I have wondered for some time why nobody is implementing an "intelligent" traffic-light-timing system. Think Google for traffic light timing.
Where I live, all of the major traffic lights have cameras attached (to automate sending of traffic tickets to red-light runners). It seems to me that these cameras could be used to analyze the routes that are being taken by vehicles and intelligently optimize the light timing based on this.
For example, let's assume that tracking vehicles by license plate number in this manner is too much an invasion of privacy, but we will allow the system to track a vehicle's route by color and size. Let's also assume that it's 3am and there are very few cars on the road.
Say a given small red vehicle leaves a given intersection at 3:05am. This system would be able to estimate how soon that vehicle should arrive at the next adjoining intersection, and adjust the timing of that light so that it's green when the vehicle arrives (other traffic notwithstanding).
Of course there's some chance that the vehicle will not actually reach that second light (e.g. it pulls into a driveway). So for each intersection, the system would actually calculate the probability of a vehicle arriving from each direction in real time.
Now let's say this vehicle's driver gets off work every weekday at 3:00-3:10am and almost always takes the same route home. Over several days, the system should be able to recognize this trend and optimize the lights such that, as soon as the vehicle is recognized on the road, this driver gets all green lights all the way home, most of the time. Thus the system could recognize not only the likelihood of a vehicle arriving at a given intersection at a specific moment, but also the likelihood of a "small red vehicle" taking a left at a given intersection, both in realtime and historically.
So that's the "one vehicle, no congestion" conceptualization. Now just extend this to include all vehicles on the road. Such a system should then be able to optimize 'global' traffic flows, based on this simple underlying probability model.
Of course, special optimizations for heavy congestion, accidents, weather, etc. could be layered on top.
I think the fundamental question, "why can't all the lights change to green for me, as I arrive at them, when there is no other traffic to consider?" is adequately answered by this proposed system. It should also be able to optimize traffic flows better than any human traffic engineer during the most congested times of day, relying especially on historical trends of all the vehicle routes it has recognized.
Since such a system could utilize existing traffic cams and pressure plates, I think it would also be (relatively) inexpensive to implement such a system.
I think you should be able to accomplish this in 5 minutes by creating a local HTML page incorporating FRAMESET/FRAME tags, along with a separate 'subpage' to be loaded into each frame containing a pseudo address bar (i.e. an INPUT TYPE=TEXT element, a Go button and a line of javascript).
Correct. However, it could be quite useful for the popular stuff. Some of the episodic torrents out there can have a following of thousands of peers, subscribed via RSS. This tech is a nice match for that.
I think it would be most useful if this tech was integrated into the mainline BT protocol or some of the more popular BT clients. I just don't see many folks switching to a dedicated client for BT based VOD/videostreams; YouTube is just too easy. But if this was integrated into common BT clients, "Swarmplayer mode" could be activated simply by attempting to preview a video file in the BT client.
This is just a variation of the regular captcha using pictures instead of letters. It has the same vulnerabilities. If used undistorted it is merely a matter of building a dictionary of pictures, if distorted it has the same strengths and weaknesses of the same distortion applied to letters How about "describe this scene"?
Visual scenes involving objects could be dynamically 3d-rendered to defeat "image dictionary" attack strategies.
For example, "the [cat] is [under] the [car]". The three bracketed terms could be replaced with a large set of nouns or verbs/prepositions.
This scene description could then be rendered from a number of different camera positions/angles; colors changed; and background or extraneous/obscuring foreground objects added.
Until computer intelligence reaches the level of human intelligence for interpreting visual information, there are still CAPTCHA methods available that can't be reliably broken using automated (non human) methods.
I watched the GMail storage space counter for 1 minute, noting the starting and ending numbers. Then I typed this into Google:
(3113.335029 - 3113.246785) MB/minute * 1 year
= 45.3240458 gigabytes
If the counter really keeps moving at this pace, those prices for extra storage are going to seem pretty steep.
It should obviate the need for a 'delete attachments' button or sorting by size, though.
Re:So if the this is completely free of charge....
on
1-800-Google Launches
·
· Score: 2, Informative
Possible motivations for Google:
1. Further promote Google's value and brand in the minds of consumers and business owners alike. This is following in the same vein as the Google Local Business Center's completely free coupons that you can create for your business to appear on Google Maps. Call it good PR.
