It's not going to happen. The problem is that there just aren't enough good mobile developers and enough developer mindshare to go around. Lots of developers don't even want to support TWO platforms, let alone three or four or five. There have never been more than two successful mobile platforms at once. First we had Palm and BlackBerry. iOS came along and Palm faded away. Then Android appeared and BlackBerry declined. But I don't see any of the current contenders displacing either iOS or Android; they're just too entrenched. Any new OS would have to be dramatically better in some important way, and neither BlackBerry 10 nor Windows Phone 8 is.
What about Symbian, you ask? Lots of Symbian phones were sold, but although it was theoretically a smartphone platform its success wasn't based on its smartness. The vast majority of Symbian devices were sold as feature phones with no real app capability, and the ones that could meaningfully use apps never got any significant number of them to choose from.
There is one possible scenario for limited success of BlackBerry 10 or Windows Phone, though the latter is not pursuing it. That would be to ignore the consumer market, which is a lost cause anyway, and double down on enterprise applications and try to become the preferred device of IT departments. The BYOD trend means it will be a challenge to turn this into sales, but it could at least get them a place in large corporations that are big on centralized IT and in security-conscious organizations. This would mean things like giving the mobile a top-class application for email, group calendar, and other enterprise uses, rolling VPN and full remote management support into the OS, including a remote desktop client, and so forth. It would also mean including enhanced security features like biometric ID into the handsets.
I am skeptical about the prospects for Mozilla Phone and Ubuntu Phone for the same reasons. It appears, however, that both of those platforms are going to make their push primarily in the third world, where the lower cost of those platforms might matter enough for one of them to gain some traction. It's not just that the software is zero-cost - so is Android - but that they're intended to offer acceptable performance on low-spec devices. The vast majority of apps for the two first-world platforms will likely never get to these platforms, though the big-name apps will.
They already did make the system builder license more transparent. (The new terms only apply to Windows 8, not to earlier versions, though they might apply to copies of Windows 8 Professional where the downgrade right to Windows 7 or Vista is exercised.) It now contains the Personal Use License clause:
System Builder product may be used:
As the operating system on a PC you build for personal use.
As an operating system running either on a local virtual machine or as an additional operating system in a separate partition.
System Builder product may not be used:
As an upgrade license for an existing underlying Windows operating system.
To legalize a non-genuine Windows operating system.
To license more than five copies of the software (in total) for commercial use.
That "more than five copies" clause is part of the Personal Use License and applies to commercial use by the builder. It does not preclude selling more than five commercial use copies to a customer.
The old "full retail license" - the really expensive version of Windows - is essentially dead and Microsoft should discontinue it. People buying computers from major manufacturers will, as they have for years, get OEM licenses. People building their own computers and small boutique computer builders will use the System Builder version. All but the tiniest corporations will use one of Microsoft's volume licensing programs, which start at five seats. There isn't a market niche remaining for the old full license.
Some Apple commercials are also about cool and not about what the device does. Most iPod advertising falls in that category, including the recent one with bouncing iPods.
You're not stuck with 128Kbps for most music. Apple and Amazon both currently sell music at 256Kbps or thereabouts. (Apple uses AAC and Amazon uses MP3; at that bitrate the difference is negligible.) Those two stores offer nearly all the recorded music that is available in the US in any form. eMusic.com uses bit rates ranging from 192Kbps to 320Kbps (and uses the superior LAME encoder) for most music. (A few things are only available at 128Kbps.) Other smaller sellers of downloads also offer higher rates now.
It's hard to do even better, but in some cases you can. HDTracks.com offers uncompressed FLAC downloads of music, much of it in high-resolution forms. (A few albums are even available as 24/192 or 24/176.4 downloads; the former typically means that the remastering was originally done for DVD-Audio and the latter was for SACD.) Some artists sell FLAC downloads, as do a few small labels.
Er, no. 200 channels of 1080p is 50 channels of 4K, barring adoption of a significantly more efficient codec than MPEG-4. 200 channels of SD is at most 20 channels of 4K, allowing for the fact that the SD channel is probably less efficiently encoded in MPEG-2.
