Well, together Debian Ubuntu and Red Hat probably compromise the majority of Linux installs these days. If two large and well respected distros can fail in such basic ways, then it's reasonable to extrapolate that smaller and presumably less professional outfits will be even more flaky. Of course you can always find some Linux distro that has a perfect track record, but like I said above, usage counts. At some point if you want the word "Linux" to be meaningful you have to start talking about the bits actually in circulation.
I wrote an app which is now on the market. The good:
Java. OK, actually I hate Java. But I'd hate Objective-C a lot more. Implementing a simple crash reporter around my core logic was about 10 lines of Java code, and it works every time. Implementing the same thing in C++ or Objective-C would be.... non-trivial. No bother with heap corruptions, etc.
The whole design of activities and intents is quite well thought out. It seemed overly complicated at first but now I appreciate it a lot more. It's also very flexible, you aren't forced to use the infrastructure if you don't want to.
Really rich APIs. Background services, maps, multimedia, power management, package management, notifications.... even a face recognizer!
The market. I see a lot of people rag on the market and the comments system. Maybe I'm biased because my app has almost universally good reviews, but it's really nice to get that instant feedback about how you're doing. It's my experience that G1 owners (and there are apparently quite a lot) are ridiculously lenient. My app is extremely simple and could use a lot of extra features, yet I consistently get really flattering comments about it. It's actually been a long time since I wrote and launched an app directly to Normal People, and it's been a refreshing experience. Publishing my app to the market was a breeze - it's instant gratification. No approval process.
... and the bad...
Java.
Documentation is rather rough in places. Precious few example apps. Non-existent HIG.
The SDK GUI editor is very basic (I believe it's much improved in 1.5, need to check it out).
... and the ugly...
Bugs. The 1.1 release improved things a lot, but as a user I still the contact list system in particular to be distressingly buggy. It's by no means unusably buggy, but I expect a much more robust experience from my phone than I would a desktop OS.
HTTP APIs. There's two, the standard Java API and then apache httpclient. Unfortunately httpclient is version 4.x, not the more mature and well known 3.x. HttpClient 4.x has almost no useful documentation and doesn't support some features that 3.x did. PAIN.
All that said, I like writing apps for Android. Eclipse is decent. Java is decent. The distribution process is decent. And it's apparently improving pretty fast.
MacOS X isn't all that secure. Professional hackers have said that the implementation of ASLR/NX on Vista is far superior to Apples.
And as for Linux? Well, it wasn't that long ago that a certain high profile distribution accidentally disabled the pRNG in its core crypto libraries... for two years. And then another high profile distro let attackers actually sign some rogue packages with their private key. I don't think anybody should be making smart comments about the security of Linux.
That leaves Vista, the result of many years of applying the Secure Development Lifecycle. Extensive fuzz testing on the APIs. Extensive security review of all features. IE uses a low privilege renderering engine like Chrome (and unlike any browser on Linux or MacOS).
This doesn't mean MacOS or Linux are bad. But Microsoft have been throwing enormous resources behind security for years now. Is it any surprise they are caught up and in many ways ahead?
The java-subset thing seems like a bad idea; and I'd be curious to know why they did it; but I don't see how a platform subset is a good basis for a lock-in strategy.
Yeah, this is garbage. Watch the "campfire" videos, a boringly large part of the presentations is given over to how you are not locked in, because AppEngine exposes the standard Java servlet container and database access APIs even though it's based on BigTable which is not a standard database. They show how the guestbook app can be taken right across to run on WebSphere with no code changes. The design of Java on AppEngine is pretty much the opposite of lockin - they've clearly put a lot of effort into ensuring a very, very different underlying system can export the standard Java APIs.
As to why it's a subset, I guess the same logic as applied to the Python implementation which is also a subset - due to the way it works the classes need to be audited for security problems. Some of the Java APIs contain native code which probably has to be rewritten or at least very carefully audited to ensure you can't break out of the sandbox. And some I guess just aren't that useful. But I don't really know the reason.
Having every application behaving like Google's would be an utter disaster.
Obviously, open sourcing it is the first step to making it a general service any app can register with. Really, a Google Updater type system should be a part of Windows for many years now. But it's not. If there's going to be an updater system in the background, there might as well be only one - and one that is robust, widely deployed, with high quality code and maintained as open source by a dedicated team of full-time engineers seems like as good a start as any.
Why does getting an update check from an IP help with profiling? I don't get this leap of logic... if I want to do ad targetting based on IP address, knowing that something behind the same IP address has Google Earth installed doesn't help me at all.
