyboard. No batteries, wires, or desk space needed.
It sits in the center of the keyboard, between the G, H, and B keys, where you can reach it with either index finger. That puts your thumb right over the mouse buttons that are under the spacebar.
They corrected some early kinks with resistance and calibration, and the trackpoints work about perfectly now.
It's true that for mouse-heavy activities, like drawing or editing graphics, it still feels a bit more fluid to use a real mouse or a tablet -- I have a lovely Graphire-4 tablet with a pressure-sensitive pen. But I never seem to use them, because almost everything I do needs the keyboard too, and the context-switch slows me down too much compared to the trackpoint.
"Lawyers for the accused delivered a brief statement at the opening of the trial, explaining that their client had be under a large amount of stress after being homebound for a year due to a snowboarding accident with nothing to do but watch television and play video games."
So, presumably he hadn't been playing the game elsewhere.
He played it at friends' houses, according to this account of the trial (which also uses the word "homebound" -??):
Daniel's lawyer, James Kersey, gave a short opening statement. He said his client had been under great stress at the time of the shooting because of a snowboarding accident that resulted in a severe staph infection. It left Daniel with such severe spinal damage that the slightest injury could leave him paralyzed.
Daniel was homebound for a year with nothing to do but watch television and play video games, Kersey said. It was during that time that he became fascinated with the Halo series and would play them for hours at friends' houses. His father forbade the games, saying that were too violent and sexually explicit.
... tax experts contend there's no such thing as a free spaceflight. Some contest sponsors provide a check to cover taxes, but that income is also taxable.
Iterate to convergence, folks.
Do tax experts also claim that this doesn't work? The only issue I can see is that to compute the total amount needed, you must know the rest of that person's adjusted gross income for the year (so you can compute their tax rate at each step of iteration).
I believe I've often read about corporations paying the taxes on perks they give to their senior executives. I always just assumed they iterated.
Followup: Confirmed in at least one scenario. I did a little poking around with Google. In the case of a life insurance perk, there is something called a Section 162 Double Bonus Plan. Here the company pays not only the premium and employee's tax on that benefit, but also the tax on the tax. Is this process really iterated further? Yes, judging by the $166,667 example in this article (since 1 + 0.4 + 0.4^2 + 0.4^3 +... = 1.66667):
What are the alternatives? The so-called 162 Double Bonus Plan is one. The company agrees to make an annual bonus of the after-tax amount necessary to pay the premium on an executive-owned life insurance policy. The company also makes a "double bonus" to cover the taxes the employee must pay on the bonus(es). Because the participant is paying taxes on the bonuses, the regulations otherwise applicable to traditional NQDC (and, in particular, the Act) are irrelevant. However, since the employer is expected to "gross up" the bonus to cover taxes, this concept is expensive and tax-inefficient. Assuming a 40% marginal tax rate, the gross cost to a Company is $166,667 to yield an after-tax bonus to the employee of $100,000.
... tax experts contend there's no such thing as a free spaceflight. Some contest sponsors provide a check to cover taxes, but that income is also taxable.
Iterate to convergence, folks.
Do tax experts also claim that this doesn't work? The only issue I can see is that to compute the total amount needed, you must know the rest of that person's adjusted gross income for the year (so you can compute their tax rate at each step of iteration).
I believe I've often read about corporations paying the taxes on perks they give to their senior executives. I always just assumed they iterated.
Why is there a story about grammar on a site whose editors can't understand the difference between "its" and "it's"?
The question mark [in the above] should be placed inside the quotation marks.
Nope, not in this example. Any professional writer would leave it outside. I think you're overgeneralizing -- American English does override the logical placement for commas and periods, but other punctuation marks like ? and ) are always left in their logical position, which may be either before or after the close-quote.
Keep in mind that Google's crawler, and presumably Yahoo!'s as well, tries to visit the more important pages first. So even if Google is missing half of the pages in Yahoo!, it probably still covers most of the ones you'd want to find.
(Indeed, covering fewer pages might be a wise strategy, other things equal, since it leaves more crawling time available to revisit known important pages that may have changed.)
As far as I know, prioritized crawling was first described in this 1998 paper by the Google folks. Note that the paper considers the original PageRank metric, which has since evolved. http://www-db.stanford.edu/pub/papers/efficient-cr awling.ps .
Thanks, bshanks, for the long and thoughtful reply. You're correct that the only thing that bothers me is the deception: conflicts of interest should always be disclosed. I like your vision of how open buzzing could be a good thing.
Your site is a good idea, though -- we may as well have a place to put a list of products which may be being astroturfed, either by this company or by others, as well as have an open forum to discuss these sort of things.