For a while, WGBH (PBS) was broadcasting the same content in both SD and HD. When viewed using a digital converter on an old TV, the HD version looked better even after downconversion to 480i by the converter. Recently they switched the SD subchannel to different content; the fact that the SD version was useless is probably the reason.
That's because the sports bar is showing all the sports from cable. The least they could do is use a broadcast source when availalble (sports bars carry lots of content that isn't broadcast); it usually looks much better as most broadcasters are devoting 12-15Mbps to their main channel.
Making cars more efficient is a desirable result. Having the owners of the efficient cars pay less tax is a good way to get that result to happen. More efficient cars are also, on average, smaller and lighter so they cause less wear to the roads.
Going to a mileage-based tax is a bad idea because it removes that incentive. If tax revenue falls too much, raise the tax rate to compensate. Yes, it will punish SUV owners - as it should.
At Harvard Extension nearly all classes are 4 credits. (There are some 2 credit language classes that I did not take.) The math I gave was accurate for that school; other schools vary.
I took some summer classes. My experience that taking one summer class was about the same total intensity as taking two spring or fall classes. The courseload for CS classes was higher than for non-CS classes (in part because most CS classes at Harvard Extension are taught at the graduate level, though you can apply them to an undergraduate degree if you like) so it might be better to take non-CS subjects in the summer.
It's also true that the course load of classes varies wildly. My own boundary case, Computer Architecture, was a full time occupation in itself. That class involved designing a simple CPU architecture, writing an emulator and an assembler for it so you could create programs for it, then writing a VHDL implementation to create a hardware implementation. All in twelve weeks.
There is some truth to your comment on giving up personal life. Fortunately for me, school itself was a rewarding form of personal life, so the limitations on other social existence didn't hurt as much as they might have.
Ten years is a bit of an overestimate for an undergraduate degree.
Where I went to school (Harvard Extension, the continuing education division of Harvard University), classes are 4 credits each and you need 128 credits to graduate. If you bring in no credits at all you need to take 32 classes for a degree. 2 per semester, 4 per year, you'll need eight years. If you add one summer class each year you get it down to just over six years.
Hint: if you do it there take only one CS class each semester! The course load in the CS classes is usually heavier than in the non-CS classes you will need for a degree.
For a Master's degree there you need 12 classes. Master's classes are tougher and they'll all be in CS so you're best off only taking one each semester, which will take six years or four years if you add summers.
Sometimes private schools have useful offerings as well. Here in the Boston area, both Harvard and BU offer evening computing classes and have degree options; at Harvard you can take a lot (but not all) of the classes as distance learning and do most of the work on your own schedule. (If the class has exams rather than or in addition to projects you have to show up in person for those.)
Computer science and software engineering are very different things. The former is largely about theory and is often heavily mathematical. The latter is about the practical things that the CMU student didn't get. Many of the best known CS programs have the same weakness; courses on the practicalities of programming are either not required or not even offered.
When your standard of comparison is the #1 film of all time, everything is going to be downhill. There have been other notable artistic accomplishments in 3D (Hugo would be my #1 pick) but nothing to rival the box office success of Avatar.
But now we have The Hobbit, a film that is destined to be a big box office success despite the mixed critical reviews. (The negatives are the thinness of the story for a three film series and some dislike of HFR; all accounts have it using 3D well.) And it's the beginning of a three film franchise. And it shows off a new technology, HFR 3D, that steps up the immersion level another notch. (Maybe too high.)
It doesn't look as though the all 3D all the time future that some people predicted will happen, but 3D remains a good option for the right kind of film and when it's done well. (The other big 3D release this year, The Avengers, is an example where 3D was not done well; it's a good film but 3D doesn't add much to it.)
But on any competent Linux distro, installing those other things happens nearly automatically (you might have to say yes to installing the dependencies) when you install Squirrelmail. So it's not really much of a pain. That's one area where Linux beats Windows like a drum.
Intel has been very supportive of the open source community. But their hardware simply don't measure up in graphic processing power. It's adequate for basic desktop use, but if you want serious 3D performance you'll have to look elsewhere.
