That's the point tho... XP replaced 98/ME (from the consumer's perspective), and offered a significant improvement over it, prompting them to change... Vista replaced XP, but did not offer a significant improvement and in many cases was actually worse... People resisted changing to vista, but by the time 7 comes out they will often have little choice.
Being "standardized" is both a good and a bad thing... One size does not fit all, and a monoculture is no good for anyone... The ability to select the best tool for the job rather than having to use a "standard" is a good thing. Data should be standardized, but how you interact with it should be a matter of choice... Roads are standard, tv signals are standard, but people drive all kinds of different vehicles and use different types of tv. But to answer your question:
Windows makes a terrible server platform, you can't strip it down to the bare minimum, you can't install and manage it from a serial console... and don't mention the "cli only mode" in windows 2008, their idea of cli only is to load the entire gui layer and then put a cmd.exe window in the middle of it... what was the point in loading the gui layer with all it's overhead just to display a cli? a windowed cli will never work over a serial console either...
On a desktop system the interface is extremely clunky, and is very much geared towards doing things their way or nothing... Their way doesn't suit me, the default ways of most linux distros don't suit me either but linux is much easier to customize.
Linux is easy to use without a command line, modern distros will let you do everything most users will ever want without a command line... And yet, many seasoned users actively choose to use the command line on linux.. Why? because in many cases it's easier, much easier for an experienced user, and much easier to explain when trying to help an unskilled user. Windows users, even experienced ones rarely use the command line mostly because the windows cli is pretty bad, but one counter example is when someone doing phone support wants an ip address from a windows user they're supporting, they almost always have them open a command prompt and type "ipconfig"... Why? because that's easier than finding the IP through the gui (which i assume can be done somehow). Because of this people get the impression linux is only usable from the cli, when in reality the cli is often the best but not the only way to do many things...
And when it comes to advanced things, a cli where you can cut+paste is much easier than regedit...
Another issue is package management, windows simply doesn't have one, on linux i can just open my package manager, search for what i want and hit install, and what i want is installed including any dependencies it has, or i can do it from the cli. Windows requires you to manually find what you want through google, trust that the download site you find is reputable (when was the last time a windows user downloaded a file from a random download site and then compared the checksums with a set published by an official site?), and then wait for the download to finish before you can manually execute and follow through with the installer. That is just a HUGE pain in the ass.
Multiple workspaces - i cant live without multiple workspaces, and all the windows implementations i've seen have sucked, mostly because no apps or even the basic window manager are designed with workspaces in mind. And yes, aside from workspaces i can't stand the way the windows window manager works, i find it clunky and inflexible.
Foreign filesystems - linux comes with support for all kinds of filesystems out of the box, windows just doesn't, and what third party filesystem drivers do exist are often poorly implemented, buggy or both... I have to read disks from macs, bsd and linux boxes all the time and occasionally misc other systems, windows just doesn't cut it, they arrogantly only support their own filesystems.
Source - I want the ability to modify the source code of the programs i use, to do things the authors never intended... I also want to be able to use new and exciting hardware, many years ago i used alpha, more recently i was using 64bit amd64 very early on and these days i would be looking at low power arm based systems, microsoft have always been playing catch up and the closed source nature of most windows apps make
If you sign up for and pay for a service, then Google are obliged to continue providing that service for the term of the contract providing you keep up your end of the bargain (ie paying for it)... They are also obliged to abide by any other terms in the contract, so as a business you would be foolish to not demand clauses that forbid Google from mining your data or doing anything else with it not directly related to providing the service.. You should also demand the ability to download all of your data at any time you wish in a standard format so that you can keep your own backup and/or have the data available to you for migrating to another service.
Companies trust their critical data and resources to other companies all the time... There are companies who specialize in storing or destroying massive quantities of hard copy documents for instance, not to mention courier or security companies.
This is actually a better situation than using MS apps, which come with no warranty and often provide no method or guarantee to get the data out in a standard format.
Obviously the best plan is to have the data on your own hardware, in formats which you have full documentation for, and using software which you have an unlimited irrevocable right to use for any purpose, but neither MS nor Google offer this right now.
I have seen a LOT of big companies which are doing everything you suggest, virus scanners on inbound email and the web proxy with no access to the outside world. They usually block the download or receiving via email of executables of any kind, or any kind of encrypted data (since they cant check the contents of it).. They also tend to either disallow removable devices entirely, or limit them to certain users and enforce virus scans on any inserted devices.
