Accuracy: avoids all those silly retractions, libel, etc.
Relevance: Information that is meaningful or interesting or useful to the reader.
Depth: not popular among the CNN/USAToday set, but much needed.
Does the net help?
Speed: yes! two words: global bandwidth
Accuracy: Maybe: Distributed information gathering and rapid feedback help jointly edit/discredit stories. But the process can also feed a "tyranny of the majority" group delusion.
Relevance: Yes: feedback scoring mechanisms (such as/. moderation system) help good stories bubble-up and bad stories drop to invisibility.
Depth: Yes: Wiki-like group editing can add depth, although it may be too slow for "news" stories (more appropriate for feature articles and thought pieces).
Overall, I'd say that grassroots journalism could be good journalism if the system can create the self-regulatory structures needed. Something like (but better than)/. moderation/metamoderation system would be needed to create distributed control over who posts stories and how they are edited or augmented over time.
Darwinian evolution (in which the genes affect reproductive success) will have a decreasing role in future. The ability to repair congenital defects, correct metabolic disorders, and cure life-threatening conditions means that natural selection does not occur with the same intensity as in the past. More people survive and reproduce would not of in the past.
The one area where Darwinian evolution may play a role is in how people respond to pharmaceuticals. Not all drugs work on all people -- some people cannot tolerate certain drugs and other people metabolize a medication so quickly that it is ineffective. These people will find themselves part of the orphan disease population -- populations that are too small to be worth the effort to develop drugs for. In time, them may succumb more frequently to medical problems and become less prevalent in the population.
What we will see is more evolution of memes (rather than genes). Memetic evolution is Lamarckian, not Darwinian. Whereas genes are markedly stable (the copy error rate is very low), memes are more malleable and tend to acquire new characteristics that are then passed on.
Thus, I would argue that Lamarckian evolution will play a bigger role in the future than Darwinian evolution. The characteristics that people (and society) acquire will be passed on to the next generation. The new technologies, new terminologies, new ideas, and new ways of living will define humanity's future and a person's life far more than does the genetic sequence of a person's DNA.
Microsoft rode IBM's coattails in driving PC adoption within business. In the 1980s, computers were primarily a business tool (PCs were very expensive, especially when costed in today's dollars). Adoption in business then drove adoption at home (PCs migrated to the home, TRS-80s did not migrate to business). Apple lost to Microsoft because it didn't have business applications or a reputation for catering to business in the way that IBM did.
Learning from this history suggests that the key to Linux' success is to get those B2B companies that market computers (and computer-based services) to adopt and promote Linux for business applications. As with the PC, businesses won't care if the OS is MS or FOSS as long as the costs are right and the risks are low (=reputable companies with a reputation for good service) -- remember that back then, "nobody got fired for buying IBM.". That may be a valid strategy today or it may not.
The problem is that those who learn from history are doomed to repeat it. MS rode the wave of computer adoption in the world and that wave clearly progressed from business applications to consumer applications (and from developed nations to developing nations). In contrast, now consumer applications (gaming, media, internet) and developing nation growth (low-cost PCs) seem to be driving the technology more so than business market. The mainstream business market (e.g., Office-running desktops and laptops) is saturated and does not really need faster processors and hotter graphics chips in the same way that consumer device do.
In this new environment, Linux may have more success in driving adoption in the embedded and consumer-electronics market (where low-cost is a key advantage and the engineering overhead of "learning linux" is less important). As long as a Linux device has a familiar GUI and appropriate compatibility (e.g., web, email, and "office" file compatibility), it can be accepted.
IBM and MS are like the big-three U.S. automakers -- dominant in the first wave of a new technology. But, somewhere out there is the Toyota of the PC industry, possibly in China or India or elsewhere. They will make a cheaper and higher-quality product (perhaps with Linux) that displaces the incumbent. Or you could say that IBM & MS are like Sears (which dominated the first wave of mass-market retailing in the U.S). Perhaps somewhere there is a nascent Wal-Mart of the PC world (which first arose in the backwoods of Arkansas to become the largest company in the world).
