See: the free market works - you get the service you want.
You fail. There is no way that "free market" describes any situation where a small number of companies are protected from competition by government regulation.
Intel doesn't fabricate quad core processors - they only make single and dual core chips. They may well be selling bad dual cores as single core processors (or not), but their chips are tested well before two dual cores get glued together into a quad core so they don't have the same situation that makes triple-core make sense for AMD.
Sure. Let's outlaw arc welders too. They can blind people. And screwdrivers, because you could stab someone. And metal forks, because you might try to use them to get stuff out of the toaster and electrocute yourself.
Or maybe some tools are dangerous, but everyone needs to accept that it's not the role of government to protect people from every little danger.
As for "going digital", suggest a digital input/output that is universally available (e.g. is input for monitors, output for cameras, input/output for computers including Linux supported, and input/output for recording devices like DVD-R, DVR, etc)... and supports HD. Hint: I doubt you can find one.
Who cares if there's a single universal input/output connector? All you need if you're actually trying to accomplish something is to connect *your specific setup* together.
Looking around quickly, I see a couple of capture cards that support s-video in - that's a component format.
I'm not quite sure why you would focus on that though - if you're doing amateur stuff composite is fine and if you need higher quality you can just go digital earlier.
I'm also trying to address it and the social problems that exacerbate it with discussion with users, like you for example. Unless vocal users are aware of real usability issues and willing to admit those problems, it does more to dissuade people from using Linux on the desktop than the problems themselves in many cases.
The vast majority of people on discussion boards like Slashdot have absolutely no influence on the development of any OS distribution. Posts that say "Linux isn't ready for mainstream users / the desktop" basically come across as trolls and/or astroturfing.
If you want to actually accomplish something useful when it comes to usability on Linux, the most effective thing that you can do is submit Ubuntu blueprints for the next release - right now is the perfect time to actually get your suggestions into 8.10 by submitting one or more blueprints (https://blueprints.launchpad.net/ubuntu/).
Yea, but when you decide to back up your collection of 200 such movies to an online service, it'd be better if the backup process takes 4 minutes rather than 7 hours.
If you get your fuel recycling going properly, then the cycle-end waste gets back to ore-level radioactivity in a couple hundred years. We have building technology that can reliably be trusted to store stuff for a couple hundred years - poured cement anywhere that isn't in a flood zone or on a tectonic fault line.
It's only with this damn fool "recycling nuclear fuel gives the terrorists nuclear bombs" nonsense that we're stuck with dangerously radioactive material 1500 years from now.
Making claims about technology not bounded in time is a poor choice. If we humans haven't managed to increase our perceptual bandwidth in, say, ten thousand years then we are an absolute failure as an intelligent species.
But in the foreseeable future you very well might be right that a terabit per second is a bit much for an average household. Compressed HDTV is only 20 Mbps. We can probably completely saturate the human sensory inputs with nearly-uncompressed data at a bandwidth on the order of a gigabit per second.
On the other hand, there may be engineering gains in using bandwidth inefficiently. For example, TOR uses significantly more bandwidth than traditional routing in order to provide anonymity. Compare processor clock cycles - in 1955 the idea of wasting them was absurd. Today, I throw away two orders of magnitude of processor efficiency every time I decide to write a program in Ruby - and it's usually the right engineering choice simply because my time is worth more than processor time is.
Here's another concept: Optimistic delivery for interactive content. Imagine that we have a full-sensory VR environment and 1 Tbps of bandwidth, but still have hundreds of milliseconds of latency (damn speed of light). We can trade off bandwidth against latency by guessing at the user's actions and pre-buffering the content that they *might* experience. They could look up, or look down, or hit the button to teleport themselves to an underwater reef palace - so load it all.
Another, even simpler point is that people frequently want to download things like movies faster than they can watch them. So if we have a 200 Mbps super-HD two-hour movie, it'll take 1.3 seconds to download on a 1 terabit link. That's sure better than the nearly two minutes it would take to download on a 10 Gbps link. So I guess that does show that a 1 Tbps link may be useful for an average household ever.
That would also require calling their damn support number and waiting on hold for 40 minutes.
Further, where I live there is a Verison / Comcast duopoly on consumer / small business grade internet connectivity. Comcast sucks a bit more than Verizon does, so my basic choices are to 1.) stick with Verizon or 2.) have no usable internet connection or 3.) get a real (dedicated line) internet connection from a legitimate provider. #3 is the correct solution, but I can't really afford a T3 by myself right now and I'm not going to be living here long enough to go through the effort of organizing a bandwidth co-op.
