> I think you are confusing 'architecting' with 'coding'. The first step of creating a software architecture doesn't involve any code whatsoever.
I don't think I said it did. I didn't bother going into great detail of what is involved in creating a blueprint or source code, because I thought I implied it when I mentioned the cost of creating a blueprint. After all, drawing random lines to create a dummy blueprint isn't that expensive nowadays. Same for writing "hello world".
> and 40% build.
Here's where I disagree.
In software, the source code is the blueprint.
A programmer writing code is just like an architect creating a blueprint. Before the blueprint, there is some design work, top-down or bottom-up or whatever is the fad of the day, talking to external parties, figuring out what is needed, what isn't, drafts, etc. Very similar in the software world.
In the Civil Engineering build phase the workers and machines build to the blueprint/detailed design, they're not normally supposed to add 10 extra entrances or remove rooms, (yes there can be variations, but if there are too many something is wrong). You often can add more machines and workers to speed the build up (within reason).
In the software build phase, the compiler builds the object code according to the source code. It's not supposed to add 10 extra features or remove required functionality. You can often add more machines and CPUs to speed the build up (within reason).
If your 40% software build phase involves programmers creating source code (blueprint), it's really analogous to the design phase of a Civil Engineering project, not the build phase.
Still think I got it wrong?
Show me why source code isn't the equivalent of a blueprint. Or show me that creating the blueprint belongs in the build phase of a civil engineering project (not talking about the "as built" drawings/documents that describe what they actually built - as opposed to what they were supposed to build;) ).
Lots of news orgs will get a bunch of quotes from people on the street etc, and only print the ones that suit their agenda. So they didn't say the stuff, someone else said it, but...:).
And why should he bother targeting a single distro? A single distro would be a small part of a small sliver ("Linux distros") of total desktop marketshare (desktop because he said "continued failure of Linux on the desktop").
He might as well target OSX. He'd make more money.
OSX has had much greater success than "Desktop Linux".
> It would be the same as trying to build a house without blueprints, just nailing up wood at one of the corners.. It will kind of look like a house when done, but will have lots of problems. Code is the same thing.
It's not really the same thing in practice though.
You see, in the software world, the blueprints compile and kinda run... Heck the plastic models and artist's impression might even run too!
Management typically sells those to the customer as v1.0 and v2.0;).
Why? Because in the software world, each blueprint costs as much to make as the Real Thing (which is the result of building/compiling blueprint #<big number>).
That's not true in the Civil Engineering world, where the Real Thing costs 10x or more. So management is more agreeable to spend the resources and time for enough design iterations just so the Real Thing is more likely to be good enough.
Basically: In the civil engineering world, the design phase cost a lot less money (and typically less time) than the build phase (<=10%?), which involves lots of people, lots of building materials and heavy machinery. In the software world, the design phase costs most of the money and time (>90%), and the build phase involves someone doing "make all" and fetching a cup of coffee.
I believe there's a difference between "design phase" project management and "build phase" project management (the build phase of a software project is managed by the CPUs, Operating System Scheduler, compiler and packager).
But most project managers (and bosses) manage software projects the way people manage the build phase of a civil engineering project.
If they really did that, I'd say they were clueless. Such a feature would increase the odds of error.
ZFS might checksum every block. But what happens when ZFS is not everywhere? Does the BIOS or whatever equivalent support ZFS checksumming for reading the boot sectors? So those sectors better be 100% or you better be turning it off for boot drives. You have to use ZFS everywhere and for everything. For example, if you ever try to image a 1TB disk without ECC, the odds of bit errors will be high. Even if ZFS can repair it - you'd only find out much later (too late?) and likely after another error prone write.
Such a feature would just be creating more opportunities for people to get things wrong.
And for what benefit?
> The motivating idea was that this would reduce the overhead involved on ECC and gain extra space.
I think the people who'd want ZFS or RAID would rather have better reliability than the 10% or so extra space.