Ugh, do you really want every app to get a multi-second delay on startup so it can check for updates? What happens if you're on a slow connection - your entire desktop grinds to a crawl thanks to the constant startup update checks. No app actually does it this way, it'd be crazy, startup time is important.
As to what stops the updater being compromised, I assume it checks whatever it downloads for a digital signature. Why would it not?
The unique ID is just a random number. How does that let Google tie your IP address to an advertising profile better than, say, a regular cookie? All this is good for is deduping update requests, to get an accurate figure for how many machines the software runs on.
If you were building an auto-updater, you'd probably be interested in knowing how many people had your app installed too. That way you know if people uninstall the app you're doing something wrong!
WoW isn't some kind of digital crack that enslaves anyone who plays. I tried WoW because a colleague told me that it was awesome and I should try it. He showed me around the world. I liked the artwork, the game world was cool. But even in that 30-60 minutes of playing, it was sort of tedious. What's more, you couldn't really get immersed in it because of the completely immature chat dialog that filled the world, and due to basic things like lack of collision detection between characters (required but not immersive). I never tried it again - saw no point.
that's really the only problem that Linux has: it's outside of people's comfort zone. The article is right: the combination of new hardware and new software is just too much for people to cope with.
That doesn't seem to stop Apple cleaning up in the US/EU laptop market. Different hardware? Check. Different software? Check. Actually I'd say MacOS is more different to Windows than the Linux desktop is.
The problem Linux has is that it's different but not in any significantly better way. What's the advantage of Linux for the average person? Is it easier than Windows? No. Is it visually slicker? Is it cooler? No. That matters a lot. Does it have a big company behind it that provides stores where you can go and talk to "geniuses" (ugh) if you get stuck? No. Is it more secure or less maintenance over the long run? No, because your average person these days gets infected due to the way they use their computer and not due to some inherent flaws of Windows, and because they can probably find a friend to fix it when they do unlike with Linux.
I've used Linux on the desktop for many years now, but I don't see any reason why I'd recommend it over Windows. For that matter I don't see much reason why I'd recommend MacOS X, except that it's less likely to have security issues today due to the way its market share is so skewed towards laptops which make for poor infection targets. But that's obviously something of a self-defeating recommendation in the long run.
If you want to replace Windows on the desktop, you need to go back to square one and rethink the whole thing from scratch. The only alternative is really marketing - take what already exists but make it prettier and cooler, ie the Apple approach. But it's been done already. Producing something fundamentally better is the only route left.
Your analysis is detailed and subtle but completely wrong, because you believe the people attack the captcha are smart. As pointed out repeatedly here, most spammers are extremely stupid and/or lazy. Go read some stories about spammers that got caught. Most of them just downloaded crap other people wrote and started using it. For instance your point number (1) - yes in theory, no in practice. Take something like Microsofts puppy dataset and see how many spammers bother to exhaust THAT possibility space.
Neither of those points is correct. Tethering apps are on the store outside of the US, and you can install apps from outside the store anyway just by downloading them with your web browser. The second point is also wrong. Developer phones can't install copy protected apps from the app store, but that's by no means the same as "can't install apps". I have tons of apps installed from the market on my dev phone.
Android is open, and it has the advantage of being built by people who know what they're doing and is thus a usable, competitive phone.
Ah. It's actually not doing well on netbooks. Windows now has 90% of the netbook market. Linux failed on netbooks for the same reasons it failed on desktops and laptops. It's got deep, fundamental problems that the people developing it are unwilling to admit to. If you look at the ridiculous software distribution mechanisms on Linux you can see the problems... apt is like an infinitely less appealing, flakier and slower-moving version of the iPhone app store. Yet the average Linux user will blindly defend the system as being superior to the one used by, gosh, every other OS ever designed and commercially successful.
It's very hard to use dramatically more electricity than your neighbour, if you try various fuses will blow. It's very easy to use dramatically more bandwidth than your neighbour, just fire up BitTorrent. If I could accidentally pull down a couple of megawatts by plugging in the wrong toaster, you'd see similar policies from electricity companies (eg "unlimited" plans that ban you from plugging in certain types of devices).
I work for a large internet company and definitely understand what bandwidth is. That doesn't mean I can divine how much my phone is going to use without some tricky mental arithmetic and a deep knowledge of how all the software I run works. I could measure it too, but I use an Android phone which updates itself over the air, and I don't know when or how big those updates are. So I can never really win.