2. Statistics. Useful not only to them, but to local business owners. Again, develops brand loyalty and is good PR.
3. As part and parcel of 1 and 2, they hope that some of these local business owners will later purchase normal ads on Google.
In short, even if the added revenue from 3 doesn't completely offset the cost of running this service, 1 and 2 add value in ways that are hard to directly quantify in monetary terms. Though I suspect they will indeed make more (regular Adwords/Adsense) revenue as a direct result of this service's availability than it costs to run it. The usual exorbitant $1.25 per 411 call is not even close to what it costs to run such a service (especially without human operators).
4. Don't be evil. And while you're at it, if you can take out a few of the other evildoers out there, all the better! Google has taken the lessons of "disruptive technology" to heart, and it's worked for them time and again.
As usual, Google wins by not being overly (or at least overtly) greedy. It's for this reason that I doubt we'll ever hear paid ads or purchased placement when calling the service.
Diebold is the leading manufacturer of ATM machines. Those machines are nigh-impossible to break into with physical access (magstripe hacks & minimart ATMs notwithstanding). Diebold has some serious security-minded people on staff, and they know what they're doing.
Which leads to the only rational conclusion: that Diebold used its extensive security knowledge to ensure that as many avenues of security penetration are available *as possible*. That so many security holes do exist is a matter of record, and it is plain to see from these reports that it would take someone with a deep understanding of security issues to enable such a variety of attack vectors.
I mean, polymorphable executable code can reside on the flash cards meant to hold the vote count totals. That functionality had to be specifically designed and implemented. This was no mistake.
"If Blanchard is right, then clearly, as average family size decreases, so will the incidence of male homosexuality."
It certainly makes sense as a "naturally selected" evolutionary outcome. As a geographical region became overpopulated with humans, those families and communities which produced fewer offspring would be less likely to overpopulate or overwork the land. Those communities whose mothers and daughters produced too many offspring would be less likely to survive into the next generation. So, mothers who were more likely to produce homosexual offspring (at least sons) after her firstborn would present an evolutionary advantage, and her genetics would be thereby "selected" in future generations.
Similarly it would be an evolutionary advantage, as a mother, to still produce able-bodied sons to work for the community, but not necessarily to have all those sons procreating all the time.
Phil Harrison:.. If you're a complete music fan and video fan, and you want to have huge amounts of digital content, then you can upgrade to whatever size of drive you like. You can put any in that you like - it is a computer, after all.
gamesindustry.biz: So that hard drive is a standard PC drive?
Phil Harrison: ATA, bog standard, yeah.
gamesindustry.biz: You're not going to be selling Sony drive upgrades?
Phil Harrison: We've got no plan to. We may offer something, but we have no plan to at the moment.
Table names (and perhaps even column names) have to be in ALL CAPS. Sorry can't name a table PersonContact, you must use PERSON_CONTACT.
Guess it's just my aesthetic pedanticness but I do not like ALL_CAPS except in my C constants. It also makes any kind of automated conversion from another DB that much more annoying.
A bad design decision that's never really been fixed. I believe it is possible to get UpperLowerCase table names in Postgres, but you have to jump through so many error-prone hoops, it's not worth the trouble.
This is my point, though -- with uTorrent, you don't need to add RAM, or demote the process priority, or for that matter, require a JRE to be loaded. In other words, if there is a far superior (IMHO) alternative tool for the job, what reason is there to carry on using the inferior tool?
If you need those features which Azureus has that uTorrent doesn't provide (and honestly there aren't many), continuing to use Azureus would make sense. At least for me though, the probable remaining reason of "being used to Azureus", or not having to make the effort of using a different product, is not compelling.
If anything, that "if it ain't broke" attitude of is part of what keeps me motivated to find the best tool I can for the job -- and be ready to switch when the best tool is replaced by something markedly better. Otherwise, I may be missing out on an opportunity. (ever try to get someone to switch from IE to a tabbed browser?)
Maybe you can elucidate for me why, if two pieces of software provide equivalent functionality, but one runs in a CPU/RAM/disk footprint a fraction the size of the other -- why you would then keep using the greedier software.
The main reasons I've heard in the past are:
- "It Doesn't Run On Non-Windows." Duh, and this discussion is not even pertinent to non Windows users.