Is bazaar slower mainly because of the architecture? The rename tracking seems to be a potentially important difference (it may not bite you now, but if it bites you way later that can be more painful:) ).
It could still head that way. Then reviewers and review sites might become important. You don't make as much from being a reviewer. But economists might say that's a sign of the market becoming more efficient;).
Nowadays I've been getting more useful reviews from food bloggers than "professional" food reviewers.
The professional food reviewers might have a more refined palate etc, but that sometimes is the problem if you're not like them at all:).
I've considered creating a site where everyone can review anything (but the review is just a score and a very short description).
Users then pick a "review POV" that matches best their "target taste", and use it to look for stuff that "target taste" might like.
The target taste doesn't have to be what you like. You could be looking for stuff someone else might like based on what you know of that person's preferences. So you find a "review POV" that likes and dislikes the same stuff, and go from there.
The trouble is finding the math and algorithms that can handle this in a large scale. The other big trouble is I'm just a lazy person with many ideas that I never ever get around to implementing. Ideas are easy, but it takes a lot of time and effort to try and implement just one idea - that's why I think patent trolls are disgusting, and patents currently cause more harm than good.
The laws. As long as the laws say something is human it's human:).
> Perhaps it is free will,
This is very important too, from a strategic POV.
It is dangerous for humans to say stuff like: 1) It's not my fault, I have no choice - I'm born like that. 2) We have no free will 3) We are just machines
Because defective machines can be discarded a lot more easily than defective humans. So even if 1-3 are true, a wise human might want to maintain the illusion that they are special;).
Anyway, to me the direction to go is to not replace humans but to augment them. There is a subtle yet significant difference between the approaches.
With one approach you'd have more and more independent autonomous AIs in charge of stuff. With another approach, the focus is for the humans to be able to do more and more (and still be responsible for what they do). Yes there'll still be automation, but the focus is different.
I don't see a great technical need for nonhuman intelligences - we already have plenty - they're in pet shops and farms. We're not doing such a great job with them as is, so it would be rather irresponsible to try to create a real AI at this point in time.
I guess I'm lucky and don't require that 5-10 second orientation period.
It is really annoying when I know exactly what window I want, but it takes longer than it has to, to get to it.
It's like driving a car with a laggy steering wheel. You can still get from point A to point B, but no way can you consider it a good car. It's crap.
Yes, multiple big displays would help, but multiple displays aren't as portable, cost more $$$ AND most importantly you wouldn't need them as much if GUIs weren't so crap!
As I have been saying, switching a window doesn't have to be so much slower than looking at another display and refocusing. The reason switching a window is significantly slower than looking at another display is because the popular Desktop GUIs aren't helping.
I regularly have to switch amongst 3 or more windows during my work (which in the current stage involves creating and updating a bunch of design documents).
Recent example: 1) Document listing firewall rules 2) Document listing firewall objects 3) Email listing a bunch of hosts and what ports the hosts require to be allowed by the firewall. 4) Notepad - because pasting stuff into notepad and recopying said stuff from it is often faster if you want to convert stuff into plain text for easy repasting- Microsoft Office does give you paste options but they can be fiddly (in some scenarios it doesn't give you the plain text option, and it actually is visibly slow for some reason). Also you can keep a bunch of text that is commonly required to be entered. So instead of typing such stuff again, you switch to notepad (left hand does alt tab, right hand gets ready for quick selection with mouse), copy, switch back, paste, repeat if necessary. Autocomplete doesn't help if the customer wants lots of stuff to start with the same prefixes (e.g. all stuff in one data center has the same prefix).
So it could be: read part of email, add address objects, add protocol objects, insert/modify rules, go back to email read next bit, and repeat. Copying stuff to+from notepad will be involved in some parts as mentioned.
In case you're wondering, yes the customer requires documents in MS Word format.