In the extended hours markets, some traders post bids or offers at outrageous prices, hoping someone will make a mistake. They will offer to buy some stock for $0.02 per share or sell some stock for 200.00 per share (that normally trades $20 a share) in hopes that someone screws up. The low cost of participating in an electronic market makes it easy to post these orders.
The real lesson is that stocks don't really have "a price" in a traditional sense. (At best, the price on the last transaction serves as a proxy, but is no guarantee of getting that price in the future). In reality, stocks have both a bid and an ask price. For thinly traded stocks (especially in the off-hours), the bid-ask spread can be very very large.
Somehow internal R&D was bad for the books- but seed money came from a different pile.
R&D is bad for the books because so much of the money is wasted on ideas that fail. Letting start-ups do the R&D, initial product roll-out and marketing lets you do R&D totally off the books because someone else foots the bill. Rather than fund 100 R&D projects and see one create a successful project, you can let 100 start-ups do it and buy the one start-up that has a good idea.
Cisco also had the advantage that buying startups cost that "nothing." Rather than buy the startup in cash, the used stock to purchase the company (it dilutes the stock a bit, but Cisco almost never bought large companies). You are right about the issue of R&D on the books, because internal R&D cost real dollars, acquired R&D (buying startups) only cost shares.
Absolutely true (up to the no-firing part). Look at all the IT Consulting and Business Process Outsourcing firms. Accenture has 100,000 employees, EDS has 117,000, Cap Gemini has 49,000. They act like the "big firm" and shift employees from project to project.
Of course when they lose a big contract or don't book enough business, they still lay people off.
Cisco pursued this strategy of buying small companies with promising technology but in the dotcom days. Effectively, Cisco let others finance the R&D and initial market testing that all start-ups perform. If the start-up failed, it was no skin off Cisco's nose. If the start-up worked and had a product (and company vision) that matched Cisco, then they bought you.
Its a great way to get innovative and market-proven technologies, but it can be a little piece-meal. Not all the great start-up techs are going to fit into your architecture or segment the market in a clean way. Also, due diligence can't uncover every problem when you take the start-up's tech and try to scale it to big-company volumes and service expectations. Finally, buying start-ups is a very public affair compared to a tightly secured internal R&D function -- the minute you buy a startup (with a product on the market), all your competitors know where you are going.
Buying start-ups is a nice tool, but it can't totally replace (only augment) an internal, integrated approach to R&D, product development and architecture development.
I'm surprised that the PSU and all the cables (like speaker/CAT5) work at all, I feel so uneducated.
Actually, I'd be more worried about the high-speed circuits in the machine. Oil does not conduct electricity, but that doesn't mean its electronically equivalent to air.
Oil has a dielectric constant of between 2 and 3 (depends on the oil) and that will affect the capacitance on and between the traces of the circuit-board. The signals will run a little slower on the board and have a bit more cross-talk. Its probably not a big deal -- the materials in the circuit board have a bigger effect -- but it could slow the signals enough to reduce reliability in a marginal design.
A very basic computer needs very little memory. I've got some older machines that are more than functional (for basic office work) that have only 16 MB of RAM and 200 MB of disk space. They're fine for word processing, spreadsheets, small databases, and email. I can even use the web, although the high level of graphics and gratuitous formatting on many websites makes it a slow experience.
Just because new machines need 1 GB RAM and 60 GB HDs, doesn't mean you can do anything with 1/10 or even 1/100 of the memory of a modern PC.
This is an interesting analysis, but unfortunately completely wrong. The thing is that the Number Field Sieve algorithm's complexity is sub- exponential in number length. (To be precise, it's O(exp(c*log(n)^(1/3)*loglog(n)^(2/3)+o(1))) ).
Thanks for the info! What is the value of c? Does it have a lower bound? This is what I am talking about. All of these t = O(???) equations have constants in them that may be lower than people think.