NVidia has offered high quality proprietary drivers, although they sometimes lack features that are in the drivers for other operating systems. Notably, Optimus support for systems with integrated graphics plus an NVidia card is missing in Linux. That's where NVidia picks up the heavy lifting for graphics-intense applications and the integrated graphics do all the basic stuff. They continue driver support for many years after a product is discontinued. If their drivers work on your system and distro, NVidia is probably the most solid choice for 3D performance. However, they have not been supportive of open source, and so the non-proprietary drivers for NVidia hardware are terrible. When the company drops support of your hardware, you're pretty much SOL; you'll have to stop upgrading your Linux distro because the last driver that supports your card probably won't work with a new Linux version.
AMD/ATI is a mixed bag. The quality of their proprietary drivers has sometimes been lacking, and they drop support of products much sooner than NVidia does. (I've seen ATI hardware that is less than three years old lose driver support.) On the other hand, AMD has been reasonably cooperative with the open source community, so the open source driver for ATI is much better than its counterpart for NVidia. If you go with a new computer with AMD graphics, you can expect to have to use the proprietary driver for the first couple of years, and at some point soon after that to be forced to switch to the open source driver. The short duration of manufacturer driver support for AMD/ATI products makes them hard to recommend for any Linux system that is also going to run Windows.
People who spend their day on the telephone obviously still need a desk phone; if your job is in sales or marketing I don't expect your phone to go away. If you do a grubby job on a factory floor your rugged wall phone still makes sense. And if you work in a call center, duh!
Cell phones and/or VOIP make more sense for the rest of the workforce. For the typical knowledge worker the first question should be whether you even need to talk to them (a highly disruptive method of communication because of the need to drop everything to answer) at all, or whether you'd do better to use some form of text communication.
Some US cell carriers block international calling unless you ask them to enable it. (Sprint is one.) It's mostly to protect clueless users from making expensive international calls and running up a large bill; calls to some countries can cost dollars per minute though most are far less costly.
Prepaid plans are another story. Many of them have no international calling capability at all, and you almost always have to contact the company to enable it if the company offers the service. There usually is no fee for just enabling international calling, though some companies offer add-on plans that give you lower international rates but charge a monthly fee.
International roaming (using your US plan from outside the US) from any US cell carrier is ludicrously expensive. If you plan to do any significant amount of calling from outside the US it's generally better to carry an unlocked GSM phone and buy a local SIM for the place or places you visit. One exception is Canada; some carriers offer reasonably priced roaming plans there.
IDE CD and DVD drives remained common for a while after SATA hard drives were the rule; motherboards cut down to one IDE port but kept the one so you could connect them. For some reason, SATA optical drives were slower to appear. I still have some IDE optical drives (or PATA if you prefer the newer name) in a couple of my computers; they still work so why replace them? (Especially the 16x DVD burner; newer ones are slightly faster but you're hitting diminishing returns at that point.) And my email server still has IDE drives for the same reason.
I keep one retro-computer working that has lots of old hardware and software that lives on my electronics bench. The primary reason is that I own a device programmer that uses a parallel port interface, and is not supported in Windows versions newer than XP. So that's an old box with IDE hard and optical drives, Windows XP, and a motherboard that still has a parallel port. All of the other Windows computers have gone to Windows 7.
My favorite keyboard ever was an NMB keyboard from the early 90s. Sadly, it finally stopped working or else I'd still be using it. It was a PS/2 interface keyboard; many current computers still have a port for those (though the current trend seems to be to have only one port instead of two, so you can have a legacy keyboard OR mouse but not both) or you can use a USB adapter.
Few people upgrade the OS on their PC, but people DO upgrade the OS on their Macs and Linux systems. Moving Windows to a yearly release cycle is an attempt to move users to more Mac-like behavior and get them to buy the annual upgrades. So the fact that the upgrades will cost money is relevant; Microsoft will presumably move to a lower price for the upgrades (just as Apple has done over time) because few users would pay the current list price every year. The discount on Windows 8 upgrades through January shows that Microsoft is testing the waters and this might well be their permanent model; a low price for a limited time to get people to upgrade right away.