Another scenario: 1) LAN switch port allocation - new version 2) LAN switch port allocation- old version (for reference) 3) Switch port mappings for servers and enclosures. 4) IP Address allocation document. 5) Email 6) Notepad
So I get a message that asks me to add a bunch of stuff and move existing stuff around. I need new IP addresses, need to change some IPs, need to change the ports (enclosure and switches) stuff gets attached to. And need to make sure I haven't made mistakes (and accidentally deleted or left out stuff).
Who said they'd be unrelated? Given you regularly use 30 windows, I'm sure you can easily think of a scenario where you could have 5-7 related windows. "alt tab" only helps for 2 active windows, maybe even 4, but after that it stops helping.
> I use screen every day. I fail to see how it offers me any quicker or more efficient task switching than Alt+TAB.
Screen has combo+<number> to switch to a particular screen. That's a lot quicker and less error-prone than "alt tab, tab, tab, tab".
Yes many people can get from point A to point B just by hopping. But that's suboptimal.
> If you really are wasting meaningful amounts of time waiting for your Desktop GUI, it's time to upgrade that 486.
That's not really the point. The point is to not be impeded by the GUI more than you have to be. Stuff doesn't have to be so crap.
I can understand when my legs take a while to move me from one point to another. But why should switching amongst 9 windows be slower than switching between two windows? I see no good reason for that. Especially when Desktop GUI developers have enough time to spend on "features" like "wobbly windows" and "shake window to show desktop". It's clear a lot of them are clueless[1].
I can understand why switching amongst 30 random windows would be slower since most human minds have difficulty tracking more than 7 plus/minus 2 objects without regrouping them into group-objects. But the GUI should allow quick access to "7 plus/minus 2".
The goal should be for the user to be more restricted by the limits of his/her mind than the limits of the GUI. Computers should augment the user's mind (not replace it).
Current GUIs might not restrict naive users, but they do restrict users like you who can actually tolerate having 30 open windows (I've seen users who have to keep closing and reopening/relaunching windows/applications because they either can't cope with having more than a few windows open, or don't understand how it could work).
Lastly, I suspect that user interfaces can go from functional/tolerable to "insanely great" when you make them feel more like an extension of the user's body. One way of doing that is to reduce the latency/delays. Thus the iphone's low latency "no tear" scrolling is more likely to feel like a part of a user's body, rather than a mere device.
[1] Google Chrome is another example of UI fail. When you middle click on a link to open a new tab, Google Chrome should always open the new tab _adjacent_ to the source tab. Currently the positioning of the new tab might as well be random (it's not random, but it's ridiculous). When you finally find an interesting link you want to read NOW, instead of later (like the previous new-tab links), it should be in an easily predictable spot. Google Chrome fails this. In the normal world, people find lost items in the last spot they look. In the google chrome world, people apparently keep searching for lost items after they find them.
You may not know how much you'd miss the benefits of a car or other vehicle till you've actually used one. Then you might feel almost crippled when you have no access to them.
If you've seen skilled "dumb terminal" or cash register workers operating the corresponding Application Specific UIs, you'll see that very many people can learn to operate UIs very effectively and efficiently.
With features like my proposal, a Desktop UI would allow such workers to operate multiple applications and multiple windows as quickly as if they were one application specific UI. You would not require a monolithic application designed for user-efficiency.
As it is, the popular Desktop UIs often get in the way with flashy "cutscenes" between actions. And you are required to click multiple times, or alt-tab, or winkey-<number> many times just to get to the window you want.
Try it yourself: switch amongst 4 different windows quickly, and try to immediately work with them the split second you switch to them. The four windows could be: email, documentation, editor, ssh to remote machine #1 (and even more related windows).
With most "popular" GUIs you can't even do that quickly - you have to move you hand from the keyboard and click. Windows 7 allows you to do that if they are different application windows with winkey+<number> (but the switch only happens on key-release not key-press which means it is slower).
You can do that with "screen" but that's a CLI interface. Then there are some tiling window managers that might allow you to do that, but this functionality can be in a "normal" GUI without affecting the beginner users.
I'd rather waste time on Slashdot than waste it on my Desktop GUI.
