True, a larger concern is that even if existing applications CAN run on x86 with just a recompile, you still need all those developers to recompile their apps. I've got more than a few apps on my phone that are no longer actively updated. I'd lose them all in a switch to x86.
Fundamentally, it comes back to the basic question: why should Apple switch? ARM's licensing fees are cheap, Apple makes their own SoCs (and they've consistently moved in the direction of doing *more* of the design themselves, rather than less) so they'd lose all control over that if they moved to an Intel SoC, and their existing platform all relies upon it.
Yes, they could switch, but Intel needs to do better than "just as good as". They need to be "better than". It's doubtful Intel could compete on price, because ARM is just licensing the design to Apple. It would have to come down to power and performance, and they'd have to be substantially better than what Apple could get out of ARM or there'd be no benefit to switching. Especially if Apple would be required to give up control of their SoC by replacing it with an entirely outsourced solution (an Intel SoC).
If anything, I think Apple is much less likely to switch to x86 for their mobile devices than an Android vendor, because with a few exceptions, most Android vendors are not nearly as vertically integrated. Most of them don't make their own SoCs, and even those that do like Samsung or Ericsson don't even use their own SoCs in some of their phones; I'm not sure if any Sony Ericsson phone has ever used an Ericsson SoC, and Samsung makes a bunch of devices like the Galaxy S II HD and the Galaxy Nexus that are using Qualcomm's Snapdragon rather than the Exynos.
Samsung expects to get that out in 2012Q2, which would be in less than 6 months. Inside of 2012, we should also see 28nm Cortex A15, as well as quad-core parts.
So, the problem is that Intel is touting these performance advantages for their next-gen part, but are comparing it to current-gen ARM stuff. An accurate comparison would be between the next-gen Atom and the next-gen ARM, since that's what it'll be competing with. Merely providing a comparable part to ARM's stuff isn't good enough, Intel has to do something BETTER (or cheaper), or there will be no reason for anybody to switch.
If Intel can offer you a part with the same cost, power, and performance as ARM, but you need to take a bunch of extra effort to switch to the new platform, why would you ever switch?
Unless that app was written with the Android NDK, or links to a library built with the NDK. Then it's running native code, and is not platform independent.
Saying that iOS runs on x86 because of the iOS simulator is like saying that Windows runs on PowerPC because of WINE. It's just an implementation of the iOS APIs, not the OS itself. The simulator just makes a set of iOS-like APIs available to Mac applications.
It's not a pre-programmed function, though. For that query, Siri resorts to Wolfram Alpha, which does a good job in solving that problem (that of tips) as well as a lot of other cases. And for the vast majority of these things, Google doesn't do anywhere near as good a job as Wolfram Alpha. Google can do some stuff with Google Calculator, and it can answer some very narrow questions with actual data, but it doesn't have anywhere near the same degree of syntax flexibility as Alpha, and the data available is much more specific.
Ask Google the population of a country, and it'll probably tell you. But if you want the population of a province/state/city, it's hit or miss if it'll know, while Wolfram Alpha has got pretty much every province/state/city you could think of. If I ask Wolfram Alpha the population of my 30k population hometown in Quebec, and spell the name wrong, it still gives me the right answer. Google can't even tell me the population of the entire province.
My point is that basically, Google has no equivalent to Wolfram Alpha. Luckily, the solution is simple: they just need to license Wolfram Alpha like Apple did. This does lead to one extra problem, though; Siri or Majel needs to know WHEN it should consult Wolfram Alpha for an answer. Using Siri, I've noticed that it doesn't always consult Wolfram Alpha even when I want it to. So there's a lot of work that goes into defining that delineation.
Right, but that just goes to show the limitations of Google Calculator versus Wolfram Alpha: Google Calculator is far more sensitive to syntax, and doesn't provide as much information. For example, it doesn't do the sum for me.
The logical answer would be that Google's competitor to Siri should just license Wolfram Alpha too.
It's more than just about search, though; I almost never use Siri for doing web searches, since it requires unlocking the phone (with a password) while most other Siri activities don't.
Apple's approach was to bring in a few different data sources for different types of queries. Yelp for restaurant reviews, for example, or Wolfram Alpha for information or calculation queries.