And yet these companies still get infected, new malware gets straight through their web filtering because it's newly written and the company providing the filter hasn't added it to their signatures yet... Lots of malware also gets through the email filter because it's embedded in proprietary formats like word and excel.. These files aren't blocked because external companies often send files in such formats, which are hard to parse and new malware is constantly being developed. All it takes is for one user to get infected, and their machine can be a source of infection for everything else... Most companies protect their borders, but the inside is extremely insecure. They use shared authentication schemes like active directory because they have large numbers of machines to manage, but still don't have the resources to ensure everything is up to date and securely configured so it's usually trivially easy to compromise a machine or two, make your way onto the domain (made easier by it running the same software and network visible apis as the member systems) and take the whole network very easily...
Most of the issues are due to outlook only really supporting it's own proprietary protocols for most things... They are designed to lock you in to exchange. That's why their support for IMAP is lousy, and they have no support for anything like CalDAV at all.
You won't be trimming off the low hanging fruit, you will simply be raising that fruit a few inches higher... And because less people will now be picking it, those people who are a bit taller will now have a lot more easily accessible fruit to pick.
What would happen, is that the prevalence of unskilled script kiddies would massively decrease, and the background scans taking place constantly would decrease... Because the perceived threats would have abated, people wouldn't bother installing updates or taking any measures to protect themselves. Also without public disclosure and/or active exploitation, software vendors would downplay the seriousness of their vulnerabilities and delay providing patches for them.
The end result of this, is that the smaller number of people who can acquire exploits, and this includes paid criminal gangs, would have a lot more power because they would no longer have to compete against the script kiddies for control of drone systems.
Incidentally, i am also against the *free* disclosure of vulnerabilities in non free software... Commercial vendors charge you a lot of money for their software, and can often be hostile or uncommunicative towards people who find bugs in their software... These people finding bugs are effectively doing their jobs for them and get nothing but grief in return, so it's no wonder that so many bug hunters are now working for criminal gangs. A lot of these vendors want you to do their beta testing for them for free, and then report the bugs privately to them so they can silently fix them not even giving you credit for the find and often not disclosing any details to the public other than perhaps providing a black box patch.
Or for a blackhat, what better way to divert the blame? Bots are plentiful, insecure windows boxes are extremely abundant and it will be easy for them to acquire more, they probably haven't even diverted all of their current resources to this attack. The machines that get wiped will likely just be reinstalled from the recovery cd that came with the machine, thus returning them to the same vulnerable state they were in before - ready to be reowned.
Incidentally, if you've ever looked at a compromised machine, there's typically lots of different pieces of malware on them, most infected boxes tend to be shared between several groups and some end up a battleground between competing groups trying to remove each others' malware.
Have a look at XdmX... It lets you run a big panoramic display spread over multiple X servers, and you can drag stuff across... You could potentially combine it with a clustering setup like openmosix so that the actual running process moves as well as the window.
I would say windows is far more disfunctional when it comes to high resolution screen management with lots of small windows... The fonts don't scale according to screen DPI, and windows is very much geared towards having one app running full screen at a time - and many of those apps (and many poorly designed websites etc) just look stupid when used at a high resolution. The mainstream Linux window managers suffer from trying to be too much like windows and having many of the same flaws... You really need a completely different WM for a small low resolution netbook than for multiple large high resolution screens, one size does not fit all.
You can buy Sun and SGI CRT monitors very cheaply these days, and they go up to 24" in size... They tended to be very good screens because they were intended for high end workstations. I used a 21" Sun for years, it took several years before i had a machine powerful enough to drive it at it's maximum resolution.
I have an old SGI 1600SW 17" monitor, which does 1600x1024 natively (widescreen) which is very similar in size and res to the older 17" macbook pro (1680x1050)... The new macbook pro 17" does 1920x1200 on the same size screen so it must be possible to make a standalone monitor at that res/size...
One of the biggest things holding back resolutions tho, is as always, windows... windows doesn't detect the monitor dpi properly, so a larger screen at the same resolution will result in everything getting bigger, while a higher resolution just results in everything getting smaller (and often unreadable) rather than simply becoming more detailed.
If it has platform specific libraries which many people use and are encouraged to use, then the end result is anything but cross platform... It's just another form of lock-in.
For.NET - most definitely no. It was about language and architecture (x86/amd64/ia64/...) independence.
Yes, it was designed to ease the transition to 64bit machines while maintaining as much lockin to windows as possible... Going along with it just diverts resources away from java, which actually is designed to be cross platform, is fully open and is far more mature.