It's not that easy to fake a coin. Most vending machines don't just rely on weight. They also check the magnetic properties of the coins, which means that counterfeits that aren't made of metals similar to the real coins won't work. That's one reason we ended up with the sandwich of metals used in the dime and quarter; that combination made them react similarly to the silver coins they replaced, as well as being the right weight and color.
The few machines that take pennies aren't very fussy (exception: coin counting machines like the Coinstar things found in some supermarkets) so we were able to replace copper pennies with plated zinc pennies without a fuss. No sane person is going to counterfeit pennies in any case; the metal value of them exceeds their face value.
A lot of them already take $1 coins. At least here in Boston they do. The fact that the Charlie Card vending machines (fare cards for public transit) give out $1 coins as change may be a factor.
Vending machine companies will love this switch because it will allow most of them to get rid of those fussy, unreliable bill slots. The ones selling high-value items (like those Charlie Card machines) will still need them for larger bills.
CPU power has continued to advance at a similar pace; it's just that a lot of it has come in ways that are less noticeable to users. More cores are a boon for servers and to the serious multitasker but don't speed up typical desktop tasks much.
Another factor is the switch from desktop to laptop systems, and now from laptop systems to more portable lightweight laptops. Although the computing power of high-end systems continues to increase, the power of the AVERAGE computer has stagnated. Today's laptop is much faster than a five year old laptop, but comparable to a five year old desktop. Today's Ultrabook with a ULV processor is comparable to a five year old heavyweight laptop.
Because of this change, software makers haven't been able to fully target the increased power of those high end desktops because they can't count on a typical user having that level of power. Professional applications like video editors and Photoshop are exceptions; they continue to push the envelope with full multithreading and GPU-assisted rendering, making them rare examples of applications that can actually take advantage of a current high-end desktop system.
True for current ARM chips; they top out at the low end of laptop performance levels. Possibly not true for the upcoming 64-bit ARM chips, especially the Cortex-A57, but we won't know until they get shipped in 2014.
It's not going to happen. The problem is that there just aren't enough good mobile developers and enough developer mindshare to go around. Lots of developers don't even want to support TWO platforms, let alone three or four or five. There have never been more than two successful mobile platforms at once. First we had Palm and BlackBerry. iOS came along and Palm faded away. Then Android appeared and BlackBerry declined. But I don't see any of the current contenders displacing either iOS or Android; they're just too entrenched. Any new OS would have to be dramatically better in some important way, and neither BlackBerry 10 nor Windows Phone 8 is.
What about Symbian, you ask? Lots of Symbian phones were sold, but although it was theoretically a smartphone platform its success wasn't based on its smartness. The vast majority of Symbian devices were sold as feature phones with no real app capability, and the ones that could meaningfully use apps never got any significant number of them to choose from.
There is one possible scenario for limited success of BlackBerry 10 or Windows Phone, though the latter is not pursuing it. That would be to ignore the consumer market, which is a lost cause anyway, and double down on enterprise applications and try to become the preferred device of IT departments. The BYOD trend means it will be a challenge to turn this into sales, but it could at least get them a place in large corporations that are big on centralized IT and in security-conscious organizations. This would mean things like giving the mobile a top-class application for email, group calendar, and other enterprise uses, rolling VPN and full remote management support into the OS, including a remote desktop client, and so forth. It would also mean including enhanced security features like biometric ID into the handsets.
I am skeptical about the prospects for Mozilla Phone and Ubuntu Phone for the same reasons. It appears, however, that both of those platforms are going to make their push primarily in the third world, where the lower cost of those platforms might matter enough for one of them to gain some traction. It's not just that the software is zero-cost - so is Android - but that they're intended to offer acceptable performance on low-spec devices. The vast majority of apps for the two first-world platforms will likely never get to these platforms, though the big-name apps will.
If it's that old I'm guessing it was a 512MB drive, not 512GB.