Google does have direct access to a lot of this information; they've got Google Local (now merged with Maps) to replace Yelp. But they don't really have a direct equivalent to Wolfram Alpha. I mean, there's Google Calculator, and a few other searches do seem to directly return information, but all of that combined pales in comparison to Wolfram Alpha.
Of course, Apple doesn't own Wolfram. Google can just make their own deal to get Majel integration with Wolfram Alpha.
In spite of the thought you can simply train any random group of primates to build this stuff, the actual construction of rockets is a highly skilled task that frankly few aerospace engineers could actually accomplish without a whole lot of experience.
SpaceX managed to do that, plus all the R&D to actually put payload into orbit, for less than the cost of a single shuttle launch.
Of course we live in such a world. If you want to launch a new communication satellite to GTO, would you rather pay $50 million to SpaceX for part of the payload of a Falcon Heavy launch, or $900 million for two shuttle launches?
Even if SpaceX didn't exist, you'd never use the space shuttle for launching payloads into orbit, because even the pre-SpaceX alternatives are enormously cheaper for getting your stuff into orbit. SpaceX isn't competing with the shuttle, that was never the intent; the shuttle isn't a super heavy launch platform anyhow. A Falcon Heavy carries five times as much payload to GTO as a shuttle while costing a quarter as much to launch. The intent of SpaceX is instead to compete with existing commercial launch systems like the Delta IV, the Ariane 5, etc.
For civilian rocketry, you need a rocket that can be reliably launched, but it doesn't have to sit on the launch pad all that long after being prepared, and can use volatile cryogenic fuels like liquid oxygen that an ICBM really doesn't use very well (even though some countries still use it for ballistic missiles).
I'm not sure if you'd qualify liquid oxygen as a fuel. It's a propellant, sure, but it's used to oxidize the fuel rather than as fuel itself. SpaceX uses Kerosene as fuel, with liquid oxygen as the oxidizer.
I don't think any modern notebook can match 0.85 kg, but you can get a 13.3" notebook at 1.1kg (Toshibz Z830 for example), and I suspect we'll see 11" notebooks hitting that 0.85 kg target at some point.
Less size and weight can be achieved without a netbook. Ultrabooks actually tend to be even better than that; they're thinner and lighter than netbooks, but feature larger screens and more performance. The downside is cost; $900-1000 is the cheapest you'll find them.
Re:next we'll hear that Dell is in trouble...
on
Dell Ditches Netbooks
·
· Score: 4, Informative
That's what the "ultrabook" class is trying to address. They (and devices like them that predate them, like the Samsung Series 9 or the Macbook Air) are largely replacing netbooks in terms of portability. They're typically 11 or 13 inches, and tend to weight 2.5 to 3.5 pounds or so. They're often lighter than a netbook, but have a much larger screen. They also tend to have proper dual-core processors, although they're the ULV (Ultra Low Voltage) kind which means they're clocked lower. Still, a dual-core i5 is still pretty decent, even if it's ULV.
The downside is price. There are tons of models available for under a grand, but some people want to get the portability for much less. We're not there yet, they still cost too much to make (all ultrabooks use SSDs, so the trick is the cost of enough flash to make that practical), but the cost will probably come down slowly over time.
Or Phalanger at http://www.php-compiler.net/ which is a GPL project that compiles to CIL bytecode, which is then run through.NET or Mono in a JIT fashion. It's a proper.NET language, so you get access to the.NET framework as well as the core PHP functions and default extensions. Unfortunately, some of the non-default extensions that they've ported (like gd2 or memcached) are not available in the GPL version.
Haven't they sort of? Google plans to use LLVM to make it portable (an intermediate form that can be translated to different instruction sets as needed), Microsoft uses CIL with Silverlight (which as much as the Linux community might not like it, is opensource and supported by the two top desktop platforms), Java has their own thing going...
If it's not DRM encrypted, there's software like Calibre that will convert between all the different formats. DRM-free eBook formats haven't been an issue for years, I don't know why everybody is so obsessed with ePub on the kindle. As the OP, I've been reading non-Amazon DRM-free ePubs for ages.
There are a lot of stuff that's annoying about the kindle, the format support is not one of them. Not being able to set my own screensaver image on my non-advertising kindle is a bigger annoyance to me.