That said, it doesn't seem that the factoring problem will become any easier, at least not before Quantum computers are built. The factoring problem is considered "the holy grail" of cryptography for 3 decades now, and there were hardly any advances in the last 15 years, despite the huge interest in the problem
True. But can future progress be extrapolated from past progress? I suspect that technical progress probably acts like a punctuated equilibrium system. A new advance creates massive innovation and progress that then hits some asymptotic limit until the next innovation hits.
The base is superflous. Factoring is approximately linear in the key size as a number, but said to be 'exponential' in the number of digits in the key.
Agreed. I was referring to the base of the exponent in the exponential formula for the factoring time. If the running time of the algorithm is t = A * B ^ N. A is a speed constant (decreasing with Moore's law). N is the number of bits in the number and B is the (perhaps misnamed) base of the exponent. For a brute-force algorithm, B = 2. For a better algorithm, B is less than 2. How much less than 2 is the issue.
The algorithms that exist must search the vast majority of the keyspace.
No, only brute force algorithms must search the vast majority of the keyspace. Can anyone prove that all possible algorithms have to do this? If, not, then factoring may be easier than we think or want.
If an author works years to create some opus, then copyright law should give them long-term control over distribution. Some authors create very few works in their lifetime and, arguably, might deserve "lifetime" protection.
But this same law is used (abused) by mass-market content creators to protect every bit of pulp/drek/shovelware they put out. The current copyright law is geared toward individual authors (with corporations only too happy to ride the coat-tails).
You could argue that patent law has the same problem - how do you afford appropriate levels of protection to ideas that take millions (or billions) of dollars to create versus those created in the blink of an eye by some patent-happy company.
Factoring numbers looks harder than it is. At first glance, it looks like adding digits makes the factoring problem exponentially harder. The question is: what is the base of the exponent. A naive analysis suggests that adding one binary digit makes the number twice as big and thus makes the factoring problem twice as hard. Such analyses are where get estimates that proclaim it will take a computer the life of the universe to factor an X-digit number.
If adding one bit to the number, makes the problem twice as hard, then the base of the exponent for the executive time is 2. But what if the base is not 2, but is only 1.01? Then, adding 200 bits to the number only makes the problem 7 times harder (1.01 ^ 200). The scary part is that we can't prove that factoring has a lower limit to the base of the exponent. It could be 1.1, 1.01, or 1.001, or 1.0001. This means that any crypto based on prime factors has an unknown vulnerability in it.
For now, prime factoring is hard, tomorrow, it might not be.
If there is one overriding reason that I hate MS Office, it is that it feels like the application was developed by a thousand independent programmers. Consistency between and within Office applications is very poor. Each feature seems to have its own UI logic, limitations, behaviors.
A bounty program is great. But if it creates a thousand independent bolt-on features, it will suck. Perhaps some high-level architect in each project can create some stub classes or documentation that define exactly what the bounty-earning feature must do and how it should conform to a set of UI guidelines.
/. discussed the Witty worm back in 2004. This analysis used UCSD Network Telescope IP block (containing 1/256 of IPv4 space) to sample the randomly spewed packets created by the worm. They were able to analyze quite a few interesting features, including the fact that the worm was jump-started by an infection of about 110 PCs at the outset, 24-hour cycles in infected/reinfected machines, and data on the distribution of bit-rates of worm transmitters.
Hybrids are a nice interim solution, but they are nothing more than a more efficient version of the internal combustion engine. A big heavy hybrid SUV is still a gas guzzler.
It's great to see people jumping on the hybrid bandwagon, but only if it leads to truly green solutions using renewable energy. If people think hybrids are the ultimate solution, they will be sadly mistaken.
TFA, further down the page, describes the user experience of a Cabir infection. The recipient must click "yes" a number of times to accept the unknown transmission, install the unknown file, and bypass a security warning about installing something from an unverified supplier. Why do people click "yes" to all this? Because if you click "No" the virus keeps trying to install itself and pester you with the messages.
Definitely reminds me of "Abort/Retry/Fail" error message of so long ago. The first time you ever see the message, you hit "retry" a few times hoping it will work. Eventually, the computer teaches you to never try "retry" because it only puts up the error message again.