They already did make the system builder license more transparent. (The new terms only apply to Windows 8, not to earlier versions, though they might apply to copies of Windows 8 Professional where the downgrade right to Windows 7 or Vista is exercised.) It now contains the Personal Use License clause: System Builder product may be used: As the operating system on a PC you build for personal use. As an operating system running either on a local virtual machine or as an additional operating system in a separate partition. System Builder product may not be used: As an upgrade license for an existing underlying Windows operating system. To legalize a non-genuine Windows operating system. To license more than five copies of the software (in total) for commercial use. That "more than five copies" clause is part of the Personal Use License and applies to commercial use by the builder. It does not preclude selling more than five commercial use copies to a customer. The old "full retail license" - the really expensive version of Windows - is essentially dead and Microsoft should discontinue it. People buying computers from major manufacturers will, as they have for years, get OEM licenses. People building their own computers and small boutique computer builders will use the System Builder version. All but the tiniest corporations will use one of Microsoft's volume licensing programs, which start at five seats. There isn't a market niche remaining for the old full license.
Some Apple commercials are also about cool and not about what the device does. Most iPod advertising falls in that category, including the recent one with bouncing iPods.
You're not stuck with 128Kbps for most music. Apple and Amazon both currently sell music at 256Kbps or thereabouts. (Apple uses AAC and Amazon uses MP3; at that bitrate the difference is negligible.) Those two stores offer nearly all the recorded music that is available in the US in any form. eMusic.com uses bit rates ranging from 192Kbps to 320Kbps (and uses the superior LAME encoder) for most music. (A few things are only available at 128Kbps.) Other smaller sellers of downloads also offer higher rates now. It's hard to do even better, but in some cases you can. HDTracks.com offers uncompressed FLAC downloads of music, much of it in high-resolution forms. (A few albums are even available as 24/192 or 24/176.4 downloads; the former typically means that the remastering was originally done for DVD-Audio and the latter was for SACD.) Some artists sell FLAC downloads, as do a few small labels.
Er, no. 200 channels of 1080p is 50 channels of 4K, barring adoption of a significantly more efficient codec than MPEG-4. 200 channels of SD is at most 20 channels of 4K, allowing for the fact that the SD channel is probably less efficiently encoded in MPEG-2.
For a while, WGBH (PBS) was broadcasting the same content in both SD and HD. When viewed using a digital converter on an old TV, the HD version looked better even after downconversion to 480i by the converter. Recently they switched the SD subchannel to different content; the fact that the SD version was useless is probably the reason.
That's because the sports bar is showing all the sports from cable. The least they could do is use a broadcast source when availalble (sports bars carry lots of content that isn't broadcast); it usually looks much better as most broadcasters are devoting 12-15Mbps to their main channel.
Making cars more efficient is a desirable result. Having the owners of the efficient cars pay less tax is a good way to get that result to happen. More efficient cars are also, on average, smaller and lighter so they cause less wear to the roads.
Going to a mileage-based tax is a bad idea because it removes that incentive. If tax revenue falls too much, raise the tax rate to compensate. Yes, it will punish SUV owners - as it should.
At Harvard Extension nearly all classes are 4 credits. (There are some 2 credit language classes that I did not take.) The math I gave was accurate for that school; other schools vary.
I took some summer classes. My experience that taking one summer class was about the same total intensity as taking two spring or fall classes. The courseload for CS classes was higher than for non-CS classes (in part because most CS classes at Harvard Extension are taught at the graduate level, though you can apply them to an undergraduate degree if you like) so it might be better to take non-CS subjects in the summer.
It's also true that the course load of classes varies wildly. My own boundary case, Computer Architecture, was a full time occupation in itself. That class involved designing a simple CPU architecture, writing an emulator and an assembler for it so you could create programs for it, then writing a VHDL implementation to create a hardware implementation. All in twelve weeks.
There is some truth to your comment on giving up personal life. Fortunately for me, school itself was a rewarding form of personal life, so the limitations on other social existence didn't hurt as much as they might have.