The mono libs (c) won't be installed by default on windows systems, or supported out of the box by the microsoft development tools, so noone will use them... Most developers who write applications using.net/c# will target them at windows and then possibly consider other platforms as an afterthought, and most such apps won't even run using mono...
As a cross platform system, mono and.net suck pretty badly, not like java where many apps written by people who never anticipated them being used on anything but windows, but which run perfectly well on linux/mac/solaris.
They have to push out a new version every year so they have something new and flashy for people to pay to upgrade to... That's the problem when you buy software from an entity that needs to keep you buying future versions to ensure a revenue stream.
Releasing the source works out extremely beneficial for them... By the time a game gets opened up, it has very little value as a commercial game anymore, but look at any modern platform that has been cracked or released open - a port of quake or doom is one of the first things to spring up. So something that has virtually no commercial value now becomes a free advertising platform and keeps your name prominent. Most games from the same era as quake are languishing as abandonware and occasionally being played under dosbox, quake runs natively on virtually anything these days. It's also only the engine that's open, the data files are not, so you can either use third party data files (like urban terror and openarena), the original demo files or buy the original data files (you will usually be able to find a dirt cheap copy of the game in a bargain bin somewhere).
I think all game companies should do this, having the source to old games is good for everyone involved and far better than games becoming abandonware that won't even run on modern systems without some form of emulation.
The problem is the games, there is no reason you couldn't connect a keyboard and mouse to the usb ports present on all modern consoles, games just need to be written to support them... And if the game is multi platform, it would presumably already have that support had they bothered to enable it for the console builds.
You can get more powerful hardware in a PC, but... Many people don't have cutting edge hardware, many people i know use hardware less powerful than an xbox360 or ps3. Because of this games cannot require the latest hardware or they lose a significant chunk of potential customers. The OS overhead wastes a significant amount of that power, especially if you run any additional background apps like av... Build a PC of comparable specs to a first gen xbox and try playing halo on it...
Console makers are shooting themselves in the foot... All the modern consoles come with USB ports, and most modern keyboards and mice are USB... There is nothing to stop someone connecting a keyboard/mouse to a console and using it to play games... As for modding, games would just need to be designed for that purpose, consoles come with hard drives, usb and memory card slots these days so it wouldn't be hard to have them look there for data files. Not sure what to say about small independent games, the console makers want to maintain an iron grip over their consoles, sony let you run mostly what you want but restrict access to the video hardware making it largely useless for independent games, and ms/nintendo don't even allow that. Perhaps they should open up the system but restrict unofficial games to being free unless you market them through official channels, and slap a big disclaimer on the screen before such a game boots...
Aside from that, as you pointed out, the PC is a terrible gaming platform... Mac would make a somewhat better platform because there are far less possible hardware configurations but having a full blown OS running under your game is far from efficient. The Amiga had the right idea, limited number of hardware configs, full featured OS available if you wanted it but games could also take over the hardware directly like on a console.
You can achieve the same level of reliability fairly cheaply, what's expensive is the throughput, which isn't necessary unless you want to support a large number of users, and if you have lots of paying customers you should be able to afford the highend kit.
Cloud computing is in essence one step worse than proprietary software, in that not only is your data locked up in proprietary formats but it's now hosted on someone else's servers too, making you even more dependent on the service provider. On the other hand, unlike software, they are providing a service with contracts guarantees... I would demand a guarantee of a certain level of uptime, and a guarantee that i can always take my data out in a standard format if i want/need to. Very few proprietary software guarantees you the ability to retain your data in a standard format that can be imported into a competing product or service.
Upper management are usually extremely non technical, which brings up a lot of problems...
As you pointed out, they are no good at identifying if someone else is technically competent... Aside from that, they are more likely to trust marketing literature and advice from people they play golf with (who are likely to have vested interests)...
Yes i will second that, responsiveness is one of the most important aspects to traders... They have special keyboards tailored to their particular trading applications so you can enter trades quicker... They have dedicated lines between sites because a vpn going over the internet would be slower... They use unencrypted and often unauthenticated protocols to reduce the overhead. They intentionally use very short or no passwords so they are quicker to enter... Security, cost, all secondary factors to the need for low latency.
MS used to run lots of ads, including banner ads on slashdot, about how the london stock exchange chose windows over linux... Those ads stopped very quickly when they had the big outage a few months ago.
Slashdot Mirror
That's the point tho...
XP replaced 98/ME (from the consumer's perspective), and offered a significant improvement over it, prompting them to change...