Exactly. Even if BzzAgent took the steps you suggested, other marketers would probably spring up who are willing to push the ethical boundaries. And the problem is not limited to product marketing: I've heard several instances of astroturfing political opinions, by advocacy groups or corporations (dare I speak the name of slashdot's Great Satan?). I think such tactics need to be tracked and exposed where possible, both as a partial cure and as a deterrent.
It's unpleasant to think of the innocents that will be caught in the cross-fire; for example, I'm a big fan of the Rock Bottom restaurant chain mentioned in the article and often evangelize it to friends. And I happened to move in July 2003. I hope I'm never mistaken for a "secret agent".
Well, suspicious friends can ask you point-blank whether you're buzzing. But Rock Bottom ought to recognize that deceptive marketing comes with a cost: if their strategy gets out, then genuine recommendations like yours do become suspect. Their choice.
The alternative is that recommendations for pretty much everything become suspect, and that wounds many more innocents.
I'm more worried about the possibility that products might be wrongly or maliciously identified as counterbuzz targets. That's one reason it's good to have an authoritative website with an appeals mechanism. Sort of a snopes.com for that urban legend going around about how Al Fresco sausages are tasty.:-)
If the word gets out that one member of a community is covertly foisting products on the rest, a general sentiment of deceit smites the social atmosphere. I feel that, although this is a perfectly legal, dare I say brilliant, marketing system, I would make it a point to rout out and publicly humiliate any Bzzers I discover.
I had the same thought. But this has to be done cooperatively to be effective. Why not infiltrate, find out which products are being buzzed, list them online, and let volunteers stamp out the astroturf they see? In other words, we need a site that plays the same role that Snopes does for urban legends.
I've just registered counterbuzz.com and created an overview page and logo. I'll also send a brief letter to the NY Times Magazine suggesting this strategy and giving the URL.
Who wants to take over development of the site? If the letter is selected for publication, it will appear in 2 weeks and send a lot of visitors to the site. We need at least a forum so volunteers can get organized.
The page says:
CounterBuzz will try to collect and verify information about which
products are being "buzzed." If someone is buzzing in a public forum
about a product listed here, you can post your own review that exposes
her and proves it with a link to CounterBuzz. If your friend is
buzzing you one-on-one, at least you can tell him how you feel about it.
The name CounterBuzz was chosen by analogy with counter-propaganda
and counter-spin. If volunteer shills can play this game, so can
volunteer debunkers.
What's wrong with buzzing? There are ethical arguments on both sides of this practice. But
the bottom line is that I deserve to know when I'm being spun.
If you claim to be my friend, or you're posting a review on Amazon,
you'd better tell me if you have a conflict of interest.
there's no good reason to put it in the filename. UNIX traditionally doesn't store this meta-data at all, and the user is left to just "know" what each file is.
Eh? The tradition in Unix is to put it in the filename (same as DOS and Windows). Your C compiler will behave differently on files with the extensions.c,.cpp,.h,.o,.s.,.i,.a. Compressing a file adds a.Z or.gz extension. Most makefiles dispatch by extension. Etc.
There are also many conventional extensions like.tar,.rpm,.ps,.pdf that are not necessarily treated specially by applications, but are taken by users to be informative.
It is true that Unix itself doesn't care what the extension is. So it's not the filesystem or other parts of the OS that care; it's specific applications or users. DOS was the same way. Windows is too, as far as I know, except that it adds a convenient little "launch" command that launches an application by file type. This is a good addition, and useful enough that I hacked up my own version in Unix.
The moon's orbit is everywhere concave towards the Sun. Therefore, the moon is a satellite of the Sun, and not a satellite of the earth. As such, perhaps it should be called a member of a binary planetary system.
Once profs have tenure their incentive to teach better is dramatically reduced. If they can get more grants doing research with no chance of being fired for imcompentent teaching then you can believe the grants will come first.
As a not-yet-tenured computer science prof, I think this is a misleading, even backward, account of how tenure affects teaching.
Do you really think that untenured profs are under pressure to teach well? At a research university, our tenure case depends much more on the ability to get grants and do influential research. Of course we untenured folk are expected to teach and do everything else well, too, but given that time is finite, we're usually advised to teach adequately without wasting too much time on it, and make sure the research is brilliant. So you ought to expect that it's untenured faculty who neglect teaching in favor of research.
You're correct that tenure may further reduce the incentive to teach well. But hey, it also reduces the incentive to get grants, do research, speak politely to one's colleagues, etc. So why do you conclude specifically that teaching will lose out to research once the pressure's off? Here are some guesses that are more logical but still misinformed:
"Tenured faculty have more freedom to allocate their time among these activities as they see fit, without the dean or tenure committee looking over their shoulder. Some may prefer research, others teaching."