I'm pretty sure the contracts you sign mention the lack of support for tethering. At least mine did. And yes the USB dongle approach is another way to do it because that way only people who want tethering pay the higher price for it. In theory you could do the same with phones but I guess it's too hard for operators to enable/disable tethering remotely depending on billing plan in a way that works across handsets and all operators.
Yes, if you want to purchase/sell a USB tethering dongle with a very expensive metered plan then you accept the associated billing problems and unhappy customers who go over the limit. That's not a mass market product though.
This is already tried for landline ISPs. Your business would encounter problems after people started encrypting everything to avoid packet shaping, then using BitTorrent to saturate your uplinks.
Yeah, great. So what happens when people do things that use lots of bandwidth without realising it? They get absurdly big charges. That leads to stories like this one. Your argument is basically, if you have crappy customer service, it's OK to set bandwidth caps and allow tethering because you don't care if the customer gets screwed anyway. Wonderful.
Gah, it's not about gouging. Why are you assuming every mobile operator in the world (cuz they practically all have the same policies) are Evil(tm)? Doesn't that strike you as rather unrealistic?
The reason tethering is disallowed is that it's the only business decision which makes sense. Simple.
Consider it from the operators perspective. They have finite mobile bandwidth, and they want to sell it to the mass market, ie, Joe Sixpack on his consumer phone. But they have a problem, the same problem landline ISPs have. Nobody, I repeat nobody understands what bandwidth is. Not Joe. Not you. Not me. It is sold to us in units of gigabytes/month, but what does that really mean? How many MP3s is that? How much web browsing? How many operating system updates? How many apps from the app store?
The fact is, consumer bandwith providers are in the unenviable position of selling a product nobody understands. They might as well sell bandwidth in pints for all the difference it'd make.
There is a simple solution for this problem - sell unlimited bandwidth plans (or plans so huge they're practically unlimited), and then use statistical models of how much bandwidth the average user gets through to set prices. Swallow the costs of the outliers and hope that on average your accounts end up a bit higher at the end of each month.
This business model works, and has allowed massive rollouts of internet connectivity across the world. There are a few things that break it. For mobile operators, tethering is one, because laptops will use so much more bandwidth than a mobile phone will. VoIP is the same - only a few people will use it, but those people will use the majority of the bandwidth dramatically raising costs for everyone. Rather than go back to selling people things they can't possibly understand, or boosting prices for everyone to subsidise the minority, they amend the contracts to read "unlimited, except no tethering and no voip" which is easy to grok even for Joe Sixpack.
If you were trying to sell bandwidth to the masses (and then deal with their billing enquiries!) you'd undoubtably do the exact same thing.
Slashdot Mirror
Windows Update does not use IE and hasn't since XP. You need to get information that isn't many years out of date.
Well, together Debian Ubuntu and Red Hat probably compromise the majority of Linux installs these days. If two large and well respected distros can fail in such basic ways, then it's reasonable to extrapolate that smaller and presumably less professional outfits will be even more flaky. Of course you can always find some Linux distro that has a perfect track record, but like I said above, usage counts. At some point if you want the word "Linux" to be meaningful you have to start talking about the bits actually in circulation.
DOS is probably more secure than both OpenBSD and Vista. At some point, you have to say "so what?". Usage does count for something.
I wrote an app which is now on the market. The good:
Java. OK, actually I hate Java. But I'd hate Objective-C a lot more. Implementing a simple crash reporter around my core logic was about 10 lines of Java code, and it works every time. Implementing the same thing in C++ or Objective-C would be .... non-trivial. No bother with heap corruptions, etc.
The whole design of activities and intents is quite well thought out. It seemed overly complicated at first but now I appreciate it a lot more. It's also very flexible, you aren't forced to use the infrastructure if you don't want to.
Really rich APIs. Background services, maps, multimedia, power management, package management, notifications .... even a face recognizer!
The market. I see a lot of people rag on the market and the comments system. Maybe I'm biased because my app has almost universally good reviews, but it's really nice to get that instant feedback about how you're doing. It's my experience that G1 owners (and there are apparently quite a lot) are ridiculously lenient. My app is extremely simple and could use a lot of extra features, yet I consistently get really flattering comments about it. It's actually been a long time since I wrote and launched an app directly to Normal People, and it's been a refreshing experience. Publishing my app to the market was a breeze - it's instant gratification. No approval process.
... and the bad ...
All that said, I like writing apps for Android. Eclipse is decent. Java is decent. The distribution process is decent. And it's apparently improving pretty fast.
No, he is probably right.
MacOS X isn't all that secure. Professional hackers have said that the implementation of ASLR/NX on Vista is far superior to Apples.
