Bingo. I've gotten really really tired of the old "user permissions" rhetoric. It doesn't help the problem at all except that it limits the amount of harm a single user can do to a multi user system. Problem is that most desktop systems in the home are accessed by a single user. Even in the enterprise world, it's more common to have a single user login per station. Privilege separation between users/admins breaks down because the value of user data is generally much greater than the value of the system configuration and program files. Most home users can just do the reformat/reinstall dance and spend some hours customizing their settings to get back to where they started (or hire someone to do it for them). Any competent IT department will most certainly have an effective backup/restore solution for getting things up and running again should a nerfarious destroy administrative data. So what's the big deal? Let's enumerate the advantages of setting up a typical multi-user system:
- Robustness: Users can't (easily) take down the entire system due to error, ignorance or deliberate malice. >> This is moot on most desktop systems.
- Safety: Users can't destroy data that is of value to others. >> This is moot in most workplaces since users generally have wide access to a LOT of shared state on mapped network drives, etc... Also think of shared resources like network, email and printers that privilege separation really doesn't protect in most applications. At the least, they can -- or programs running under their name can -- make a nuisance of themselves. So really all you get is a limitation of liability.
Unfortunately even when logged in as a user, there is a lot of value in the data that can be accessed and altered (else why have it). Damage to this smaller portion of property can do a lot of harm. Backups are only a partial solution to this problem but most users don't have the discipline to manage them. Worse, corruption may not become apparent until long after the tapes have been recycled.
- Privacy: Users can't see or access data that they shouldn't (for whatever reasons). >> This is useful in a large organization but not so much at home or on an employee's private workstation. If there's only one user account... who cares if you log in as Administrator?
What else is left? Computer security is still at a laughable state! The very idea that privileges can even be assigned to users on a static basis is broken. There must be more control over how data is created, located, manipulated, shared, verified, reconciled, and recovered so that it becomes impossible (or impractical) for malicious scripts or user error to wreak such havoc as they can now even in authenticated user domains. Until that's solved the whole admin/user issue is rather academic.
I'm going to go off on a limb and mention that Microsoft's RDP performs way better over low bandwidth / high latency connections than X does. This seems to be in large part because applications under RDP are not stalled by round trip messages as they are with X. X applications exhibit extremely long load times and are for the most part totally unusable except on a fast low-latency local network.
VNC doesn't even come close to the same level of responsiveness under the same network conditions. It tries very hard, but there's only so much that can be done with (compressed) bitmap image transfers alone.
In the long term, it would be neat if some RDP-type solution were implemented using similar techniques as VoIP. Allowing for some packet dropping and prioritization could significantly improve responsiveness over its current levels. RDP still appears to be much better than VNC at flow control. With VNC, one must wait for the entire screen dump to be transferred, even if most of it has become irrelevant due to user interactions.
Now, I've been arguing based on slow / unresponsive network effects. Why do I worry about this when 802.11 wireless implementations are usually pretty fast?
We want to avoid significant degredation when sharing a wireless network, or when physically near its fringes.
We want to run high-bandwidth multimedia applications, and it is important that their performance degrade gradually. This requires adaptive algorithms for prioritizing and queueing audio/video streams.
I see no impediment to attaching multiple smart displays to a central computer (mainframe terminal style) or to multiple computers. So sharing of resources is a problem.
Local area wireless networks are great, but I envision non-local uses for these things too.
While we're at it. Eliminate the scroll bars and any other scroll points altogether! Why should I have to move the mouse pointer to a specific point on the screen and click on some widget to do scrolling at all?
IMHO, the advent of the mouse-wheel and the dedication of the middle mouse button for scrolling purposes should set those pesky scrolling widgets on the way of the dodo.
How could it be any simpler? Just click the middle mouse button ANYWHERE within a scrollable pane, then move the mouse around, click again when done. Heck, if you make the wheel into a trackball then you don't even need the modal click to make this work.
As for scrollable views within scrollable views, pick the most enclosed one given the current mouse position -- besides, that's a horrible UI design idea.
Slashdot Mirror
Bingo. I've gotten really really tired of the old "user permissions" rhetoric. It doesn't help the problem at all except that it limits the amount of harm a single user can do to a multi user system. Problem is that most desktop systems in the home are accessed by a single user. Even in the enterprise world, it's more common to have a single user login per station. Privilege separation between users/admins breaks down because the value of user data is generally much greater than the value of the system configuration and program files. Most home users can just do the reformat/reinstall dance and spend some hours customizing their settings to get back to where they started (or hire someone to do it for them). Any competent IT department will most certainly have an effective backup/restore solution for getting things up and running again should a nerfarious destroy administrative data. So what's the big deal? Let's enumerate the advantages of setting up a typical multi-user system:
- Robustness: Users can't (easily) take down the entire system due to error, ignorance or deliberate malice.
>> This is moot on most desktop systems.
- Safety: Users can't destroy data that is of value to others.
>> This is moot in most workplaces since users generally have wide access to a LOT of shared state on mapped network drives, etc... Also think of shared resources like network, email and printers that privilege separation really doesn't protect in most applications. At the least, they can -- or programs running under their name can -- make a nuisance of themselves. So really all you get is a limitation of liability.
Unfortunately even when logged in as a user, there is a lot of value in the data that can be accessed and altered (else why have it). Damage to this smaller portion of property can do a lot of harm. Backups are only a partial solution to this problem but most users don't have the discipline to manage them. Worse, corruption may not become apparent until long after the tapes have been recycled.
- Privacy: Users can't see or access data that they shouldn't (for whatever reasons).
>> This is useful in a large organization but not so much at home or on an employee's private workstation. If there's only one user account... who cares if you log in as Administrator?
What else is left? Computer security is still at a laughable state! The very idea that privileges can even be assigned to users on a static basis is broken. There must be more control over how data is created, located, manipulated, shared, verified, reconciled, and recovered so that it becomes impossible (or impractical) for malicious scripts or user error to wreak such havoc as they can now even in authenticated user domains. Until that's solved the whole admin/user issue is rather academic.
*Grar* Sorry for this silly rant...
While we're at it. Eliminate the scroll bars and any other scroll points altogether! Why should I have to move the mouse pointer to a specific point on the screen and click on some widget to do scrolling at all?
IMHO, the advent of the mouse-wheel and the dedication of the middle mouse button for scrolling purposes should set those pesky scrolling widgets on the way of the dodo.
How could it be any simpler? Just click the middle mouse button ANYWHERE within a scrollable pane, then move the mouse around, click again when done. Heck, if you make the wheel into a trackball then you don't even need the modal click to make this work.
As for scrollable views within scrollable views, pick the most enclosed one given the current mouse position -- besides, that's a horrible UI design idea.