Ten years is a bit of an overestimate for an undergraduate degree. Where I went to school (Harvard Extension, the continuing education division of Harvard University), classes are 4 credits each and you need 128 credits to graduate. If you bring in no credits at all you need to take 32 classes for a degree. 2 per semester, 4 per year, you'll need eight years. If you add one summer class each year you get it down to just over six years. Hint: if you do it there take only one CS class each semester! The course load in the CS classes is usually heavier than in the non-CS classes you will need for a degree. For a Master's degree there you need 12 classes. Master's classes are tougher and they'll all be in CS so you're best off only taking one each semester, which will take six years or four years if you add summers.
Sometimes private schools have useful offerings as well. Here in the Boston area, both Harvard and BU offer evening computing classes and have degree options; at Harvard you can take a lot (but not all) of the classes as distance learning and do most of the work on your own schedule. (If the class has exams rather than or in addition to projects you have to show up in person for those.)
Computer science and software engineering are very different things. The former is largely about theory and is often heavily mathematical. The latter is about the practical things that the CMU student didn't get. Many of the best known CS programs have the same weakness; courses on the practicalities of programming are either not required or not even offered.
I knew you could...
When your standard of comparison is the #1 film of all time, everything is going to be downhill. There have been other notable artistic accomplishments in 3D (Hugo would be my #1 pick) but nothing to rival the box office success of Avatar.
But now we have The Hobbit, a film that is destined to be a big box office success despite the mixed critical reviews. (The negatives are the thinness of the story for a three film series and some dislike of HFR; all accounts have it using 3D well.) And it's the beginning of a three film franchise. And it shows off a new technology, HFR 3D, that steps up the immersion level another notch. (Maybe too high.)
It doesn't look as though the all 3D all the time future that some people predicted will happen, but 3D remains a good option for the right kind of film and when it's done well. (The other big 3D release this year, The Avengers, is an example where 3D was not done well; it's a good film but 3D doesn't add much to it.)
The poor internet speeds and low data caps in Australia will keep physical DVD rental alive for a while.
But on any competent Linux distro, installing those other things happens nearly automatically (you might have to say yes to installing the dependencies) when you install Squirrelmail. So it's not really much of a pain. That's one area where Linux beats Windows like a drum.
Intel has been very supportive of the open source community. But their hardware simply don't measure up in graphic processing power. It's adequate for basic desktop use, but if you want serious 3D performance you'll have to look elsewhere.
NVidia has offered high quality proprietary drivers, although they sometimes lack features that are in the drivers for other operating systems. Notably, Optimus support for systems with integrated graphics plus an NVidia card is missing in Linux. That's where NVidia picks up the heavy lifting for graphics-intense applications and the integrated graphics do all the basic stuff. They continue driver support for many years after a product is discontinued. If their drivers work on your system and distro, NVidia is probably the most solid choice for 3D performance. However, they have not been supportive of open source, and so the non-proprietary drivers for NVidia hardware are terrible. When the company drops support of your hardware, you're pretty much SOL; you'll have to stop upgrading your Linux distro because the last driver that supports your card probably won't work with a new Linux version.
AMD/ATI is a mixed bag. The quality of their proprietary drivers has sometimes been lacking, and they drop support of products much sooner than NVidia does. (I've seen ATI hardware that is less than three years old lose driver support.) On the other hand, AMD has been reasonably cooperative with the open source community, so the open source driver for ATI is much better than its counterpart for NVidia. If you go with a new computer with AMD graphics, you can expect to have to use the proprietary driver for the first couple of years, and at some point soon after that to be forced to switch to the open source driver. The short duration of manufacturer driver support for AMD/ATI products makes them hard to recommend for any Linux system that is also going to run Windows.
People who spend their day on the telephone obviously still need a desk phone; if your job is in sales or marketing I don't expect your phone to go away. If you do a grubby job on a factory floor your rugged wall phone still makes sense. And if you work in a call center, duh!
Cell phones and/or VOIP make more sense for the rest of the workforce. For the typical knowledge worker the first question should be whether you even need to talk to them (a highly disruptive method of communication because of the need to drop everything to answer) at all, or whether you'd do better to use some form of text communication.
Some US cell carriers block international calling unless you ask them to enable it. (Sprint is one.) It's mostly to protect clueless users from making expensive international calls and running up a large bill; calls to some countries can cost dollars per minute though most are far less costly.