This virus is social engineering at its best, just like the whiny kid in the grocery store. Keep pestering until they say "yes."
Until these things have run for a few thousand hours, we won't know if they have burn-in problems. I would expect the carbon nanotubes to erode at the ends due to the extremely high electronic fields and for carbon tube fragments to eventually poison the phosphors.
TFA touts that this "could offer... longer lifetimes," but I'm always cynical of breathless promises of what some future technology "could" bring.
Technology like this will take off when we can make flexible circuitry that can conform to the inner surface of the retina in a moving eyeball. The ultimate artificial retina would have a photosensitive array on one surface and a nerve-stimulating grid on the other surface. A small transducer coil elsewhere on the body would provide power to the unit.
Finding a new elastomeric polymer with conductive/semi-conductive properties (think stretchy OLED polymers) would help make this happen. Or perhaps blending silicon and silicone could be possible. I could also see using RFID manufacturing technologies (which can handle silicon dies smaller than 0.3 mm) to create arrays of semiconductor circuits embedded in a flexible polymer plastic matrix.
Once the electronics becomes flexible enough, they could be used in a wide array of neurological repair/augmentation applications.
I've been hearing about reuse of code and the development of stable shared libraries for the past 20 years and its probably been going on for longer than that. Why don't people, especially OS and application people, create, debug, and reuse a set of overflow-proof buffer-handling libraries? The libraries could include a range of forked versions for different usage patterns (e.g., big buffers of small data objects, small buffers of big data objects, buffers optimized for variable or fixed size, buffers optimized for frequent writes/sorts/reads/etc. Why is that so hard?
Every buffer-overflow exploit is just evidence of re-invention of a bug-filled wheel.
Current multispectral stuff is more than 300m? Interesting, it would miss entire steams, even rivers and stuff.
Good point and that seems plausible, but is not entirely true. With a good pixel-mixing analysis you can resolve stuff inside the pixel. The key is having a clean spectral model for the terrain versus water and being able to say that a given pixel looks like its 90% trees and 10% fresh water. "Unmix" enough pixels and you can string them together to find streams smaller than 300 m wide that cross an expanse of forest or grass-land. A string of 10% water pixels in a 300 m resolution image is probably a 30 meter-wide stream. Moreover, the plants around water often different substantially from the plants in the drier surrounding areas -- making it even easily to infer the existence of creeks that are far smaller than the resolution limit.
MERIS grabs data in up to 15 spectral bands with 16 bits per band per pixel. Its only has a 1/2 megapixel imager (842 x 691), but the RAW images are 17.5 MB.
Multispectral data is great for identifying ground cover (e.g, classifying the types of plants, health of plants, minerals, etc. on the ground). Sometimes, it's more valuable to know the materials on the ground than to see the geometric detail.
MS Office is a monopoly because it is a monopoly -- its not the best set of applications, but it's prevalence forces anyone who is a supplier to conform to the IT choices of key clients. A two-person shop can't convince a $1 billion organization to change its IT infrastructure.
Sadly, we have enough minor glitches between our Mac Office documents and our PC-using clients to scare us away from Open Office for documents to/from clients (i.e., 95% of our office application work). The only consolation is that we get paid the extra labor for dealing with Word/Excel/Powerpoint and the well documented annoyances. Even so, what enrages me the most is that a recent upgrade from Mac Office 98 to Mac Office 2004 proved that MS has done almost nothing to fix long-standing bugs in 6 years and 3 versions (and has added a few new "annoyances").
I may be a wimp in not taking the OpenOffice plunge, but when your livelihood depends on being "easy to work with" its hard to take risks that could mean the loss of 33% or more of your income.
I can only hope that the tech is given with a clear release from any future patent claims by MS. Without that, any company that uses MS technology is very vulnerable to future claims for use of MS' intellectual property.
Otherwise, its a good move by MS to expand the economic pie of the MS universe.
Slashdot Mirror
- Speed: the "new" in news.