Slashdot Mirror
Nope, it was the 2D version.
Good point about the 3D though- I was considering watching the 3D version. If it's 12fps, it's going to be really bad...
> I think you are confusing 'architecting' with 'coding'. The first step of creating a software architecture doesn't involve any code whatsoever.
;) ).
I don't think I said it did. I didn't bother going into great detail of what is involved in creating a blueprint or source code, because I thought I implied it when I mentioned the cost of creating a blueprint. After all, drawing random lines to create a dummy blueprint isn't that expensive nowadays. Same for writing "hello world".
> and 40% build.
Here's where I disagree.
In software, the source code is the blueprint.
A programmer writing code is just like an architect creating a blueprint. Before the blueprint, there is some design work, top-down or bottom-up or whatever is the fad of the day, talking to external parties, figuring out what is needed, what isn't, drafts, etc. Very similar in the software world.
In the Civil Engineering build phase the workers and machines build to the blueprint/detailed design, they're not normally supposed to add 10 extra entrances or remove rooms, (yes there can be variations, but if there are too many something is wrong). You often can add more machines and workers to speed the build up (within reason).
In the software build phase, the compiler builds the object code according to the source code. It's not supposed to add 10 extra features or remove required functionality. You can often add more machines and CPUs to speed the build up (within reason).
If your 40% software build phase involves programmers creating source code (blueprint), it's really analogous to the design phase of a Civil Engineering project, not the build phase.
Still think I got it wrong?
Show me why source code isn't the equivalent of a blueprint. Or show me that creating the blueprint belongs in the build phase of a civil engineering project (not talking about the "as built" drawings/documents that describe what they actually built - as opposed to what they were supposed to build
They do that in print too.
... :).
Lots of news orgs will get a bunch of quotes from people on the street etc, and only print the ones that suit their agenda. So they didn't say the stuff, someone else said it, but
And why should he bother targeting a single distro? A single distro would be a small part of a small sliver ("Linux distros") of total desktop marketshare (desktop because he said "continued failure of Linux on the desktop").
He might as well target OSX. He'd make more money.
OSX has had much greater success than "Desktop Linux".
> It would be the same as trying to build a house without blueprints, just nailing up wood at one of the corners.. It will kind of look like a house when done, but will have lots of problems. Code is the same thing.
;).
:).
It's not really the same thing in practice though.
You see, in the software world, the blueprints compile and kinda run... Heck the plastic models and artist's impression might even run too!
Management typically sells those to the customer as v1.0 and v2.0
Why? Because in the software world, each blueprint costs as much to make as the Real Thing (which is the result of building/compiling blueprint #<big number>).
That's not true in the Civil Engineering world, where the Real Thing costs 10x or more. So management is more agreeable to spend the resources and time for enough design iterations just so the Real Thing is more likely to be good enough.
Basically:
In the civil engineering world, the design phase cost a lot less money (and typically less time) than the build phase (<=10%?), which involves lots of people, lots of building materials and heavy machinery.
In the software world, the design phase costs most of the money and time (>90%), and the build phase involves someone doing "make all" and fetching a cup of coffee.
I believe there's a difference between "design phase" project management and "build phase" project management (the build phase of a software project is managed by the CPUs, Operating System Scheduler, compiler and packager).
But most project managers (and bosses) manage software projects the way people manage the build phase of a civil engineering project.
And that's why software tends to be crappier
I guess you didn't experience this then: http://support.microsoft.com/kb/216641
For some reason I found the show too jerky during scenes with lots of rapid movement.
;).
Maybe it was due to the movie being projected at 24fps or that cinema had some problem with their projection system.
Nowadays, 24fps is "barely playable" by video game standards
If they really did that, I'd say they were clueless. Such a feature would increase the odds of error.
;).