Prepaid plans are another story. Many of them have no international calling capability at all, and you almost always have to contact the company to enable it if the company offers the service. There usually is no fee for just enabling international calling, though some companies offer add-on plans that give you lower international rates but charge a monthly fee.
International roaming (using your US plan from outside the US) from any US cell carrier is ludicrously expensive. If you plan to do any significant amount of calling from outside the US it's generally better to carry an unlocked GSM phone and buy a local SIM for the place or places you visit. One exception is Canada; some carriers offer reasonably priced roaming plans there.
IDE CD and DVD drives remained common for a while after SATA hard drives were the rule; motherboards cut down to one IDE port but kept the one so you could connect them. For some reason, SATA optical drives were slower to appear. I still have some IDE optical drives (or PATA if you prefer the newer name) in a couple of my computers; they still work so why replace them? (Especially the 16x DVD burner; newer ones are slightly faster but you're hitting diminishing returns at that point.) And my email server still has IDE drives for the same reason. I keep one retro-computer working that has lots of old hardware and software that lives on my electronics bench. The primary reason is that I own a device programmer that uses a parallel port interface, and is not supported in Windows versions newer than XP. So that's an old box with IDE hard and optical drives, Windows XP, and a motherboard that still has a parallel port. All of the other Windows computers have gone to Windows 7. My favorite keyboard ever was an NMB keyboard from the early 90s. Sadly, it finally stopped working or else I'd still be using it. It was a PS/2 interface keyboard; many current computers still have a port for those (though the current trend seems to be to have only one port instead of two, so you can have a legacy keyboard OR mouse but not both) or you can use a USB adapter.
Few people upgrade the OS on their PC, but people DO upgrade the OS on their Macs and Linux systems. Moving Windows to a yearly release cycle is an attempt to move users to more Mac-like behavior and get them to buy the annual upgrades. So the fact that the upgrades will cost money is relevant; Microsoft will presumably move to a lower price for the upgrades (just as Apple has done over time) because few users would pay the current list price every year. The discount on Windows 8 upgrades through January shows that Microsoft is testing the waters and this might well be their permanent model; a low price for a limited time to get people to upgrade right away.
It's not that easy to fake a coin. Most vending machines don't just rely on weight. They also check the magnetic properties of the coins, which means that counterfeits that aren't made of metals similar to the real coins won't work. That's one reason we ended up with the sandwich of metals used in the dime and quarter; that combination made them react similarly to the silver coins they replaced, as well as being the right weight and color.
The few machines that take pennies aren't very fussy (exception: coin counting machines like the Coinstar things found in some supermarkets) so we were able to replace copper pennies with plated zinc pennies without a fuss. No sane person is going to counterfeit pennies in any case; the metal value of them exceeds their face value.
A lot of them already take $1 coins. At least here in Boston they do. The fact that the Charlie Card vending machines (fare cards for public transit) give out $1 coins as change may be a factor.
Vending machine companies will love this switch because it will allow most of them to get rid of those fussy, unreliable bill slots. The ones selling high-value items (like those Charlie Card machines) will still need them for larger bills.
CPU power has continued to advance at a similar pace; it's just that a lot of it has come in ways that are less noticeable to users. More cores are a boon for servers and to the serious multitasker but don't speed up typical desktop tasks much.
Another factor is the switch from desktop to laptop systems, and now from laptop systems to more portable lightweight laptops. Although the computing power of high-end systems continues to increase, the power of the AVERAGE computer has stagnated. Today's laptop is much faster than a five year old laptop, but comparable to a five year old desktop. Today's Ultrabook with a ULV processor is comparable to a five year old heavyweight laptop.
Because of this change, software makers haven't been able to fully target the increased power of those high end desktops because they can't count on a typical user having that level of power. Professional applications like video editors and Photoshop are exceptions; they continue to push the envelope with full multithreading and GPU-assisted rendering, making them rare examples of applications that can actually take advantage of a current high-end desktop system.
True for current ARM chips; they top out at the low end of laptop performance levels. Possibly not true for the upcoming 64-bit ARM chips, especially the Cortex-A57, but we won't know until they get shipped in 2014.