- "It Is Not Open Source". OK, but it is Free, so this is really just the ol' argument of Open Source vs. Demonstrable Performance. And the argument that Azureus "could" become better with user contributions is a moot point - if it DOES become better then use Azureus but until then, it actually IS NOT better.
- "My Grandma Is Already Used to Azureus". Well, I think it's healthy to be able to adapt and make the best of what is available. Maybe others disagree with me.
Fact: Azureus is a CPU and RAM hog. Now granted, give it enough CPU and RAM to work with and obviously you won't notice an impact on system performance. OTOH, try to play any recent 3d game while Azureus is busy, on virtually any system -- you'll find it quickly becomes untenable.
uTorrent does ~99% of what Azureus does, but somehow manages to do it all in a 110k binary, while having virtually no RAM or CPU footprint. (I'm downloading multiple torrents with it just now -- and it's consuming 0-1% cpu, 4,240 kb RAM in task mgr).
However.. some recent changes to uTorrent (latest betas, http://utorrent.com/download/beta/) seem to have rocketed it WAY ahead of the pack in actual transfer completion times. Maybe it's just me, but I'm seeing 5-10x faster overall time for torrents to complete. This appears to have nothing to do with peak bandwidth and everything to do with how quickly uTorrent can connect to peers and begin downloading.
With Azureus I'm accustomed to 10-30 seconds for each peer to establish connection, and another ~10 seconds in the best case to begin actual data transfer. In contrast, with the latest uTorrent beta, I am seeing connections establish in 1-2 seconds, and data begin transferring roughly 1 second after that. The result appears to be that, while my peak transfer rate is about the same as before, uTorrent is managing to keep the average transfer rate consistently high throughout the download. This makes sense, since BT is all about connecting to and switching between peers constantly as it distributes the traffic load. If you've got a relay race going and all the runners are the same speed, but one team takes an extra 30 seconds at each handoff of the baton, you know who's coming in first.
I've switched to uTorrent after being a longtime Azureus fan, because you can not have Azureus running while also playing any recent video game. It's a real memory and CPU hog.
uTorrent by contrast takes virtually no CPU, no RAM, and for that matter only ~110k of disk space. And it has about 99% of the useful features of Azureus. Only really miss a couple of things, like the "swarm" tab -- and that's just for fun really.
Running Azureus and HL2 at the same time would render the game virtually unplayable for me. uTorrent, on the other hand, doesn't even make a dent in the FPS. In fact I play DOD:Source all the time and forget to kill uTorrent first - and I've never seen a problem in either FPS or lag as a result.
"Before the time of the great Purification, they will make metal roads for iron horses and hang metal ropes in the air."
"First they will bring back pieces of the Moon which will upset the balance and unleash disastrous forces."
"Near the day of the Great Purification, there will be cobwebs spun back and forth in the sky."
"The Purification will begin shortly after humans build a great house in the sky. By then there will be fires everywhere and greedy, selfish, power-mad leaders, internal wars."
Frankly, I think PostgreSQL would considerably more popular if it simply had a better name. Not only is the meaning of the name not apparent (etymology anyone?), but it's not even clear how it should be pronounced.
Then again, FireBird isn't nearly as popular as either MySQL Or Postgres. But Postgres is the clear winner in breadth and power of features.
Well, this librarian just doesn't get it. He thinks that by making the texts searchable, that readers will consequently read just the 1-2 sentence snippets and that will be the end of it.
In reality, readers will actually just locate the texts pertaining to their search, and then read those texts in their entirety outside of the search results page -- inasmuch as they choose to. Presumably, scholarly readers would choose to read the entire texts more often than others.
Following his reasoning, we would all have incomplete understandings of many paperback novels which we've flipped through in a bookstore. Which may in fact be true, if we didn't read the whole book -- but we read all that we *wanted* to. And we will continue to do so with the emergence of Google Print.
Slashdot Mirror
What about basic HTML mode? Or does that constitute 'disabling Javascript'?
Does it sound good?
I have wondered for some time why nobody is implementing an "intelligent" traffic-light-timing system. Think Google for traffic light timing.
Where I live, all of the major traffic lights have cameras attached (to automate sending of traffic tickets to red-light runners). It seems to me that these cameras could be used to analyze the routes that are being taken by vehicles and intelligently optimize the light timing based on this.
For example, let's assume that tracking vehicles by license plate number in this manner is too much an invasion of privacy, but we will allow the system to track a vehicle's route by color and size. Let's also assume that it's 3am and there are very few cars on the road.