- Accuracy: avoids all those silly retractions, libel, etc.
- Relevance: Information that is meaningful or interesting or useful to the reader.
- Depth: not popular among the CNN/USAToday set, but much needed.
Does the net help?- Speed: yes! two words: global bandwidth
- Accuracy: Maybe: Distributed information gathering and rapid feedback help jointly edit/discredit stories. But the process can also feed a "tyranny of the majority" group delusion.
- Relevance: Yes: feedback scoring mechanisms (such as
/. moderation system) help good stories bubble-up and bad stories drop to invisibility.
- Depth: Yes: Wiki-like group editing can add depth, although it may be too slow for "news" stories (more appropriate for feature articles and thought pieces).
Overall, I'd say that grassroots journalism could be good journalism if the system can create the self-regulatory structures needed. Something like (but better than)Darwinian evolution (in which the genes affect reproductive success) will have a decreasing role in future. The ability to repair congenital defects, correct metabolic disorders, and cure life-threatening conditions means that natural selection does not occur with the same intensity as in the past. More people survive and reproduce would not of in the past.
The one area where Darwinian evolution may play a role is in how people respond to pharmaceuticals. Not all drugs work on all people -- some people cannot tolerate certain drugs and other people metabolize a medication so quickly that it is ineffective. These people will find themselves part of the orphan disease population -- populations that are too small to be worth the effort to develop drugs for. In time, them may succumb more frequently to medical problems and become less prevalent in the population.
What we will see is more evolution of memes (rather than genes). Memetic evolution is Lamarckian, not Darwinian. Whereas genes are markedly stable (the copy error rate is very low), memes are more malleable and tend to acquire new characteristics that are then passed on.
Thus, I would argue that Lamarckian evolution will play a bigger role in the future than Darwinian evolution. The characteristics that people (and society) acquire will be passed on to the next generation. The new technologies, new terminologies, new ideas, and new ways of living will define humanity's future and a person's life far more than does the genetic sequence of a person's DNA.
Microsoft rode IBM's coattails in driving PC adoption within business. In the 1980s, computers were primarily a business tool (PCs were very expensive, especially when costed in today's dollars). Adoption in business then drove adoption at home (PCs migrated to the home, TRS-80s did not migrate to business). Apple lost to Microsoft because it didn't have business applications or a reputation for catering to business in the way that IBM did.
Learning from this history suggests that the key to Linux' success is to get those B2B companies that market computers (and computer-based services) to adopt and promote Linux for business applications. As with the PC, businesses won't care if the OS is MS or FOSS as long as the costs are right and the risks are low (=reputable companies with a reputation for good service) -- remember that back then, "nobody got fired for buying IBM.". That may be a valid strategy today or it may not.
The problem is that those who learn from history are doomed to repeat it. MS rode the wave of computer adoption in the world and that wave clearly progressed from business applications to consumer applications (and from developed nations to developing nations). In contrast, now consumer applications (gaming, media, internet) and developing nation growth (low-cost PCs) seem to be driving the technology more so than business market. The mainstream business market (e.g., Office-running desktops and laptops) is saturated and does not really need faster processors and hotter graphics chips in the same way that consumer device do.
In this new environment, Linux may have more success in driving adoption in the embedded and consumer-electronics market (where low-cost is a key advantage and the engineering overhead of "learning linux" is less important). As long as a Linux device has a familiar GUI and appropriate compatibility (e.g., web, email, and "office" file compatibility), it can be accepted.
IBM and MS are like the big-three U.S. automakers -- dominant in the first wave of a new technology. But, somewhere out there is the Toyota of the PC industry, possibly in China or India or elsewhere. They will make a cheaper and higher-quality product (perhaps with Linux) that displaces the incumbent. Or you could say that IBM & MS are like Sears (which dominated the first wave of mass-market retailing in the U.S). Perhaps somewhere there is a nascent Wal-Mart of the PC world (which first arose in the backwoods of Arkansas to become the largest company in the world).