This is sort of true, but there are many other pressures on how profs allocate their time.
Taking away the tenure issue doesn't suddenly free us up to do what we like. There are always a zillion things that have to be taken care of today.
"Tenured faculty have no incentive to do anything, and 'allocate their time' to sitting on their duffs. So they teach worse and do less research."
Fortunately, this cynical proposition doesn't at all match my experience of CS profs. Faculty who make it to tenure tend to continue working very hard. They generally have a strong work ethic and identify with their jobs. Usually they also have a sense of duty toward their undergrad students, grad students, and grant sponsors.
You may be thinking: Ok, tenured profs may keep working hard, but won't they spend all their time on research once they can get away with it? Don't professors just prefer research? Obviously profs at research universities do enjoy research -- after all, we emerged from a Ph.D. still liking it enough that we took a job where independent research was required. But we also enjoy teaching (or at least mentorship) -- otherwise we'd have gone off to do research in industry, at twice the salary and half the hassle. So don't assume we'd all just like to ditch the teaching in favor of research as soon as we can.
Now, it may well be that your original conclusion happens to be right, and tenured profs do focus more on research than untenured profs. But here are three possible explanations for such a correlation (if it exists):
Causal explanation (which you proposed in your post): "Their priorities changed as soon as they got job protection."
Doubtful (based on conversations with many tenured faculty). Tenure is unlikely to change my priorities at all. I do what I do because I like to do it and because other people (especially students) depend on me to do it. If you get tenure, nothing changes, and you're very relieved about that. You just go on doing what you were doing before.
Selection bias explanation: "The profs who spend all their time teaching don't get tenure. So the ones who are left after tenure were the ones who preferred research in the first place."
This explanation has some merit (since a few great, committed teachers do get thrown away like this), but it doesn't go too far. In my experience, most faculty who are denied tenure were trying to play the research game and are not noticeably better teachers than the
The great thing about Bayesian filtering is that it's adaptive. So they would have to dramatically increase the rate at which they discover and use filter-killing tricks for this to work.
Hmmm. Many anti-spammers seem to assume that advertisers will keep sending the same kind of spam, just superficially doctored by the spam-sender to evade the latest anti-spam heuristics.
But I worry that eventually, some companies that advertise via spam will learn to speak in a human voice. Surely this is possible for some products or scams. Advertisements don't have to look like advertisements, especially if they are only trying to pique your interest in a product that you will then go buy (or vote for) offline.
Even you will have to read the message carefully to realize that it's unsolicited bulk email. In such cases, we can't expect good accuracy from Bayesian filters, and the message will take more of your time.
Basically, advertisers adapt. A parallel example: If we get too good at zapping TV commercials with our TiVOs, they'll switch to more insidious product placement in the shows, so that the commercials are indistinguishable from the content.
Collaborative spam-filtering methods like Vipul's Razor might hold more promise. But the character of spam could shift to evade these filters, too. Spam might eventually come to resemble a bigger form of junk snailmail, or telemarketing -- where there are lots more advertisers but each one does a better job of targeting to a smaller list of customers (thanks to database companies like Experian).
By flying under the radar with smaller lists,
an advertiser might be able to stay out of the
database of known spams. (With a small list, few recipients may bother to report the spam, so you can't distinguish it from solicited bulk mail that has been accidentally or maliciously reported as spam by several people.)
In the long run, I think we have to solve spam
in the email architecture. I've always thought hashcash was the most promising idea, and it is now being pursued at Microsoft Research. There are also more radical proposals like Tripoli.
Slashdot Mirror
The earliest one I know of was in April 2000 on the 14-story science library at Brown University. Info & videos here:
http://bastilleweb.techhouse.org/
I remember the press coverage at the time. Steve Wozniak flew in from CA to play.
Intelligence agencies have been doing this sort of thing for decades, giving slightly different versions of a sensitive document to suspected spies or places where possible spies might have access to it, with some subtle changes in the words, seeing which one gets leaked or appears elsewhere. Tom Clancy coined the term Canary trap for the technique. Patriot Games was published in 1987, but its real-world use for exposing information leaks most likely predates the novel.
But the classic Canary Trap requires someone to modify the document manually, which is hard to do on a large scale. Here it is being done automatically by an algorithm.
However, I am aware of published methods for this problem dating back to 2001 by Mikhail Atallah at Purdue. In fact Atallah received a patent for followup work in 2007, a year after the Amazon patent was filed.