ZFS might checksum every block. But what happens when ZFS is not everywhere? Does the BIOS or whatever equivalent support ZFS checksumming for reading the boot sectors? So those sectors better be 100% or you better be turning it off for boot drives. You have to use ZFS everywhere and for everything. For example, if you ever try to image a 1TB disk without ECC, the odds of bit errors will be high. Even if ZFS can repair it - you'd only find out much later (too late?) and likely after another error prone write.
Such a feature would just be creating more opportunities for people to get things wrong.
And for what benefit?
> The motivating idea was that this would reduce the overhead involved on ECC and gain extra space.
I think the people who'd want ZFS or RAID would rather have better reliability than the 10% or so extra space.
Even if they don't know it at first
If it's not an architectural/design issue, then the bits that are slow could be rewritten to be faster.
Is bazaar slower mainly because of the architecture? The rename tracking seems to be a potentially important difference (it may not bite you now, but if it bites you way later that can be more painful :) ).
So far what are the technical disadvantages to using Bazaar instead of git or something else?
Antibiotics can also damage you permanently. So it's just a matter of picking the lesser evil :).
If your body is losing the battle, go to the hospital and get help.
Otherwise, avoid the antibiotics if you can especially the more toxic ones. Some doctors are a lot more "antibiotic/drug happy" than others.
> They're already breaking one law, so I don't expect them to have any concerns about breaking another one to avoid getting caught.
If their friends are in Mexico when they raise false alarms in USA, are they breaking any Mexican laws?
It could still head that way. Then reviewers and review sites might become important. You don't make as much from being a reviewer. But economists might say that's a sign of the market becoming more efficient ;).
:).
Nowadays I've been getting more useful reviews from food bloggers than "professional" food reviewers.
The professional food reviewers might have a more refined palate etc, but that sometimes is the problem if you're not like them at all
I've considered creating a site where everyone can review anything (but the review is just a score and a very short description).
Users then pick a "review POV" that matches best their "target taste", and use it to look for stuff that "target taste" might like.
The target taste doesn't have to be what you like. You could be looking for stuff someone else might like based on what you know of that person's preferences. So you find a "review POV" that likes and dislikes the same stuff, and go from there.
The trouble is finding the math and algorithms that can handle this in a large scale. The other big trouble is I'm just a lazy person with many ideas that I never ever get around to implementing. Ideas are easy, but it takes a lot of time and effort to try and implement just one idea - that's why I think patent trolls are disgusting, and patents currently cause more harm than good.
> So what makes us human?
:).
;).
The laws. As long as the laws say something is human it's human
> Perhaps it is free will,
This is very important too, from a strategic POV.
It is dangerous for humans to say stuff like:
1) It's not my fault, I have no choice - I'm born like that.
2) We have no free will
3) We are just machines
Because defective machines can be discarded a lot more easily than defective humans. So even if 1-3 are true, a wise human might want to maintain the illusion that they are special
Anyway, to me the direction to go is to not replace humans but to augment them. There is a subtle yet significant difference between the approaches.
With one approach you'd have more and more independent autonomous AIs in charge of stuff. With another approach, the focus is for the humans to be able to do more and more (and still be responsible for what they do). Yes there'll still be automation, but the focus is different.
I don't see a great technical need for nonhuman intelligences - we already have plenty - they're in pet shops and farms. We're not doing such a great job with them as is, so it would be rather irresponsible to try to create a real AI at this point in time.
> and will seat eight.
Plus one goat and two chickens.
> you may want to look to them to see an alternative method of dealing with attacks.
Yeah it should be the USA - y'know, the Land of the Free and the Home of the Brave. That sort of thing.
Not the TSA - Terrified States of America:
Passengers stay in your seats, keep your hands in sight, and no pillows allowed...
Agh. Mod me down, I can't read.
He said "almost 2 people". And there are 86400 seconds per day.
Your last line is correct though.
The accounting is a bit different if the weapon is used to forcibly take resources from some other country.
I guess I'm lucky and don't require that 5-10 second orientation period.
It is really annoying when I know exactly what window I want, but it takes longer than it has to, to get to it.