Say a given small red vehicle leaves a given intersection at 3:05am. This system would be able to estimate how soon that vehicle should arrive at the next adjoining intersection, and adjust the timing of that light so that it's green when the vehicle arrives (other traffic notwithstanding).
Of course there's some chance that the vehicle will not actually reach that second light (e.g. it pulls into a driveway). So for each intersection, the system would actually calculate the probability of a vehicle arriving from each direction in real time.
Now let's say this vehicle's driver gets off work every weekday at 3:00-3:10am and almost always takes the same route home. Over several days, the system should be able to recognize this trend and optimize the lights such that, as soon as the vehicle is recognized on the road, this driver gets all green lights all the way home, most of the time. Thus the system could recognize not only the likelihood of a vehicle arriving at a given intersection at a specific moment, but also the likelihood of a "small red vehicle" taking a left at a given intersection, both in realtime and historically.
So that's the "one vehicle, no congestion" conceptualization. Now just extend this to include all vehicles on the road. Such a system should then be able to optimize 'global' traffic flows, based on this simple underlying probability model.
Of course, special optimizations for heavy congestion, accidents, weather, etc. could be layered on top.
I think the fundamental question, "why can't all the lights change to green for me, as I arrive at them, when there is no other traffic to consider?" is adequately answered by this proposed system. It should also be able to optimize traffic flows better than any human traffic engineer during the most congested times of day, relying especially on historical trends of all the vehicle routes it has recognized.
Since such a system could utilize existing traffic cams and pressure plates, I think it would also be (relatively) inexpensive to implement such a system.
Your post was so banal, it made me write this.
I think you should be able to accomplish this in 5 minutes by creating a local HTML page incorporating FRAMESET/FRAME tags, along with a separate 'subpage' to be loaded into each frame containing a pseudo address bar (i.e. an INPUT TYPE=TEXT element, a Go button and a line of javascript).
Correct. However, it could be quite useful for the popular stuff. Some of the episodic torrents out there can have a following of thousands of peers, subscribed via RSS. This tech is a nice match for that.
I think it would be most useful if this tech was integrated into the mainline BT protocol or some of the more popular BT clients. I just don't see many folks switching to a dedicated client for BT based VOD/videostreams; YouTube is just too easy. But if this was integrated into common BT clients, "Swarmplayer mode" could be activated simply by attempting to preview a video file in the BT client.
I believe they are trying to address this problem by adding NPC 'helpers' to the weaker side in PVP/RVR battles where the population is unbalanced.
That won't help the "ghost town" phenomenon that can result from population distribution problems, but it should hopefully make RvR "fair".
Shortest path from disney to fuck
The Walt Disney Company
Motion Picture Association of America film rating system
Fuck
2 clicks needed
Visual scenes involving objects could be dynamically 3d-rendered to defeat "image dictionary" attack strategies.
For example, "the [cat] is [under] the [car]". The three bracketed terms could be replaced with a large set of nouns or verbs/prepositions.
This scene description could then be rendered from a number of different camera positions/angles; colors changed; and background or extraneous/obscuring foreground objects added.
Until computer intelligence reaches the level of human intelligence for interpreting visual information, there are still CAPTCHA methods available that can't be reliably broken using automated (non human) methods.
http://openvpn.net/
I watched the GMail storage space counter for 1 minute, noting the starting and ending numbers. Then I typed this into Google:
(3113.335029 - 3113.246785) MB/minute * 1 year
= 45.3240458 gigabytes
If the counter really keeps moving at this pace, those prices for extra storage are going to seem pretty steep.
It should obviate the need for a 'delete attachments' button or sorting by size, though.
Possible motivations for Google:
1. Further promote Google's value and brand in the minds of consumers and business owners alike. This is following in the same vein as the Google Local Business Center's completely free coupons that you can create for your business to appear on Google Maps. Call it good PR.
2. Statistics. Useful not only to them, but to local business owners. Again, develops brand loyalty and is good PR.
3. As part and parcel of 1 and 2, they hope that some of these local business owners will later purchase normal ads on Google.
In short, even if the added revenue from 3 doesn't completely offset the cost of running this service, 1 and 2 add value in ways that are hard to directly quantify in monetary terms. Though I suspect they will indeed make more (regular Adwords/Adsense) revenue as a direct result of this service's availability than it costs to run it. The usual exorbitant $1.25 per 411 call is not even close to what it costs to run such a service (especially without human operators).