Here are a few hundred papers on the subject, via Google Scholar. Some adjust whitespace, some modify images of the text, and some attempt fairly sophisticated syntactic analysis and restructuring of selected sentences.
I apologize that I haven't read the Amazon patent, or read the prior literature carefully, or gone to law school, so I can't comment on whether the patent seems valid or not.
My Dell desktop from 1999 has been running like the wind again since last week, when I reverted it to its 2002 state from backup tape. It goes superfast now that it's virus-free, off the network, and running old apps on Windows 98.
I was only trying to recover some old files before junking an unusable machine, but I may keep it around now as a non-networked machine for the kids.
Johns Hopkins University, near Washington, DC, offers a master's degree in Security Informatics. This is through their Information Security Institute, which was founded several years ago and includes several well-known CS faculty.
The curriculum includes many technological courses (theoretical and applied crypto, network design, network protocols, red-teaming, etc.), but also some public policy courses. I'm guessing that their graduates will be prime candidates for these jobs.
Of course, major in CS first.
I use a keyboard with an IBM trackpoint so i don't keep moving my right hand between keyboard and mouse.
It takes a little to get used to it, but it worth the try!
http://www.pc.ibm.com/ww/healthycomputing/trkpnt.html
I heartily second this. I've been using Thinkpads (T-series) for the past 10 years. The trackpoint is great -- you can navigate quickly and precisely while keeping your hands on the keyboard. No batteries, wires, or desk space needed.
It sits in the center of the keyboard, between the G, H, and B keys, where you can reach it with either index finger. That puts your thumb right over the mouse buttons that are under the spacebar.
They corrected some early kinks with resistance and calibration, and the trackpoints work about perfectly now.
It's true that for mouse-heavy activities, like drawing or editing graphics, it still feels a bit more fluid to use a real mouse or a tablet -- I have a lovely Graphire-4 tablet with a pressure-sensitive pen. But I never seem to use them, because almost everything I do needs the keyboard too, and the context-switch slows me down too much compared to the trackpoint.
"Lawyers for the accused delivered a brief statement at the opening of the trial, explaining that their client had be under a large amount of stress after being homebound for a year due to a snowboarding accident with nothing to do but watch television and play video games."
So, presumably he hadn't been playing the game elsewhere.
He played it at friends' houses, according to this account of the trial (which also uses the word "homebound" -??):
Do tax experts also claim that this doesn't work? The only issue I can see is that to compute the total amount needed, you must know the rest of that person's adjusted gross income for the year (so you can compute their tax rate at each step of iteration).
I believe I've often read about corporations paying the taxes on perks they give to their senior executives. I always just assumed they iterated.
Nope, not in this example. Any professional writer would leave it outside. I think you're overgeneralizing -- American English does override the logical placement for commas and periods, but other punctuation marks like ? and ) are always left in their logical position, which may be either before or after the close-quote.
Source if you need one: http://webster.commnet.edu/grammar/marks/quotation .htm
Keep in mind that Google's crawler, and presumably Yahoo!'s as well, tries to visit the more important pages first. So even if Google is missing half of the pages in Yahoo!, it probably still covers most of the ones you'd want to find.
r awling.ps .
(Indeed, covering fewer pages might be a wise strategy, other things equal, since it leaves more crawling time available to revisit known important pages that may have changed.)
As far as I know, prioritized crawling was first described in this 1998 paper by the Google folks. Note that the paper considers the original PageRank metric, which has since evolved. http://www-db.stanford.edu/pub/papers/efficient-c
Thanks, bshanks, for the long and thoughtful reply. You're correct that the only thing that bothers me is the deception: conflicts of interest should always be disclosed. I like your vision of how open buzzing could be a good thing.
Exactly. Even if BzzAgent took the steps you suggested, other marketers would probably spring up who are willing to push the ethical boundaries. And the problem is not limited to product marketing: I've heard several instances of astroturfing political opinions, by advocacy groups or corporations (dare I speak the name of slashdot's Great Satan?). I think such tactics need to be tracked and exposed where possible, both as a partial cure and as a deterrent.
Well, suspicious friends can ask you point-blank whether you're buzzing. But Rock Bottom ought to recognize that deceptive marketing comes with a cost: if their strategy gets out, then genuine recommendations like yours do become suspect. Their choice.
The alternative is that recommendations for pretty much everything become suspect, and that wounds many more innocents.
I'm more worried about the possibility that products might be wrongly or maliciously identified as counterbuzz targets. That's one reason it's good to have an authoritative website with an appeals mechanism. Sort of a snopes.com for that urban legend going around about how Al Fresco sausages are tasty. :-)
I had the same thought. But this has to be done cooperatively to be effective. Why not infiltrate, find out which products are being buzzed, list them online, and let volunteers stamp out the astroturf they see? In other words, we need a site that plays the same role that Snopes does for urban legends.