In the extended hours markets, some traders post bids or offers at outrageous prices, hoping someone will make a mistake. They will offer to buy some stock for $0.02 per share or sell some stock for 200.00 per share (that normally trades $20 a share) in hopes that someone screws up. The low cost of participating in an electronic market makes it easy to post these orders.
The real lesson is that stocks don't really have "a price" in a traditional sense. (At best, the price on the last transaction serves as a proxy, but is no guarantee of getting that price in the future). In reality, stocks have both a bid and an ask price. For thinly traded stocks (especially in the off-hours), the bid-ask spread can be very very large.
Buyer (and seller) beware.
I hope that this data will motivates webmasters and site designers to create more universally viewable sites.
Somehow internal R&D was bad for the books- but seed money came from a different pile.
R&D is bad for the books because so much of the money is wasted on ideas that fail. Letting start-ups do the R&D, initial product roll-out and marketing lets you do R&D totally off the books because someone else foots the bill. Rather than fund 100 R&D projects and see one create a successful project, you can let 100 start-ups do it and buy the one start-up that has a good idea.
Cisco also had the advantage that buying startups cost that "nothing." Rather than buy the startup in cash, the used stock to purchase the company (it dilutes the stock a bit, but Cisco almost never bought large companies). You are right about the issue of R&D on the books, because internal R&D cost real dollars, acquired R&D (buying startups) only cost shares.
Absolutely true (up to the no-firing part). Look at all the IT Consulting and Business Process Outsourcing firms. Accenture has 100,000 employees, EDS has 117,000, Cap Gemini has 49,000. They act like the "big firm" and shift employees from project to project.
Of course when they lose a big contract or don't book enough business, they still lay people off.
Cisco pursued this strategy of buying small companies with promising technology but in the dotcom days. Effectively, Cisco let others finance the R&D and initial market testing that all start-ups perform. If the start-up failed, it was no skin off Cisco's nose. If the start-up worked and had a product (and company vision) that matched Cisco, then they bought you.
Its a great way to get innovative and market-proven technologies, but it can be a little piece-meal. Not all the great start-up techs are going to fit into your architecture or segment the market in a clean way. Also, due diligence can't uncover every problem when you take the start-up's tech and try to scale it to big-company volumes and service expectations. Finally, buying start-ups is a very public affair compared to a tightly secured internal R&D function -- the minute you buy a startup (with a product on the market), all your competitors know where you are going.
Buying start-ups is a nice tool, but it can't totally replace (only augment) an internal, integrated approach to R&D, product development and architecture development.
I'm surprised that the PSU and all the cables (like speaker/CAT5) work at all, I feel so uneducated.
Actually, I'd be more worried about the high-speed circuits in the machine. Oil does not conduct electricity, but that doesn't mean its electronically equivalent to air.
Oil has a dielectric constant of between 2 and 3 (depends on the oil) and that will affect the capacitance on and between the traces of the circuit-board. The signals will run a little slower on the board and have a bit more cross-talk. Its probably not a big deal -- the materials in the circuit board have a bigger effect -- but it could slow the signals enough to reduce reliability in a marginal design.
A very basic computer needs very little memory. I've got some older machines that are more than functional (for basic office work) that have only 16 MB of RAM and 200 MB of disk space. They're fine for word processing, spreadsheets, small databases, and email. I can even use the web, although the high level of graphics and gratuitous formatting on many websites makes it a slow experience.
Just because new machines need 1 GB RAM and 60 GB HDs, doesn't mean you can do anything with 1/10 or even 1/100 of the memory of a modern PC.
This is an interesting analysis, but unfortunately completely wrong. The thing is that the Number Field Sieve algorithm's complexity is sub- exponential in number length. (To be precise, it's O(exp(c*log(n)^(1/3)*loglog(n)^(2/3)+o(1))) ).
Thanks for the info! What is the value of c? Does it have a lower bound? This is what I am talking about. All of these t = O(???) equations have constants in them that may be lower than people think.