Well, Kodak didn't make any digital cameras of any kind themselves, so far as I can tell, but if you're asking in a general sense, I don't think EVFs (electronic viewfinders) are sufficiently advanced to completely replace DSLRs.
Sony's latest cameras like the SLT-A55 are technically not DSLRs, as they eschew the optical viewfinder for an EVF, but they have some shortcomings... I'm not an expert, but I'll do my best to explain.
In terms of "why viewfinder in general", that one is easy; LCD panels like you'd find on a point & shoot are useless outdoors. This is starting to change (mostly driven by smartphone development), but there's still an enormous loss of fidelity even if you can tell what the screen is showing; the image on the screen, under bright sunlight, looks nothing like what the camera will capture. There's also probably various other issues about the relative amount of your vision the image takes up when looking through a viewfinder versus at an LCD panel on the back of a camera. Certainly nobody has ever made a camera with a rear LCD sufficiently high res to properly manually focus, although that's not a technology limitation (probably a cost issue).
In terms of why optical viewfinders rather than electronic (as in, why not just use the sensor), there are a few answers. For one thing, higher end cameras don't use the sensor image for autofocus, so they need to redirect light to the autofocus sensors anyhow, and at that point you've done most of the work for implementing a DSLR anyhow. There are cameras out there that use EVFs (electronic viewfinders), but without dedicated autofocus sensors, and their autofocus performance isn't even remotely as good (or as fast), since it can only do contrast-based AF and not phase-based AF. The other answer is that EVF technology just wasn't there yet. EVFs were invariably laggy (the image you see was lagging behind), and too low-res to let the photographer accurately manually focus. EVFs have improved a lot, but the Sony SLT-A55 is the first camera that manages to get around these issues. I mention the A55 not as some sort of advertisement for Sony (I prefer Canon cameras myself), but because they're legitimately the first company to actually pull it off. But the fact that the latest camera in that lineup is the very first one to pull it off kind of goes to show how new the technology to affordably pull it off is. Basically, the A55 uses the image from the sensor for the EVF, and a translucent mirror to simultaneously direct 30% of the light towards the autofocus sensors. You'll notice, though, that the A55 looks like a DSLR even though it's not; that's because the mechanics are similar, even if the viewfinder isn't optical.
So, the A55 is, I believe, the first camera to get perceptibly zero latency on the EVF, still have a dedicated autofocus array, and sufficiently high resolution on the EVF for manually focusing (although it's still not at the level where it's indistinguishable from reality). But there are limitations too. For one thing, the transparent mirror that redirects 30% of the light to the autofocus array is permanently in place. Your sensor will always get 70% less light than a comparable DSLR, and there are obviously consequences to that. For another thing, it's not quite perfect yet; there's no lag in the image on the EVF, but there's a big lag about it detecting if you're looking through the EVF, switching on the display (IIRC reviews pegged it at half a second or a second or so, enough to be annoying). And also, Sony probably has patents on the transparent mirror thing, which could impede other companies from doing the same thing.
In the end, the technical challenges about replacing an optical viewfinder with an electronic one make it a much harder challenge than you seem to think, and most camera companies obviously don't think it's yet worth the effort. It's certainly much harder to make than a DSLR. I think we'll slowly see more cameras moving in that direction, but it's only just becoming possible to do so.
That patent describes putting together a few existing components with no particularly original innovation, it does not appear to be a description of Kodak manufacturing and selling a DSLR camera (which I'd argue "make the first DSLR" would entail). Heck, the patent would seem to be invalid due to prior art, although at least some of the prior art may have been classified.
Kodak didn't invent the CCD (That was AT&T in 1969), they didn't first use it for imaging (That was also AT&T, 1971), they didn't first use it for capturing an actual picture (that was Fairchild in 1974 led by future failed Apple CEO Gil Amelio), and the use of CCD digital cameras predates their patent (U.S. NRO recon satellite, 1976). By the time their 1978 patent came around (which didn't result in any actual camera product), it had all been done before.