4. Don't be evil. And while you're at it, if you can take out a few of the other evildoers out there, all the better! Google has taken the lessons of "disruptive technology" to heart, and it's worked for them time and again.
As usual, Google wins by not being overly (or at least overtly) greedy. It's for this reason that I doubt we'll ever hear paid ads or purchased placement when calling the service.
Diebold is the leading manufacturer of ATM machines. Those machines are nigh-impossible to break into with physical access (magstripe hacks & minimart ATMs notwithstanding). Diebold has some serious security-minded people on staff, and they know what they're doing.
Which leads to the only rational conclusion: that Diebold used its extensive security knowledge to ensure that as many avenues of security penetration are available *as possible*. That so many security holes do exist is a matter of record, and it is plain to see from these reports that it would take someone with a deep understanding of security issues to enable such a variety of attack vectors.
I mean, polymorphable executable code can reside on the flash cards meant to hold the vote count totals. That functionality had to be specifically designed and implemented. This was no mistake.
Please watch Hacking Democracy: http://youtube.com/watch?v=yj530xW7PLE
Well, I'm not a biologist but there have been studies which do suggest a biological basis for a homosexuality-overpopulation correlation.
8 10591.html
For example here's a study showing that sons with older brothers are more likely to be homosexual.
http://www.smh.com.au/articles/2003/03/27/1048653
"If Blanchard is right, then clearly, as average family size decreases, so will the incidence of male homosexuality."
It certainly makes sense as a "naturally selected" evolutionary outcome. As a geographical region became overpopulated with humans, those families and communities which produced fewer offspring would be less likely to overpopulate or overwork the land. Those communities whose mothers and daughters produced too many offspring would be less likely to survive into the next generation. So, mothers who were more likely to produce homosexual offspring (at least sons) after her firstborn would present an evolutionary advantage, and her genetics would be thereby "selected" in future generations.
Similarly it would be an evolutionary advantage, as a mother, to still produce able-bodied sons to work for the community, but not necessarily to have all those sons procreating all the time.
Phil Harrison: .. If you're a complete music fan and video fan, and you want to have huge amounts of digital content, then you can upgrade to whatever size of drive you like. You can put any in that you like - it is a computer, after all.
= 16969
gamesindustry.biz: So that hard drive is a standard PC drive?
Phil Harrison: ATA, bog standard, yeah.
gamesindustry.biz: You're not going to be selling Sony drive upgrades?
Phil Harrison: We've got no plan to. We may offer something, but we have no plan to at the moment.
http://www.gamesindustry.biz/content_page.php?aid
Out of curiosity, what does the SQL spec say about mixed case / case sensitivity on db objects?
Ok I'll RTFS if you insist but you already seem to know..
Hmm, my mistake. Perhaps I was hallucinating.
If so, it wasn't a very pleasant hallucination.
Table names (and perhaps even column names) have to be in ALL CAPS. Sorry can't name a table PersonContact, you must use PERSON_CONTACT.
Guess it's just my aesthetic pedanticness but I do not like ALL_CAPS except in my C constants. It also makes any kind of automated conversion from another DB that much more annoying.
A bad design decision that's never really been fixed. I believe it is possible to get UpperLowerCase table names in Postgres, but you have to jump through so many error-prone hoops, it's not worth the trouble.
This is my point, though -- with uTorrent, you don't need to add RAM, or demote the process priority, or for that matter, require a JRE to be loaded. In other words, if there is a far superior (IMHO) alternative tool for the job, what reason is there to carry on using the inferior tool?
If you need those features which Azureus has that uTorrent doesn't provide (and honestly there aren't many), continuing to use Azureus would make sense. At least for me though, the probable remaining reason of "being used to Azureus", or not having to make the effort of using a different product, is not compelling.
If anything, that "if it ain't broke" attitude of is part of what keeps me motivated to find the best tool I can for the job -- and be ready to switch when the best tool is replaced by something markedly better. Otherwise, I may be missing out on an opportunity. (ever try to get someone to switch from IE to a tabbed browser?)
Maybe you can elucidate for me why, if two pieces of software provide equivalent functionality, but one runs in a CPU/RAM/disk footprint a fraction the size of the other -- why you would then keep using the greedier software.