With today's computers, I can't think of any way for a single household to usefully saturate a one terabit per second link. And I'm the guy who always calls people short sighted idiots for saying that X amount of bandwidth is "good enough". In order to even *generate* that traffic you'd need 1000 computers with gigabit ethernet and obscene routing hardware.
For today, I'd be happy with a gigabit connection. I feel pretty safe in saying that I won't be needing any faster than that, personally, this year.
Looking into it, they do have a way to opt out of this "service" -- although if you're not at least reasonably competent with making TCP/IP configuration changes on a home router, don't bother; it involves looking up and modifying IP addresses. Not a big deal to most/.ers, I'd say, but a nightmare for the general public.
The opt-out instructions don't work, at least here in eastern Massachussetts. And there's no way to complain about it short of calling tech support and waiting on hold for 40 minutes.
Any sort of engineer, scientist, or technician should be using the units that make their math the easiest. For computer science, power-of-two units for storage is correct. If we CS people have to occasionally write KiB (pronounced "kay" or "kilobyte") every once and a while when talking to communication engineers, that's fine - but I could care less about confusing the general public by 2% on hard disk size.
People can't go off touting Ubuntu as "more usable" and "ready for mainstream users" and then stick their heads in the sand when real problems are pointed out.
If operating systems can't be touted as "ready for mainstream users" while *any* difficult use cases exist then no operating system qualifies. Mac OS, every version of Windows, and every Linux distro all have rough edges - and they always will if you're picky enough.
Ubuntu is usable by mainstream users in practice today. For some use cases it's better than the other choices, for other use cases it's worse. It still has places to improve, and identifying those places and submitting well thought out feature suggestions is helpful. Saying that "it's not ready for mainstream users" is just flamebait, in the same way that saying that "Macs will never be mainstream because they only run on expensive Mac hardware" is.
Apparently you think that Mac OS application bundles are a Panacea. That's not obviously the case. There are some use cases where other solutions are better. Further, your implication that Linux package management is worse than Windows packages because it's worse than app bundles is silly.
There are trade-offs in package formats, and although I agree that Linux does still have some usability issues when it comes to packaging the issue isn't as clear cut as you're making it out to be.
As an active user of Ubuntu, I can assure you that directing other users to packages in the repositories by IM or email is the simplest use case ever - tell them the package name and they're all set. You can complain about other use cases all you want, but using that one as an example is absurd.
The thing is, a lot of people who aren't developers might want to run cutting edge software or just normal software that isn't packaged for their distro.
Sure. I see that issue and agree that it's a potential problem. I'm just saying that an SDK isn't a good example.
I guess what I'm frustrated by is the Linux development community's tendency to excuse usability flaws by claiming they are "features" to stop users from doing things if they aren't experts or by claiming it is good to keep it hard to use to keep non-expert users off the platform out of elitism.
I'm frustrated with some users' tendency to complain about usability issues that either 1.) haven't existed for years or 2.) have workable solutions that the users obviously have made no attempt to find or 3.) really are related to expert-only tasks that are properly designed to be convenient for experts (i.e. building a custom kernel).
The issue of packaging is a real one, but in practice it's also a solved one. For a good example, try to install Skype on Ubuntu by clicking the skype.com download link. A good example of a general executable installer is Google Earth.
The problem with this is you still lose the ability to keep it up do date because package managers don't know where it came from (no standard internal URL for checking for updates) It is better than nothing, or even what is provided right now, but as you point out, it is less than ideal.
This is one of those "the perfect is the enemy of the good" things. Sure, lots of things could be done better - but that doesn't mean that people who are trying to accomplish things today shouldn't use the (perfectly functional) methods that are available right now.
Most people probably cannot though.
Just to clarify, we aren't talking about most people. We're talking about developers who have decided to use cutting-edge SDKs. People who can't handle building packages from source are, almost by definition, not developers.
--
Right now, there are clear and obvious ways for developers who are releasing programs for Ubuntu and other Linux distributions to package their stuff so that users can easily install it. The main problem is that those developers need to actually *do it* - sure, the distros need to keep improving the options, but the real problem today is developers not using what's available.
The issues that you mention are real and still not cleanly resolved in practice.
For Ubuntu, it's mostly an issue of websites offering downloads simply needing to offer 1.) a.deb package if there is no package in Ubuntu or 2.) an apt-url link if there is a package in the Ubuntu repositories. The question is then "That solves Ubuntu, but what about the other Linux distros" - and there's no widely accepted good answer to that.
The cleanest general answer is to provide a self-extracting installer (like loki-setup) and having it default to installing in the current user's home directory. Then it feels just like the Windows install process for a single-user Linux system. That's what proprietary commercial Linux games like UT2k4 have always done.