Vista replaced XP, but did not offer a significant improvement and in many cases was actually worse...
People resisted changing to vista, but by the time 7 comes out they will often have little choice.
Being "standardized" is both a good and a bad thing... One size does not fit all, and a monoculture is no good for anyone... The ability to select the best tool for the job rather than having to use a "standard" is a good thing. Data should be standardized, but how you interact with it should be a matter of choice... Roads are standard, tv signals are standard, but people drive all kinds of different vehicles and use different types of tv.
But to answer your question:
Windows makes a terrible server platform, you can't strip it down to the bare minimum, you can't install and manage it from a serial console... and don't mention the "cli only mode" in windows 2008, their idea of cli only is to load the entire gui layer and then put a cmd.exe window in the middle of it... what was the point in loading the gui layer with all it's overhead just to display a cli? a windowed cli will never work over a serial console either...
On a desktop system the interface is extremely clunky, and is very much geared towards doing things their way or nothing... Their way doesn't suit me, the default ways of most linux distros don't suit me either but linux is much easier to customize.
Linux is easy to use without a command line, modern distros will let you do everything most users will ever want without a command line... And yet, many seasoned users actively choose to use the command line on linux.. Why? because in many cases it's easier, much easier for an experienced user, and much easier to explain when trying to help an unskilled user. Windows users, even experienced ones rarely use the command line mostly because the windows cli is pretty bad, but one counter example is when someone doing phone support wants an ip address from a windows user they're supporting, they almost always have them open a command prompt and type "ipconfig"... Why? because that's easier than finding the IP through the gui (which i assume can be done somehow).
Because of this people get the impression linux is only usable from the cli, when in reality the cli is often the best but not the only way to do many things...
And when it comes to advanced things, a cli where you can cut+paste is much easier than regedit...
Another issue is package management, windows simply doesn't have one, on linux i can just open my package manager, search for what i want and hit install, and what i want is installed including any dependencies it has, or i can do it from the cli. Windows requires you to manually find what you want through google, trust that the download site you find is reputable (when was the last time a windows user downloaded a file from a random download site and then compared the checksums with a set published by an official site?), and then wait for the download to finish before you can manually execute and follow through with the installer. That is just a HUGE pain in the ass.
Multiple workspaces - i cant live without multiple workspaces, and all the windows implementations i've seen have sucked, mostly because no apps or even the basic window manager are designed with workspaces in mind. And yes, aside from workspaces i can't stand the way the windows window manager works, i find it clunky and inflexible.
Foreign filesystems - linux comes with support for all kinds of filesystems out of the box, windows just doesn't, and what third party filesystem drivers do exist are often poorly implemented, buggy or both... I have to read disks from macs, bsd and linux boxes all the time and occasionally misc other systems, windows just doesn't cut it, they arrogantly only support their own filesystems.
Source - I want the ability to modify the source code of the programs i use, to do things the authors never intended... I also want to be able to use new and exciting hardware, many years ago i used alpha, more recently i was using 64bit amd64 very early on and these days i would be looking at low power arm based systems, microsoft have always been playing catch up and the closed source nature of most windows apps make
If you sign up for and pay for a service, then Google are obliged to continue providing that service for the term of the contract providing you keep up your end of the bargain (ie paying for it)...
They are also obliged to abide by any other terms in the contract, so as a business you would be foolish to not demand clauses that forbid Google from mining your data or doing anything else with it not directly related to providing the service.. You should also demand the ability to download all of your data at any time you wish in a standard format so that you can keep your own backup and/or have the data available to you for migrating to another service.
Companies trust their critical data and resources to other companies all the time... There are companies who specialize in storing or destroying massive quantities of hard copy documents for instance, not to mention courier or security companies.
This is actually a better situation than using MS apps, which come with no warranty and often provide no method or guarantee to get the data out in a standard format.
Obviously the best plan is to have the data on your own hardware, in formats which you have full documentation for, and using software which you have an unlimited irrevocable right to use for any purpose, but neither MS nor Google offer this right now.
I have seen a LOT of big companies which are doing everything you suggest, virus scanners on inbound email and the web proxy with no access to the outside world. They usually block the download or receiving via email of executables of any kind, or any kind of encrypted data (since they cant check the contents of it).. They also tend to either disallow removable devices entirely, or limit them to certain users and enforce virus scans on any inserted devices.