Regardless of what it competed with (and MiniDisc only hit the market six years before flash MP3 players, maybe four or five years for properly portable MiniDisc), it still failed. Yes, it was more successful in Asia (a friend from Singapore brought his with him while studying here), particularly Japan, but even there they were never a huge success since the hardware never got cheap (even in Japan) and there were limited albums available on MD.
I'm not saying they didn't have the occasional success, only that they lost the vast majority of them. Sony wasn't really the primary inventor of the CD either; Philips was, and Philips also had a hand in the laserdisc standard that gave birth to CDs.
The quality difference wasn't terribly obvious; they were both doing 250 lines of resolution. And Sony never fixed the runtime issue; they got it up to 5 hours only by reducing the quality significantly (much like EP/SLP on VHS, which got it up to almost 11 hours). There were other shortcomings; Sony couldn't make beta small enough for compact tapes, so they couldn't use beta in smaller camcorders like VHS-C. Then there was the whole "it's our format so we can charge more" issue that hurt it, but really, I'm not so sure Beta was so much better. It was, in the most important way, a technologically inferior solution. The runtime was too short for home movies, too short to record sports games...
Re:Horse and buggy companies didn't make it either
on
The Rise and Fall of Kodak
·
· Score: 3, Interesting
Right, but at that point they should have been making their own cameras instead of just the sensor. They had a lock on the entire camera market at one point, but for some reason it seems that they never actually made their own digital cameras, just rebranded or added on to other companies'.
The make-your-own-camera-module thing became more important as cellphones started getting cameras integrated (even when they were crap, a lot of phones had them, and there was money there even then). It wasn't (and isn't) realistic for Kodak to have made their own cellphone, but they could have gotten a chunk of the camera module market. At this point, that's probably even a much bigger market than the rest of the camera market combined; every cellphone, tablet, handheld game console sold, they all have camera modules, and Kodak isn't the one making them. Sony makes a lot of them, even for their competitors. And I've no idea who makes the camera modules in the 3DS, but it's got *THREE* of the things. More and more cellphones these days have at least two cameras...
Slashdot Mirror
True, a larger concern is that even if existing applications CAN run on x86 with just a recompile, you still need all those developers to recompile their apps. I've got more than a few apps on my phone that are no longer actively updated. I'd lose them all in a switch to x86.
Fundamentally, it comes back to the basic question: why should Apple switch? ARM's licensing fees are cheap, Apple makes their own SoCs (and they've consistently moved in the direction of doing *more* of the design themselves, rather than less) so they'd lose all control over that if they moved to an Intel SoC, and their existing platform all relies upon it.
Yes, they could switch, but Intel needs to do better than "just as good as". They need to be "better than". It's doubtful Intel could compete on price, because ARM is just licensing the design to Apple. It would have to come down to power and performance, and they'd have to be substantially better than what Apple could get out of ARM or there'd be no benefit to switching. Especially if Apple would be required to give up control of their SoC by replacing it with an entirely outsourced solution (an Intel SoC).
If anything, I think Apple is much less likely to switch to x86 for their mobile devices than an Android vendor, because with a few exceptions, most Android vendors are not nearly as vertically integrated. Most of them don't make their own SoCs, and even those that do like Samsung or Ericsson don't even use their own SoCs in some of their phones; I'm not sure if any Sony Ericsson phone has ever used an Ericsson SoC, and Samsung makes a bunch of devices like the Galaxy S II HD and the Galaxy Nexus that are using Qualcomm's Snapdragon rather than the Exynos.
Today: 1.5GHz 45nm dual-core Cortex A9 or 1.3GHz 40nm quad-core Cortex A9
3-6 months: 2.0GHz 32nm Cortex A15.
Samsung expects to get that out in 2012Q2, which would be in less than 6 months. Inside of 2012, we should also see 28nm Cortex A15, as well as quad-core parts.
So, the problem is that Intel is touting these performance advantages for their next-gen part, but are comparing it to current-gen ARM stuff. An accurate comparison would be between the next-gen Atom and the next-gen ARM, since that's what it'll be competing with. Merely providing a comparable part to ARM's stuff isn't good enough, Intel has to do something BETTER (or cheaper), or there will be no reason for anybody to switch.
If Intel can offer you a part with the same cost, power, and performance as ARM, but you need to take a bunch of extra effort to switch to the new platform, why would you ever switch?