As for frequently-changing development tools, the standard on all free Unix-like system is.tar.gz source archives. Developers are expected to be prepared to deal with them. I'm a developer myself, and I can handle having to type "./configure && make && make install" a couple times to get a new SDK installed - you can too.
They are either expected to do research and know what hardware to purchase, or they are expected to know nothing and be able to install a 'Unix Desktop' without any problems.
The first category is called a "hobbyist", and has no trouble installing Linux. The second category (that knows nothing and yet can install an OS) is largely a myth.
As for "the masses", they buy computers with the OS pre-installed.
Not at all. Hardware that runs Linux perfectly is readily available, and the fact that some wifi chipsets don't work with Linux has been well known for years. This shouldn't be news to anyone who's considering installing any Linux distro themselves.
Not that most people would even consider installing an OS themselves - for the majority of the "unwashed masses", a new OS is something that comes with a new computer. Ubuntu pre-installed on a Dell works just as well as Windows pre-installed on a Dell or OS X pre-installed on a Mac.
And I see frequently normal users go to technically minded friends or even the computer shop to upgrade, if they bother upgrading at all. Mostly I see them throw away their old spyware-clogged Windows machines and buy new ones.
But even if I accept your point that they do upgrade, upgrading an OS isn't the same as installing it from scratch. It's unlikely that any Linux distro will ever make the install-over-Windows as simple as an upgrade from one version of Windows to another is when it goes smoothly.
The installation process for Ubuntu today is easily as simple as the installation process for Windows Vista or Mac OS X. I've personally installed both Ubuntu and Vista on fresh machines in the last two weeks, so I know what I'm talking about. And the upgrade process between versions of Ubuntu makes it look like Apple and Microsoft are stuck back in the dark ages. Comparative ease of use in installation hasn't been the issue for years.
I've been thinking of giving it another go on a (different) spare PC I have. As a semi-technical person, I might grow to love Linux but if I can't get simple stuff to work I'll just dump it for another five years.
Or you could just buy a new machine with Ubuntu installed and have it just work. Installing it on random hardware is *your choice*, and if it fails to work because you have unsupported hardware it's *your fault*. Ubuntu works great on supported hardware, and everyone knows that there are a couple pieces of hardware that simply won't work.
Really, hearing stories about WinModems and Dell wireless not working stopped being interesting in like 2001. If you can handle installing the OS, you can handle installing a PCI card with hardware that works.
By this logic, Windows (which runs on the most random hardware) is almost awesome!
My random hardware is a... Playstation 3. My random hardware is a... Sega Dreamcast. My random hardware is a... Linksys wireless router. Looks like Windows (Vista or XP) is 0 for 3. My random hardware is an... iPod. My random hardware is a... SparcStation.
Windows only seems to work on a lot of hardware when you constrain yourself to hardware that was basically intended to work as/with a Windows PC.
Why bother when Windows works just fine with ANY wireless interface.
Why bother with Windows when NetBSD supports any 32-bit microprocessor?
Finding wireless cards that work great under Ubuntu (or processors that work great under Windows) isn't hard. It makes a lot of sense to select the hardware that you need to run the software that you want to use.
Slashdot Mirror
You fail. There is no way that "free market" describes any situation where a small number of companies are protected from competition by government regulation.
Intel doesn't fabricate quad core processors - they only make single and dual core chips. They may well be selling bad dual cores as single core processors (or not), but their chips are tested well before two dual cores get glued together into a quad core so they don't have the same situation that makes triple-core make sense for AMD.
Sure. Let's outlaw arc welders too. They can blind people. And screwdrivers, because you could stab someone. And metal forks, because you might try to use them to get stuff out of the toaster and electrocute yourself.
Or maybe some tools are dangerous, but everyone needs to accept that it's not the role of government to protect people from every little danger.
Who cares if there's a single universal input/output connector? All you need if you're actually trying to accomplish something is to connect *your specific setup* together.
Looking around quickly, I see a couple of capture cards that support s-video in - that's a component format.
I'm not quite sure why you would focus on that though - if you're doing amateur stuff composite is fine and if you need higher quality you can just go digital earlier.
Ha!
The vast majority of people on discussion boards like Slashdot have absolutely no influence on the development of any OS distribution. Posts that say "Linux isn't ready for mainstream users / the desktop" basically come across as trolls and/or astroturfing.
If you want to actually accomplish something useful when it comes to usability on Linux, the most effective thing that you can do is submit Ubuntu blueprints for the next release - right now is the perfect time to actually get your suggestions into 8.10 by submitting one or more blueprints (https://blueprints.launchpad.net/ubuntu/).