And yet these companies still get infected, new malware gets straight through their web filtering because it's newly written and the company providing the filter hasn't added it to their signatures yet... Lots of malware also gets through the email filter because it's embedded in proprietary formats like word and excel.. These files aren't blocked because external companies often send files in such formats, which are hard to parse and new malware is constantly being developed. All it takes is for one user to get infected, and their machine can be a source of infection for everything else... Most companies protect their borders, but the inside is extremely insecure. They use shared authentication schemes like active directory because they have large numbers of machines to manage, but still don't have the resources to ensure everything is up to date and securely configured so it's usually trivially easy to compromise a machine or two, make your way onto the domain (made easier by it running the same software and network visible apis as the member systems) and take the whole network very easily...
Most of the issues are due to outlook only really supporting it's own proprietary protocols for most things... They are designed to lock you in to exchange. That's why their support for IMAP is lousy, and they have no support for anything like CalDAV at all.
You won't be trimming off the low hanging fruit, you will simply be raising that fruit a few inches higher... And because less people will now be picking it, those people who are a bit taller will now have a lot more easily accessible fruit to pick.
What would happen, is that the prevalence of unskilled script kiddies would massively decrease, and the background scans taking place constantly would decrease... Because the perceived threats would have abated, people wouldn't bother installing updates or taking any measures to protect themselves. Also without public disclosure and/or active exploitation, software vendors would downplay the seriousness of their vulnerabilities and delay providing patches for them.
The end result of this, is that the smaller number of people who can acquire exploits, and this includes paid criminal gangs, would have a lot more power because they would no longer have to compete against the script kiddies for control of drone systems.
Incidentally, i am also against the *free* disclosure of vulnerabilities in non free software... Commercial vendors charge you a lot of money for their software, and can often be hostile or uncommunicative towards people who find bugs in their software... These people finding bugs are effectively doing their jobs for them and get nothing but grief in return, so it's no wonder that so many bug hunters are now working for criminal gangs.
A lot of these vendors want you to do their beta testing for them for free, and then report the bugs privately to them so they can silently fix them not even giving you credit for the find and often not disclosing any details to the public other than perhaps providing a black box patch.
Or for a blackhat, what better way to divert the blame?
Bots are plentiful, insecure windows boxes are extremely abundant and it will be easy for them to acquire more, they probably haven't even diverted all of their current resources to this attack.
The machines that get wiped will likely just be reinstalled from the recovery cd that came with the machine, thus returning them to the same vulnerable state they were in before - ready to be reowned.
Incidentally, if you've ever looked at a compromised machine, there's typically lots of different pieces of malware on them, most infected boxes tend to be shared between several groups and some end up a battleground between competing groups trying to remove each others' malware.
Have a look at XdmX... It lets you run a big panoramic display spread over multiple X servers, and you can drag stuff across...
You could potentially combine it with a clustering setup like openmosix so that the actual running process moves as well as the window.
I would say windows is far more disfunctional when it comes to high resolution screen management with lots of small windows... The fonts don't scale according to screen DPI, and windows is very much geared towards having one app running full screen at a time - and many of those apps (and many poorly designed websites etc) just look stupid when used at a high resolution.
The mainstream Linux window managers suffer from trying to be too much like windows and having many of the same flaws... You really need a completely different WM for a small low resolution netbook than for multiple large high resolution screens, one size does not fit all.
You can buy Sun and SGI CRT monitors very cheaply these days, and they go up to 24" in size... They tended to be very good screens because they were intended for high end workstations. I used a 21" Sun for years, it took several years before i had a machine powerful enough to drive it at it's maximum resolution.
I have an old SGI 1600SW 17" monitor, which does 1600x1024 natively (widescreen) which is very similar in size and res to the older 17" macbook pro (1680x1050)... The new macbook pro 17" does 1920x1200 on the same size screen so it must be possible to make a standalone monitor at that res/size...
One of the biggest things holding back resolutions tho, is as always, windows... windows doesn't detect the monitor dpi properly, so a larger screen at the same resolution will result in everything getting bigger, while a higher resolution just results in everything getting smaller (and often unreadable) rather than simply becoming more detailed.
If it has platform specific libraries which many people use and are encouraged to use, then the end result is anything but cross platform... It's just another form of lock-in.
For .NET - most definitely no. It was about language and architecture (x86/amd64/ia64/...) independence.
Yes, it was designed to ease the transition to 64bit machines while maintaining as much lockin to windows as possible... Going along with it just diverts resources away from java, which actually is designed to be cross platform, is fully open and is far more mature.