And as for Linux? Well, it wasn't that long ago that a certain high profile distribution accidentally disabled the pRNG in its core crypto libraries ... for two years. And then another high profile distro let attackers actually sign some rogue packages with their private key. I don't think anybody should be making smart comments about the security of Linux.
That leaves Vista, the result of many years of applying the Secure Development Lifecycle. Extensive fuzz testing on the APIs. Extensive security review of all features. IE uses a low privilege renderering engine like Chrome (and unlike any browser on Linux or MacOS).
This doesn't mean MacOS or Linux are bad. But Microsoft have been throwing enormous resources behind security for years now. Is it any surprise they are caught up and in many ways ahead?
Yeah, this is garbage. Watch the "campfire" videos, a boringly large part of the presentations is given over to how you are not locked in, because AppEngine exposes the standard Java servlet container and database access APIs even though it's based on BigTable which is not a standard database. They show how the guestbook app can be taken right across to run on WebSphere with no code changes. The design of Java on AppEngine is pretty much the opposite of lockin - they've clearly put a lot of effort into ensuring a very, very different underlying system can export the standard Java APIs.
As to why it's a subset, I guess the same logic as applied to the Python implementation which is also a subset - due to the way it works the classes need to be audited for security problems. Some of the Java APIs contain native code which probably has to be rewritten or at least very carefully audited to ensure you can't break out of the sandbox. And some I guess just aren't that useful. But I don't really know the reason.
Obviously, open sourcing it is the first step to making it a general service any app can register with. Really, a Google Updater type system should be a part of Windows for many years now. But it's not. If there's going to be an updater system in the background, there might as well be only one - and one that is robust, widely deployed, with high quality code and maintained as open source by a dedicated team of full-time engineers seems like as good a start as any.
Why does getting an update check from an IP help with profiling? I don't get this leap of logic ... if I want to do ad targetting based on IP address, knowing that something behind the same IP address has Google Earth installed doesn't help me at all.
Ugh, do you really want every app to get a multi-second delay on startup so it can check for updates? What happens if you're on a slow connection - your entire desktop grinds to a crawl thanks to the constant startup update checks. No app actually does it this way, it'd be crazy, startup time is important.
As to what stops the updater being compromised, I assume it checks whatever it downloads for a digital signature. Why would it not?
The unique ID is just a random number. How does that let Google tie your IP address to an advertising profile better than, say, a regular cookie? All this is good for is deduping update requests, to get an accurate figure for how many machines the software runs on.
If you were building an auto-updater, you'd probably be interested in knowing how many people had your app installed too. That way you know if people uninstall the app you're doing something wrong!
You didn't take the wireless router down from its mast first?!?
WoW isn't some kind of digital crack that enslaves anyone who plays. I tried WoW because a colleague told me that it was awesome and I should try it. He showed me around the world. I liked the artwork, the game world was cool. But even in that 30-60 minutes of playing, it was sort of tedious. What's more, you couldn't really get immersed in it because of the completely immature chat dialog that filled the world, and due to basic things like lack of collision detection between characters (required but not immersive). I never tried it again - saw no point.
That doesn't seem to stop Apple cleaning up in the US/EU laptop market. Different hardware? Check. Different software? Check. Actually I'd say MacOS is more different to Windows than the Linux desktop is.
The problem Linux has is that it's different but not in any significantly better way. What's the advantage of Linux for the average person? Is it easier than Windows? No. Is it visually slicker? Is it cooler? No. That matters a lot. Does it have a big company behind it that provides stores where you can go and talk to "geniuses" (ugh) if you get stuck? No. Is it more secure or less maintenance over the long run? No, because your average person these days gets infected due to the way they use their computer and not due to some inherent flaws of Windows, and because they can probably find a friend to fix it when they do unlike with Linux.
I've used Linux on the desktop for many years now, but I don't see any reason why I'd recommend it over Windows. For that matter I don't see much reason why I'd recommend MacOS X, except that it's less likely to have security issues today due to the way its market share is so skewed towards laptops which make for poor infection targets. But that's obviously something of a self-defeating recommendation in the long run.
If you want to replace Windows on the desktop, you need to go back to square one and rethink the whole thing from scratch. The only alternative is really marketing - take what already exists but make it prettier and cooler, ie the Apple approach. But it's been done already. Producing something fundamentally better is the only route left.
Your analysis is detailed and subtle but completely wrong, because you believe the people attack the captcha are smart. As pointed out repeatedly here, most spammers are extremely stupid and/or lazy. Go read some stories about spammers that got caught. Most of them just downloaded crap other people wrote and started using it. For instance your point number (1) - yes in theory, no in practice. Take something like Microsofts puppy dataset and see how many spammers bother to exhaust THAT possibility space.