That said, it doesn't seem that the factoring problem will become any easier, at least not before Quantum computers are built. The factoring problem is considered "the holy grail" of cryptography for 3 decades now, and there were hardly any advances in the last 15 years, despite the huge interest in the problem
True. But can future progress be extrapolated from past progress? I suspect that technical progress probably acts like a punctuated equilibrium system. A new advance creates massive innovation and progress that then hits some asymptotic limit until the next innovation hits.
The base is superflous. Factoring is approximately linear in the key size as a number, but said to be 'exponential' in the number of digits in the key.
Agreed. I was referring to the base of the exponent in the exponential formula for the factoring time. If the running time of the algorithm is t = A * B ^ N. A is a speed constant (decreasing with Moore's law). N is the number of bits in the number and B is the (perhaps misnamed) base of the exponent. For a brute-force algorithm, B = 2. For a better algorithm, B is less than 2. How much less than 2 is the issue.
The algorithms that exist must search the vast majority of the keyspace.
No, only brute force algorithms must search the vast majority of the keyspace. Can anyone prove that all possible algorithms have to do this? If, not, then factoring may be easier than we think or want.
The base has nothing to do with it.
Agreed, sorry for the confusion.
If an author works years to create some opus, then copyright law should give them long-term control over distribution. Some authors create very few works in their lifetime and, arguably, might deserve "lifetime" protection.
But this same law is used (abused) by mass-market content creators to protect every bit of pulp/drek/shovelware they put out. The current copyright law is geared toward individual authors (with corporations only too happy to ride the coat-tails).
You could argue that patent law has the same problem - how do you afford appropriate levels of protection to ideas that take millions (or billions) of dollars to create versus those created in the blink of an eye by some patent-happy company.
Factoring numbers looks harder than it is. At first glance, it looks like adding digits makes the factoring problem exponentially harder. The question is: what is the base of the exponent. A naive analysis suggests that adding one binary digit makes the number twice as big and thus makes the factoring problem twice as hard. Such analyses are where get estimates that proclaim it will take a computer the life of the universe to factor an X-digit number.
If adding one bit to the number, makes the problem twice as hard, then the base of the exponent for the executive time is 2. But what if the base is not 2, but is only 1.01? Then, adding 200 bits to the number only makes the problem 7 times harder (1.01 ^ 200). The scary part is that we can't prove that factoring has a lower limit to the base of the exponent. It could be 1.1, 1.01, or 1.001, or 1.0001. This means that any crypto based on prime factors has an unknown vulnerability in it.
For now, prime factoring is hard, tomorrow, it might not be.
If there is one overriding reason that I hate MS Office, it is that it feels like the application was developed by a thousand independent programmers. Consistency between and within Office applications is very poor. Each feature seems to have its own UI logic, limitations, behaviors.
A bounty program is great. But if it creates a thousand independent bolt-on features, it will suck. Perhaps some high-level architect in each project can create some stub classes or documentation that define exactly what the bounty-earning feature must do and how it should conform to a set of UI guidelines.
/. discussed the Witty worm back in 2004. This analysis used UCSD Network Telescope IP block (containing 1/256 of IPv4 space) to sample the randomly spewed packets created by the worm. They were able to analyze quite a few interesting features, including the fact that the worm was jump-started by an infection of about 110 PCs at the outset, 24-hour cycles in infected/reinfected machines, and data on the distribution of bit-rates of worm transmitters.
Hybrids are a nice interim solution, but they are nothing more than a more efficient version of the internal combustion engine. A big heavy hybrid SUV is still a gas guzzler.
It's great to see people jumping on the hybrid bandwagon, but only if it leads to truly green solutions using renewable energy. If people think hybrids are the ultimate solution, they will be sadly mistaken.
TFA, further down the page, describes the user experience of a Cabir infection. The recipient must click "yes" a number of times to accept the unknown transmission, install the unknown file, and bypass a security warning about installing something from an unverified supplier. Why do people click "yes" to all this? Because if you click "No" the virus keeps trying to install itself and pester you with the messages.