Unless that app was written with the Android NDK, or links to a library built with the NDK. Then it's running native code, and is not platform independent.
Saying that iOS runs on x86 because of the iOS simulator is like saying that Windows runs on PowerPC because of WINE. It's just an implementation of the iOS APIs, not the OS itself. The simulator just makes a set of iOS-like APIs available to Mac applications.
It's not a pre-programmed function, though. For that query, Siri resorts to Wolfram Alpha, which does a good job in solving that problem (that of tips) as well as a lot of other cases. And for the vast majority of these things, Google doesn't do anywhere near as good a job as Wolfram Alpha. Google can do some stuff with Google Calculator, and it can answer some very narrow questions with actual data, but it doesn't have anywhere near the same degree of syntax flexibility as Alpha, and the data available is much more specific.
Ask Google the population of a country, and it'll probably tell you. But if you want the population of a province/state/city, it's hit or miss if it'll know, while Wolfram Alpha has got pretty much every province/state/city you could think of. If I ask Wolfram Alpha the population of my 30k population hometown in Quebec, and spell the name wrong, it still gives me the right answer. Google can't even tell me the population of the entire province.
My point is that basically, Google has no equivalent to Wolfram Alpha. Luckily, the solution is simple: they just need to license Wolfram Alpha like Apple did. This does lead to one extra problem, though; Siri or Majel needs to know WHEN it should consult Wolfram Alpha for an answer. Using Siri, I've noticed that it doesn't always consult Wolfram Alpha even when I want it to. So there's a lot of work that goes into defining that delineation.
Right, but that just goes to show the limitations of Google Calculator versus Wolfram Alpha: Google Calculator is far more sensitive to syntax, and doesn't provide as much information. For example, it doesn't do the sum for me.
The logical answer would be that Google's competitor to Siri should just license Wolfram Alpha too.
Sure, here's a common one I use it for:
"15% tip on $14.56"
Siri:
tip percentage: 15%
amount without tip: C$ 14.56
amount of tip: C$ 2.18
amount with tip C$ 16.74
Google:
Walmart~Starbucks Via Instant Coffee 14ct. Only $14.56 *Two Flavors!
It's more than just about search, though; I almost never use Siri for doing web searches, since it requires unlocking the phone (with a password) while most other Siri activities don't.
Apple's approach was to bring in a few different data sources for different types of queries. Yelp for restaurant reviews, for example, or Wolfram Alpha for information or calculation queries.
Google does have direct access to a lot of this information; they've got Google Local (now merged with Maps) to replace Yelp. But they don't really have a direct equivalent to Wolfram Alpha. I mean, there's Google Calculator, and a few other searches do seem to directly return information, but all of that combined pales in comparison to Wolfram Alpha.
Of course, Apple doesn't own Wolfram. Google can just make their own deal to get Majel integration with Wolfram Alpha.
In spite of the thought you can simply train any random group of primates to build this stuff, the actual construction of rockets is a highly skilled task that frankly few aerospace engineers could actually accomplish without a whole lot of experience.
SpaceX managed to do that, plus all the R&D to actually put payload into orbit, for less than the cost of a single shuttle launch.
Of course we live in such a world. If you want to launch a new communication satellite to GTO, would you rather pay $50 million to SpaceX for part of the payload of a Falcon Heavy launch, or $900 million for two shuttle launches?
Even if SpaceX didn't exist, you'd never use the space shuttle for launching payloads into orbit, because even the pre-SpaceX alternatives are enormously cheaper for getting your stuff into orbit. SpaceX isn't competing with the shuttle, that was never the intent; the shuttle isn't a super heavy launch platform anyhow. A Falcon Heavy carries five times as much payload to GTO as a shuttle while costing a quarter as much to launch. The intent of SpaceX is instead to compete with existing commercial launch systems like the Delta IV, the Ariane 5, etc.
For civilian rocketry, you need a rocket that can be reliably launched, but it doesn't have to sit on the launch pad all that long after being prepared, and can use volatile cryogenic fuels like liquid oxygen that an ICBM really doesn't use very well (even though some countries still use it for ballistic missiles).