Neither of those points is correct. Tethering apps are on the store outside of the US, and you can install apps from outside the store anyway just by downloading them with your web browser. The second point is also wrong. Developer phones can't install copy protected apps from the app store, but that's by no means the same as "can't install apps". I have tons of apps installed from the market on my dev phone.
Android is open, and it has the advantage of being built by people who know what they're doing and is thus a usable, competitive phone.
Ah. It's actually not doing well on netbooks. Windows now has 90% of the netbook market. Linux failed on netbooks for the same reasons it failed on desktops and laptops. It's got deep, fundamental problems that the people developing it are unwilling to admit to. If you look at the ridiculous software distribution mechanisms on Linux you can see the problems ... apt is like an infinitely less appealing, flakier and slower-moving version of the iPhone app store. Yet the average Linux user will blindly defend the system as being superior to the one used by, gosh, every other OS ever designed and commercially successful.
It's very hard to use dramatically more electricity than your neighbour, if you try various fuses will blow. It's very easy to use dramatically more bandwidth than your neighbour, just fire up BitTorrent. If I could accidentally pull down a couple of megawatts by plugging in the wrong toaster, you'd see similar policies from electricity companies (eg "unlimited" plans that ban you from plugging in certain types of devices).
I work for a large internet company and definitely understand what bandwidth is. That doesn't mean I can divine how much my phone is going to use without some tricky mental arithmetic and a deep knowledge of how all the software I run works. I could measure it too, but I use an Android phone which updates itself over the air, and I don't know when or how big those updates are. So I can never really win.
They also need to be powered, upgraded to the newest kit and spectrum needs to be licensed. I agree, that's an overly simplistic analysis.
I'm pretty sure the contracts you sign mention the lack of support for tethering. At least mine did. And yes the USB dongle approach is another way to do it because that way only people who want tethering pay the higher price for it. In theory you could do the same with phones but I guess it's too hard for operators to enable/disable tethering remotely depending on billing plan in a way that works across handsets and all operators.
Yes, if you want to purchase/sell a USB tethering dongle with a very expensive metered plan then you accept the associated billing problems and unhappy customers who go over the limit. That's not a mass market product though.
This is already tried for landline ISPs. Your business would encounter problems after people started encrypting everything to avoid packet shaping, then using BitTorrent to saturate your uplinks.
Yeah, great. So what happens when people do things that use lots of bandwidth without realising it? They get absurdly big charges. That leads to stories like this one. Your argument is basically, if you have crappy customer service, it's OK to set bandwidth caps and allow tethering because you don't care if the customer gets screwed anyway. Wonderful.
I wasn't attempting to explain the cost of SMS which I agree is too high. Your point is irrelevant to mine.
Gah, it's not about gouging. Why are you assuming every mobile operator in the world (cuz they practically all have the same policies) are Evil(tm)? Doesn't that strike you as rather unrealistic?
The reason tethering is disallowed is that it's the only business decision which makes sense. Simple.
Consider it from the operators perspective. They have finite mobile bandwidth, and they want to sell it to the mass market, ie, Joe Sixpack on his consumer phone. But they have a problem, the same problem landline ISPs have. Nobody, I repeat nobody understands what bandwidth is. Not Joe. Not you. Not me. It is sold to us in units of gigabytes/month, but what does that really mean? How many MP3s is that? How much web browsing? How many operating system updates? How many apps from the app store?
The fact is, consumer bandwith providers are in the unenviable position of selling a product nobody understands. They might as well sell bandwidth in pints for all the difference it'd make.
There is a simple solution for this problem - sell unlimited bandwidth plans (or plans so huge they're practically unlimited), and then use statistical models of how much bandwidth the average user gets through to set prices. Swallow the costs of the outliers and hope that on average your accounts end up a bit higher at the end of each month.
This business model works, and has allowed massive rollouts of internet connectivity across the world. There are a few things that break it. For mobile operators, tethering is one, because laptops will use so much more bandwidth than a mobile phone will. VoIP is the same - only a few people will use it, but those people will use the majority of the bandwidth dramatically raising costs for everyone. Rather than go back to selling people things they can't possibly understand, or boosting prices for everyone to subsidise the minority, they amend the contracts to read "unlimited, except no tethering and no voip" which is easy to grok even for Joe Sixpack.
If you were trying to sell bandwidth to the masses (and then deal with their billing enquiries!) you'd undoubtably do the exact same thing.