Definitely reminds me of "Abort/Retry/Fail" error message of so long ago. The first time you ever see the message, you hit "retry" a few times hoping it will work. Eventually, the computer teaches you to never try "retry" because it only puts up the error message again.
This virus is social engineering at its best, just like the whiny kid in the grocery store. Keep pestering until they say "yes."
Until these things have run for a few thousand hours, we won't know if they have burn-in problems. I would expect the carbon nanotubes to erode at the ends due to the extremely high electronic fields and for carbon tube fragments to eventually poison the phosphors.
... longer lifetimes," but I'm always cynical of breathless promises of what some future technology "could" bring.
TFA touts that this "could offer
Technology like this will take off when we can make flexible circuitry that can conform to the inner surface of the retina in a moving eyeball. The ultimate artificial retina would have a photosensitive array on one surface and a nerve-stimulating grid on the other surface. A small transducer coil elsewhere on the body would provide power to the unit.
Finding a new elastomeric polymer with conductive/semi-conductive properties (think stretchy OLED polymers) would help make this happen. Or perhaps blending silicon and silicone could be possible. I could also see using RFID manufacturing technologies (which can handle silicon dies smaller than 0.3 mm) to create arrays of semiconductor circuits embedded in a flexible polymer plastic matrix.
Once the electronics becomes flexible enough, they could be used in a wide array of neurological repair/augmentation applications.
I've been hearing about reuse of code and the development of stable shared libraries for the past 20 years and its probably been going on for longer than that. Why don't people, especially OS and application people, create, debug, and reuse a set of overflow-proof buffer-handling libraries? The libraries could include a range of forked versions for different usage patterns (e.g., big buffers of small data objects, small buffers of big data objects, buffers optimized for variable or fixed size, buffers optimized for frequent writes/sorts/reads/etc. Why is that so hard?
Every buffer-overflow exploit is just evidence of re-invention of a bug-filled wheel.
Current multispectral stuff is more than 300m? Interesting, it would miss entire steams, even rivers and stuff.
Good point and that seems plausible, but is not entirely true. With a good pixel-mixing analysis you can resolve stuff inside the pixel. The key is having a clean spectral model for the terrain versus water and being able to say that a given pixel looks like its 90% trees and 10% fresh water. "Unmix" enough pixels and you can string them together to find streams smaller than 300 m wide that cross an expanse of forest or grass-land. A string of 10% water pixels in a 300 m resolution image is probably a 30 meter-wide stream. Moreover, the plants around water often different substantially from the plants in the drier surrounding areas -- making it even easily to infer the existence of creeks that are far smaller than the resolution limit.
MERIS grabs data in up to 15 spectral bands with 16 bits per band per pixel. Its only has a 1/2 megapixel imager (842 x 691), but the RAW images are 17.5 MB.
Multispectral data is great for identifying ground cover (e.g, classifying the types of plants, health of plants, minerals, etc. on the ground). Sometimes, it's more valuable to know the materials on the ground than to see the geometric detail.
MS Office is a monopoly because it is a monopoly -- its not the best set of applications, but it's prevalence forces anyone who is a supplier to conform to the IT choices of key clients. A two-person shop can't convince a $1 billion organization to change its IT infrastructure.
Sadly, we have enough minor glitches between our Mac Office documents and our PC-using clients to scare us away from Open Office for documents to/from clients (i.e., 95% of our office application work). The only consolation is that we get paid the extra labor for dealing with Word/Excel/Powerpoint and the well documented annoyances. Even so, what enrages me the most is that a recent upgrade from Mac Office 98 to Mac Office 2004 proved that MS has done almost nothing to fix long-standing bugs in 6 years and 3 versions (and has added a few new "annoyances").
I may be a wimp in not taking the OpenOffice plunge, but when your livelihood depends on being "easy to work with" its hard to take risks that could mean the loss of 33% or more of your income.
I can only hope that the tech is given with a clear release from any future patent claims by MS. Without that, any company that uses MS technology is very vulnerable to future claims for use of MS' intellectual property.
Otherwise, its a good move by MS to expand the economic pie of the MS universe.