I'm not sure if you'd qualify liquid oxygen as a fuel. It's a propellant, sure, but it's used to oxidize the fuel rather than as fuel itself. SpaceX uses Kerosene as fuel, with liquid oxygen as the oxidizer.
I don't think any modern notebook can match 0.85 kg, but you can get a 13.3" notebook at 1.1kg (Toshibz Z830 for example), and I suspect we'll see 11" notebooks hitting that 0.85 kg target at some point.
Less size and weight can be achieved without a netbook. Ultrabooks actually tend to be even better than that; they're thinner and lighter than netbooks, but feature larger screens and more performance. The downside is cost; $900-1000 is the cheapest you'll find them.
That's what the "ultrabook" class is trying to address. They (and devices like them that predate them, like the Samsung Series 9 or the Macbook Air) are largely replacing netbooks in terms of portability. They're typically 11 or 13 inches, and tend to weight 2.5 to 3.5 pounds or so. They're often lighter than a netbook, but have a much larger screen. They also tend to have proper dual-core processors, although they're the ULV (Ultra Low Voltage) kind which means they're clocked lower. Still, a dual-core i5 is still pretty decent, even if it's ULV.
The downside is price. There are tons of models available for under a grand, but some people want to get the portability for much less. We're not there yet, they still cost too much to make (all ultrabooks use SSDs, so the trick is the cost of enough flash to make that practical), but the cost will probably come down slowly over time.
Or Phalanger at http://www.php-compiler.net/ which is a GPL project that compiles to CIL bytecode, which is then run through .NET or Mono in a JIT fashion. It's a proper .NET language, so you get access to the .NET framework as well as the core PHP functions and default extensions. Unfortunately, some of the non-default extensions that they've ported (like gd2 or memcached) are not available in the GPL version.
If you can natively execute LLVM after an initial translation step, why would you want to JIT it?
Truly, you have a dizzying intellect.
Haven't they sort of? Google plans to use LLVM to make it portable (an intermediate form that can be translated to different instruction sets as needed), Microsoft uses CIL with Silverlight (which as much as the Linux community might not like it, is opensource and supported by the two top desktop platforms), Java has their own thing going...
If it's not DRM encrypted, there's software like Calibre that will convert between all the different formats. DRM-free eBook formats haven't been an issue for years, I don't know why everybody is so obsessed with ePub on the kindle. As the OP, I've been reading non-Amazon DRM-free ePubs for ages.
There are a lot of stuff that's annoying about the kindle, the format support is not one of them. Not being able to set my own screensaver image on my non-advertising kindle is a bigger annoyance to me.
Well, Kodak didn't make any digital cameras of any kind themselves, so far as I can tell, but if you're asking in a general sense, I don't think EVFs (electronic viewfinders) are sufficiently advanced to completely replace DSLRs.
Sony's latest cameras like the SLT-A55 are technically not DSLRs, as they eschew the optical viewfinder for an EVF, but they have some shortcomings... I'm not an expert, but I'll do my best to explain.
In terms of "why viewfinder in general", that one is easy; LCD panels like you'd find on a point & shoot are useless outdoors. This is starting to change (mostly driven by smartphone development), but there's still an enormous loss of fidelity even if you can tell what the screen is showing; the image on the screen, under bright sunlight, looks nothing like what the camera will capture. There's also probably various other issues about the relative amount of your vision the image takes up when looking through a viewfinder versus at an LCD panel on the back of a camera. Certainly nobody has ever made a camera with a rear LCD sufficiently high res to properly manually focus, although that's not a technology limitation (probably a cost issue).
In terms of why optical viewfinders rather than electronic (as in, why not just use the sensor), there are a few answers. For one thing, higher end cameras don't use the sensor image for autofocus, so they need to redirect light to the autofocus sensors anyhow, and at that point you've done most of the work for implementing a DSLR anyhow. There are cameras out there that use EVFs (electronic viewfinders), but without dedicated autofocus sensors, and their autofocus performance isn't even remotely as good (or as fast), since it can only do contrast-based AF and not phase-based AF. The other answer is that EVF technology just wasn't there yet. EVFs were invariably laggy (the image you see was lagging behind), and too low-res to let the photographer accurately manually focus. EVFs have improved a lot, but the Sony SLT-A55 is the first camera that manages to get around these issues. I mention the A55 not as some sort of advertisement for Sony (I prefer Canon cameras myself), but because they're legitimately the first company to actually pull it off. But the fact that the latest camera in that lineup is the very first one to pull it off kind of goes to show how new the technology to affordably pull it off is. Basically, the A55 uses the image from the sensor for the EVF, and a translucent mirror to simultaneously direct 30% of the light towards the autofocus sensors. You'll notice, though, that the A55 looks like a DSLR even though it's not; that's because the mechanics are similar, even if the viewfinder isn't optical.