The mono libs (c) won't be installed by default on windows systems, or supported out of the box by the microsoft development tools, so noone will use them... Most developers who write applications using .net/c# will target them at windows and then possibly consider other platforms as an afterthought, and most such apps won't even run using mono...
As a cross platform system, mono and .net suck pretty badly, not like java where many apps written by people who never anticipated them being used on anything but windows, but which run perfectly well on linux/mac/solaris.
Having root is far more valuable than having the source code...
You can steal the source with root...
You could steal other things with root...
You could disrupt/modify trades with root...
The source code will only teach you so much...
They have to push out a new version every year so they have something new and flashy for people to pay to upgrade to... That's the problem when you buy software from an entity that needs to keep you buying future versions to ensure a revenue stream.
Releasing the source works out extremely beneficial for them... By the time a game gets opened up, it has very little value as a commercial game anymore, but look at any modern platform that has been cracked or released open - a port of quake or doom is one of the first things to spring up. So something that has virtually no commercial value now becomes a free advertising platform and keeps your name prominent.
Most games from the same era as quake are languishing as abandonware and occasionally being played under dosbox, quake runs natively on virtually anything these days.
It's also only the engine that's open, the data files are not, so you can either use third party data files (like urban terror and openarena), the original demo files or buy the original data files (you will usually be able to find a dirt cheap copy of the game in a bargain bin somewhere).
I think all game companies should do this, having the source to old games is good for everyone involved and far better than games becoming abandonware that won't even run on modern systems without some form of emulation.
The problem is the games, there is no reason you couldn't connect a keyboard and mouse to the usb ports present on all modern consoles, games just need to be written to support them... And if the game is multi platform, it would presumably already have that support had they bothered to enable it for the console builds.
You can get more powerful hardware in a PC, but...
Many people don't have cutting edge hardware, many people i know use hardware less powerful than an xbox360 or ps3. Because of this games cannot require the latest hardware or they lose a significant chunk of potential customers.
The OS overhead wastes a significant amount of that power, especially if you run any additional background apps like av... Build a PC of comparable specs to a first gen xbox and try playing halo on it...
Console makers are shooting themselves in the foot... All the modern consoles come with USB ports, and most modern keyboards and mice are USB... There is nothing to stop someone connecting a keyboard/mouse to a console and using it to play games...
As for modding, games would just need to be designed for that purpose, consoles come with hard drives, usb and memory card slots these days so it wouldn't be hard to have them look there for data files.
Not sure what to say about small independent games, the console makers want to maintain an iron grip over their consoles, sony let you run mostly what you want but restrict access to the video hardware making it largely useless for independent games, and ms/nintendo don't even allow that. Perhaps they should open up the system but restrict unofficial games to being free unless you market them through official channels, and slap a big disclaimer on the screen before such a game boots...
Aside from that, as you pointed out, the PC is a terrible gaming platform... Mac would make a somewhat better platform because there are far less possible hardware configurations but having a full blown OS running under your game is far from efficient. The Amiga had the right idea, limited number of hardware configs, full featured OS available if you wanted it but games could also take over the hardware directly like on a console.
You can achieve the same level of reliability fairly cheaply, what's expensive is the throughput, which isn't necessary unless you want to support a large number of users, and if you have lots of paying customers you should be able to afford the highend kit.
Cloud computing is in essence one step worse than proprietary software, in that not only is your data locked up in proprietary formats but it's now hosted on someone else's servers too, making you even more dependent on the service provider.
On the other hand, unlike software, they are providing a service with contracts guarantees... I would demand a guarantee of a certain level of uptime, and a guarantee that i can always take my data out in a standard format if i want/need to. Very few proprietary software guarantees you the ability to retain your data in a standard format that can be imported into a competing product or service.
Upper management are usually extremely non technical, which brings up a lot of problems...
As you pointed out, they are no good at identifying if someone else is technically competent...
Aside from that, they are more likely to trust marketing literature and advice from people they play golf with (who are likely to have vested interests)...
Yes i will second that, responsiveness is one of the most important aspects to traders...
They have special keyboards tailored to their particular trading applications so you can enter trades quicker...
They have dedicated lines between sites because a vpn going over the internet would be slower...
They use unencrypted and often unauthenticated protocols to reduce the overhead.
They intentionally use very short or no passwords so they are quicker to enter...
Security, cost, all secondary factors to the need for low latency.
MS used to run lots of ads, including banner ads on slashdot, about how the london stock exchange chose windows over linux... Those ads stopped very quickly when they had the big outage a few months ago.