I've just registered counterbuzz.com and created an overview page and logo. I'll also send a brief letter to the NY Times Magazine suggesting this strategy and giving the URL.
Who wants to take over development of the site? If the letter is selected for publication, it will appear in 2 weeks and send a lot of visitors to the site. We need at least a forum so volunteers can get organized.
The page says:
See the site for more discussion.
There are also many conventional extensions like .tar, .rpm, .ps, .pdf that are not necessarily treated specially by applications, but are taken by users to be informative.
It is true that Unix itself doesn't care what the extension is. So it's not the filesystem or other parts of the OS that care; it's specific applications or users. DOS was the same way. Windows is too, as far as I know, except that it adds a convenient little "launch" command that launches an application by file type. This is a good addition, and useful enough that I hacked up my own version in Unix.
Here are pictures and discussion of the moon's orbit about the sun.
As a not-yet-tenured computer science prof, I think this is a misleading, even backward, account of how tenure affects teaching.
Do you really think that untenured profs are under pressure to teach well? At a research university, our tenure case depends much more on the ability to get grants and do influential research. Of course we untenured folk are expected to teach and do everything else well, too, but given that time is finite, we're usually advised to teach adequately without wasting too much time on it, and make sure the research is brilliant. So you ought to expect that it's untenured faculty who neglect teaching in favor of research.
You're correct that tenure may further reduce the incentive to teach well. But hey, it also reduces the incentive to get grants, do research, speak politely to one's colleagues, etc. So why do you conclude specifically that teaching will lose out to research once the pressure's off? Here are some guesses that are more logical but still misinformed:
This is sort of true, but there are many other pressures on how profs allocate their time. Taking away the tenure issue doesn't suddenly free us up to do what we like. There are always a zillion things that have to be taken care of today.
Fortunately, this cynical proposition doesn't at all match my experience of CS profs. Faculty who make it to tenure tend to continue working very hard. They generally have a strong work ethic and identify with their jobs. Usually they also have a sense of duty toward their undergrad students, grad students, and grant sponsors.
You may be thinking: Ok, tenured profs may keep working hard, but won't they spend all their time on research once they can get away with it? Don't professors just prefer research? Obviously profs at research universities do enjoy research -- after all, we emerged from a Ph.D. still liking it enough that we took a job where independent research was required. But we also enjoy teaching (or at least mentorship) -- otherwise we'd have gone off to do research in industry, at twice the salary and half the hassle. So don't assume we'd all just like to ditch the teaching in favor of research as soon as we can.
Now, it may well be that your original conclusion happens to be right, and tenured profs do focus more on research than untenured profs. But here are three possible explanations for such a correlation (if it exists):
Doubtful (based on conversations with many tenured faculty). Tenure is unlikely to change my priorities at all. I do what I do because I like to do it and because other people (especially students) depend on me to do it. If you get tenure, nothing changes, and you're very relieved about that. You just go on doing what you were doing before.
This explanation has some merit (since a few great, committed teachers do get thrown away like this), but it doesn't go too far. In my experience, most faculty who are denied tenure were trying to play the research game and are not noticeably better teachers than the
But I worry that eventually, some companies that advertise via spam will learn to speak in a human voice. Surely this is possible for some products or scams. Advertisements don't have to look like advertisements, especially if they are only trying to pique your interest in a product that you will then go buy (or vote for) offline.
Even you will have to read the message carefully to realize that it's unsolicited bulk email. In such cases, we can't expect good accuracy from Bayesian filters, and the message will take more of your time.
Basically, advertisers adapt. A parallel example: If we get too good at zapping TV commercials with our TiVOs, they'll switch to more insidious product placement in the shows, so that the commercials are indistinguishable from the content.
Collaborative spam-filtering methods like Vipul's Razor might hold more promise. But the character of spam could shift to evade these filters, too. Spam might eventually come to resemble a bigger form of junk snailmail, or telemarketing -- where there are lots more advertisers but each one does a better job of targeting to a smaller list of customers (thanks to database companies like Experian). By flying under the radar with smaller lists, an advertiser might be able to stay out of the database of known spams. (With a small list, few recipients may bother to report the spam, so you can't distinguish it from solicited bulk mail that has been accidentally or maliciously reported as spam by several people.)
In the long run, I think we have to solve spam in the email architecture. I've always thought hashcash was the most promising idea, and it is now being pursued at Microsoft Research. There are also more radical proposals like Tripoli.