Yea, but when you decide to back up your collection of 200 such movies to an online service, it'd be better if the backup process takes 4 minutes rather than 7 hours.
If you get your fuel recycling going properly, then the cycle-end waste gets back to ore-level radioactivity in a couple hundred years. We have building technology that can reliably be trusted to store stuff for a couple hundred years - poured cement anywhere that isn't in a flood zone or on a tectonic fault line.
It's only with this damn fool "recycling nuclear fuel gives the terrorists nuclear bombs" nonsense that we're stuck with dangerously radioactive material 1500 years from now.
Making claims about technology not bounded in time is a poor choice. If we humans haven't managed to increase our perceptual bandwidth in, say, ten thousand years then we are an absolute failure as an intelligent species.
But in the foreseeable future you very well might be right that a terabit per second is a bit much for an average household. Compressed HDTV is only 20 Mbps. We can probably completely saturate the human sensory inputs with nearly-uncompressed data at a bandwidth on the order of a gigabit per second.
On the other hand, there may be engineering gains in using bandwidth inefficiently. For example, TOR uses significantly more bandwidth than traditional routing in order to provide anonymity. Compare processor clock cycles - in 1955 the idea of wasting them was absurd. Today, I throw away two orders of magnitude of processor efficiency every time I decide to write a program in Ruby - and it's usually the right engineering choice simply because my time is worth more than processor time is.
Here's another concept: Optimistic delivery for interactive content. Imagine that we have a full-sensory VR environment and 1 Tbps of bandwidth, but still have hundreds of milliseconds of latency (damn speed of light). We can trade off bandwidth against latency by guessing at the user's actions and pre-buffering the content that they *might* experience. They could look up, or look down, or hit the button to teleport themselves to an underwater reef palace - so load it all.
Another, even simpler point is that people frequently want to download things like movies faster than they can watch them. So if we have a 200 Mbps super-HD two-hour movie, it'll take 1.3 seconds to download on a 1 terabit link. That's sure better than the nearly two minutes it would take to download on a 10 Gbps link. So I guess that does show that a 1 Tbps link may be useful for an average household ever.
That would also require calling their damn support number and waiting on hold for 40 minutes.
Further, where I live there is a Verison / Comcast duopoly on consumer / small business grade internet connectivity. Comcast sucks a bit more than Verizon does, so my basic choices are to 1.) stick with Verizon or 2.) have no usable internet connection or 3.) get a real (dedicated line) internet connection from a legitimate provider. #3 is the correct solution, but I can't really afford a T3 by myself right now and I'm not going to be living here long enough to go through the effort of organizing a bandwidth co-op.
With today's computers, I can't think of any way for a single household to usefully saturate a one terabit per second link. And I'm the guy who always calls people short sighted idiots for saying that X amount of bandwidth is "good enough". In order to even *generate* that traffic you'd need 1000 computers with gigabit ethernet and obscene routing hardware.
For today, I'd be happy with a gigabit connection. I feel pretty safe in saying that I won't be needing any faster than that, personally, this year.
The opt-out instructions don't work, at least here in eastern Massachussetts. And there's no way to complain about it short of calling tech support and waiting on hold for 40 minutes.
Any sort of engineer, scientist, or technician should be using the units that make their math the easiest. For computer science, power-of-two units for storage is correct. If we CS people have to occasionally write KiB (pronounced "kay" or "kilobyte") every once and a while when talking to communication engineers, that's fine - but I could care less about confusing the general public by 2% on hard disk size.
If operating systems can't be touted as "ready for mainstream users" while *any* difficult use cases exist then no operating system qualifies. Mac OS, every version of Windows, and every Linux distro all have rough edges - and they always will if you're picky enough.
Ubuntu is usable by mainstream users in practice today. For some use cases it's better than the other choices, for other use cases it's worse. It still has places to improve, and identifying those places and submitting well thought out feature suggestions is helpful. Saying that "it's not ready for mainstream users" is just flamebait, in the same way that saying that "Macs will never be mainstream because they only run on expensive Mac hardware" is.
Apparently you think that Mac OS application bundles are a Panacea. That's not obviously the case. There are some use cases where other solutions are better. Further, your implication that Linux package management is worse than Windows packages because it's worse than app bundles is silly.
There are trade-offs in package formats, and although I agree that Linux does still have some usability issues when it comes to packaging the issue isn't as clear cut as you're making it out to be.
As an active user of Ubuntu, I can assure you that directing other users to packages in the repositories by IM or email is the simplest use case ever - tell them the package name and they're all set. You can complain about other use cases all you want, but using that one as an example is absurd.