The main reasons I've heard in the past are:
- "It Doesn't Run On Non-Windows." Duh, and this discussion is not even pertinent to non Windows users.
- "It Is Not Open Source". OK, but it is Free, so this is really just the ol' argument of Open Source vs. Demonstrable Performance. And the argument that Azureus "could" become better with user contributions is a moot point - if it DOES become better then use Azureus but until then, it actually IS NOT better.
- "My Grandma Is Already Used to Azureus". Well, I think it's healthy to be able to adapt and make the best of what is available. Maybe others disagree with me.
Fact: Azureus is a CPU and RAM hog. Now granted, give it enough CPU and RAM to work with and obviously you won't notice an impact on system performance. OTOH, try to play any recent 3d game while Azureus is busy, on virtually any system -- you'll find it quickly becomes untenable.
.. some recent changes to uTorrent (latest betas, http://utorrent.com/download/beta/) seem to have rocketed it WAY ahead of the pack in actual transfer completion times. Maybe it's just me, but I'm seeing 5-10x faster overall time for torrents to complete. This appears to have nothing to do with peak bandwidth and everything to do with how quickly uTorrent can connect to peers and begin downloading.
uTorrent does ~99% of what Azureus does, but somehow manages to do it all in a 110k binary, while having virtually no RAM or CPU footprint. (I'm downloading multiple torrents with it just now -- and it's consuming 0-1% cpu, 4,240 kb RAM in task mgr).
However
With Azureus I'm accustomed to 10-30 seconds for each peer to establish connection, and another ~10 seconds in the best case to begin actual data transfer. In contrast, with the latest uTorrent beta, I am seeing connections establish in 1-2 seconds, and data begin transferring roughly 1 second after that. The result appears to be that, while my peak transfer rate is about the same as before, uTorrent is managing to keep the average transfer rate consistently high throughout the download. This makes sense, since BT is all about connecting to and switching between peers constantly as it distributes the traffic load. If you've got a relay race going and all the runners are the same speed, but one team takes an extra 30 seconds at each handoff of the baton, you know who's coming in first.
I'd be interested in hearing if anyone else is seeing this kind of dramatic improvement.
http://utorrent.com/download/beta/
I've switched to uTorrent after being a longtime Azureus fan, because you can not have Azureus running while also playing any recent video game. It's a real memory and CPU hog.
uTorrent by contrast takes virtually no CPU, no RAM, and for that matter only ~110k of disk space. And it has about 99% of the useful features of Azureus. Only really miss a couple of things, like the "swarm" tab -- and that's just for fun really.
Running Azureus and HL2 at the same time would render the game virtually unplayable for me. uTorrent, on the other hand, doesn't even make a dent in the FPS. In fact I play DOD:Source all the time and forget to kill uTorrent first - and I've never seen a problem in either FPS or lag as a result.
"Before the time of the great Purification, they will make metal roads for iron horses and hang metal ropes in the air."
"First they will bring back pieces of the Moon which will upset the balance and unleash disastrous forces."
"Near the day of the Great Purification, there will be cobwebs spun back and forth in the sky."
"The Purification will begin shortly after humans build a great house in the sky. By then there will be fires everywhere and greedy, selfish, power-mad leaders, internal wars."
~ pre-Colombian Hopi prophecies
Just crazy talk?
http://deoxy.org/omega.htm
That's really not different than working as particle physicist or renowned literature writing, per se.
Frankly, I think PostgreSQL would considerably more popular if it simply had a better name. Not only is the meaning of the name not apparent (etymology anyone?), but it's not even clear how it should be pronounced.
Then again, FireBird isn't nearly as popular as either MySQL Or Postgres. But Postgres is the clear winner in breadth and power of features.
Well, this librarian just doesn't get it. He thinks that by making the texts searchable, that readers will consequently read just the 1-2 sentence snippets and that will be the end of it.
In reality, readers will actually just locate the texts pertaining to their search, and then read those texts in their entirety outside of the search results page -- inasmuch as they choose to. Presumably, scholarly readers would choose to read the entire texts more often than others.
Following his reasoning, we would all have incomplete understandings of many paperback novels which we've flipped through in a bookstore. Which may in fact be true, if we didn't read the whole book -- but we read all that we *wanted* to. And we will continue to do so with the emergence of Google Print.