So, the A55 is, I believe, the first camera to get perceptibly zero latency on the EVF, still have a dedicated autofocus array, and sufficiently high resolution on the EVF for manually focusing (although it's still not at the level where it's indistinguishable from reality). But there are limitations too. For one thing, the transparent mirror that redirects 30% of the light to the autofocus array is permanently in place. Your sensor will always get 70% less light than a comparable DSLR, and there are obviously consequences to that. For another thing, it's not quite perfect yet; there's no lag in the image on the EVF, but there's a big lag about it detecting if you're looking through the EVF, switching on the display (IIRC reviews pegged it at half a second or a second or so, enough to be annoying). And also, Sony probably has patents on the transparent mirror thing, which could impede other companies from doing the same thing.
In the end, the technical challenges about replacing an optical viewfinder with an electronic one make it a much harder challenge than you seem to think, and most camera companies obviously don't think it's yet worth the effort. It's certainly much harder to make than a DSLR. I think we'll slowly see more cameras moving in that direction, but it's only just becoming possible to do so.
That patent describes putting together a few existing components with no particularly original innovation, it does not appear to be a description of Kodak manufacturing and selling a DSLR camera (which I'd argue "make the first DSLR" would entail). Heck, the patent would seem to be invalid due to prior art, although at least some of the prior art may have been classified.
Kodak didn't invent the CCD (That was AT&T in 1969), they didn't first use it for imaging (That was also AT&T, 1971), they didn't first use it for capturing an actual picture (that was Fairchild in 1974 led by future failed Apple CEO Gil Amelio), and the use of CCD digital cameras predates their patent (U.S. NRO recon satellite, 1976). By the time their 1978 patent came around (which didn't result in any actual camera product), it had all been done before.
Regardless of what it competed with (and MiniDisc only hit the market six years before flash MP3 players, maybe four or five years for properly portable MiniDisc), it still failed. Yes, it was more successful in Asia (a friend from Singapore brought his with him while studying here), particularly Japan, but even there they were never a huge success since the hardware never got cheap (even in Japan) and there were limited albums available on MD.
I'm not saying they didn't have the occasional success, only that they lost the vast majority of them. Sony wasn't really the primary inventor of the CD either; Philips was, and Philips also had a hand in the laserdisc standard that gave birth to CDs.
The quality difference wasn't terribly obvious; they were both doing 250 lines of resolution. And Sony never fixed the runtime issue; they got it up to 5 hours only by reducing the quality significantly (much like EP/SLP on VHS, which got it up to almost 11 hours). There were other shortcomings; Sony couldn't make beta small enough for compact tapes, so they couldn't use beta in smaller camcorders like VHS-C. Then there was the whole "it's our format so we can charge more" issue that hurt it, but really, I'm not so sure Beta was so much better. It was, in the most important way, a technologically inferior solution. The runtime was too short for home movies, too short to record sports games...
Right, but at that point they should have been making their own cameras instead of just the sensor. They had a lock on the entire camera market at one point, but for some reason it seems that they never actually made their own digital cameras, just rebranded or added on to other companies'.
The make-your-own-camera-module thing became more important as cellphones started getting cameras integrated (even when they were crap, a lot of phones had them, and there was money there even then). It wasn't (and isn't) realistic for Kodak to have made their own cellphone, but they could have gotten a chunk of the camera module market. At this point, that's probably even a much bigger market than the rest of the camera market combined; every cellphone, tablet, handheld game console sold, they all have camera modules, and Kodak isn't the one making them. Sony makes a lot of them, even for their competitors. And I've no idea who makes the camera modules in the 3DS, but it's got *THREE* of the things. More and more cellphones these days have at least two cameras...