Sure. I see that issue and agree that it's a potential problem. I'm just saying that an SDK isn't a good example.
I'm frustrated with some users' tendency to complain about usability issues that either 1.) haven't existed for years or 2.) have workable solutions that the users obviously have made no attempt to find or 3.) really are related to expert-only tasks that are properly designed to be convenient for experts (i.e. building a custom kernel).
The issue of packaging is a real one, but in practice it's also a solved one. For a good example, try to install Skype on Ubuntu by clicking the skype.com download link. A good example of a general executable installer is Google Earth.
This is one of those "the perfect is the enemy of the good" things. Sure, lots of things could be done better - but that doesn't mean that people who are trying to accomplish things today shouldn't use the (perfectly functional) methods that are available right now.
Just to clarify, we aren't talking about most people. We're talking about developers who have decided to use cutting-edge SDKs. People who can't handle building packages from source are, almost by definition, not developers.
--
Right now, there are clear and obvious ways for developers who are releasing programs for Ubuntu and other Linux distributions to package their stuff so that users can easily install it. The main problem is that those developers need to actually *do it* - sure, the distros need to keep improving the options, but the real problem today is developers not using what's available.
The issues that you mention are real and still not cleanly resolved in practice.
For Ubuntu, it's mostly an issue of websites offering downloads simply needing to offer 1.) a .deb package if there is no package in Ubuntu or 2.) an apt-url link if there is a package in the Ubuntu repositories. The question is then "That solves Ubuntu, but what about the other Linux distros" - and there's no widely accepted good answer to that.
The cleanest general answer is to provide a self-extracting installer (like loki-setup) and having it default to installing in the current user's home directory. Then it feels just like the Windows install process for a single-user Linux system. That's what proprietary commercial Linux games like UT2k4 have always done.
As for frequently-changing development tools, the standard on all free Unix-like system is .tar.gz source archives. Developers are expected to be prepared to deal with them. I'm a developer myself, and I can handle having to type "./configure && make && make install" a couple times to get a new SDK installed - you can too.
The first category is called a "hobbyist", and has no trouble installing Linux. The second category (that knows nothing and yet can install an OS) is largely a myth.
As for "the masses", they buy computers with the OS pre-installed.
Not at all. Hardware that runs Linux perfectly is readily available, and the fact that some wifi chipsets don't work with Linux has been well known for years. This shouldn't be news to anyone who's considering installing any Linux distro themselves.
Not that most people would even consider installing an OS themselves - for the majority of the "unwashed masses", a new OS is something that comes with a new computer. Ubuntu pre-installed on a Dell works just as well as Windows pre-installed on a Dell or OS X pre-installed on a Mac.
And I see frequently normal users go to technically minded friends or even the computer shop to upgrade, if they bother upgrading at all. Mostly I see them throw away their old spyware-clogged Windows machines and buy new ones.
But even if I accept your point that they do upgrade, upgrading an OS isn't the same as installing it from scratch. It's unlikely that any Linux distro will ever make the install-over-Windows as simple as an upgrade from one version of Windows to another is when it goes smoothly.
The installation process for Ubuntu today is easily as simple as the installation process for Windows Vista or Mac OS X. I've personally installed both Ubuntu and Vista on fresh machines in the last two weeks, so I know what I'm talking about. And the upgrade process between versions of Ubuntu makes it look like Apple and Microsoft are stuck back in the dark ages. Comparative ease of use in installation hasn't been the issue for years.
Or you could just buy a new machine with Ubuntu installed and have it just work. Installing it on random hardware is *your choice*, and if it fails to work because you have unsupported hardware it's *your fault*. Ubuntu works great on supported hardware, and everyone knows that there are a couple pieces of hardware that simply won't work.
Really, hearing stories about WinModems and Dell wireless not working stopped being interesting in like 2001. If you can handle installing the OS, you can handle installing a PCI card with hardware that works.
My random hardware is a... Playstation 3. My random hardware is a... Sega Dreamcast. My random hardware is a... Linksys wireless router. Looks like Windows (Vista or XP) is 0 for 3. My random hardware is an... iPod. My random hardware is a... SparcStation.
Windows only seems to work on a lot of hardware when you constrain yourself to hardware that was basically intended to work as/with a Windows PC.
Why bother with Windows when NetBSD supports any 32-bit microprocessor?
Finding wireless cards that work great under Ubuntu (or processors that work great under Windows) isn't hard. It makes a lot of sense to select the hardware that you need to run the software that you want to use.
Thanks for posting and clarifying.
That seems to be basically what the article says too. I wonder if Slashdot editors actually read stories before posting them with flamebait summaries?