It's like driving a car with a laggy steering wheel. You can still get from point A to point B, but no way can you consider it a good car. It's crap.
Yes, multiple big displays would help, but multiple displays aren't as portable, cost more $$$ AND most importantly you wouldn't need them as much if GUIs weren't so crap!
As I have been saying, switching a window doesn't have to be so much slower than looking at another display and refocusing. The reason switching a window is significantly slower than looking at another display is because the popular Desktop GUIs aren't helping.
I regularly have to switch amongst 3 or more windows during my work (which in the current stage involves creating and updating a bunch of design documents).
Recent example:
1) Document listing firewall rules
2) Document listing firewall objects
3) Email listing a bunch of hosts and what ports the hosts require to be allowed by the firewall.
4) Notepad - because pasting stuff into notepad and recopying said stuff from it is often faster if you want to convert stuff into plain text for easy repasting- Microsoft Office does give you paste options but they can be fiddly (in some scenarios it doesn't give you the plain text option, and it actually is visibly slow for some reason). Also you can keep a bunch of text that is commonly required to be entered. So instead of typing such stuff again, you switch to notepad (left hand does alt tab, right hand gets ready for quick selection with mouse), copy, switch back, paste, repeat if necessary. Autocomplete doesn't help if the customer wants lots of stuff to start with the same prefixes (e.g. all stuff in one data center has the same prefix).
So it could be: read part of email, add address objects, add protocol objects, insert/modify rules, go back to email read next bit, and repeat. Copying stuff to+from notepad will be involved in some parts as mentioned.
In case you're wondering, yes the customer requires documents in MS Word format.
Another scenario:
1) LAN switch port allocation - new version
2) LAN switch port allocation- old version (for reference)
3) Switch port mappings for servers and enclosures.
4) IP Address allocation document.
5) Email
6) Notepad
So I get a message that asks me to add a bunch of stuff and move existing stuff around. I need new IP addresses, need to change some IPs, need to change the ports (enclosure and switches) stuff gets attached to. And need to make sure I haven't made mistakes (and accidentally deleted or left out stuff).
> - a large number of of unrelated windows
Who said they'd be unrelated? Given you regularly use 30 windows, I'm sure you can easily think of a scenario where you could have 5-7 related windows. "alt tab" only helps for 2 active windows, maybe even 4, but after that it stops helping.
> I use screen every day. I fail to see how it offers me any quicker or more efficient task switching than Alt+TAB.
Screen has combo+<number> to switch to a particular screen. That's a lot quicker and less error-prone than "alt tab, tab, tab, tab".
Yes many people can get from point A to point B just by hopping. But that's suboptimal.
> If you really are wasting meaningful amounts of time waiting for your Desktop GUI, it's time to upgrade that 486.
That's not really the point. The point is to not be impeded by the GUI more than you have to be. Stuff doesn't have to be so crap.
I can understand when my legs take a while to move me from one point to another. But why should switching amongst 9 windows be slower than switching between two windows? I see no good reason for that. Especially when Desktop GUI developers have enough time to spend on "features" like "wobbly windows" and "shake window to show desktop". It's clear a lot of them are clueless[1].
I can understand why switching amongst 30 random windows would be slower since most human minds have difficulty tracking more than 7 plus/minus 2 objects without regrouping them into group-objects. But the GUI should allow quick access to "7 plus/minus 2".
The goal should be for the user to be more restricted by the limits of his/her mind than the limits of the GUI. Computers should augment the user's mind (not replace it).
Current GUIs might not restrict naive users, but they do restrict users like you who can actually tolerate having 30 open windows (I've seen users who have to keep closing and reopening/relaunching windows/applications because they either can't cope with having more than a few windows open, or don't understand how it could work).
Lastly, I suspect that user interfaces can go from functional/tolerable to "insanely great" when you make them feel more like an extension of the user's body. One way of doing that is to reduce the latency/delays. Thus the iphone's low latency "no tear" scrolling is more likely to feel like a part of a user's body, rather than a mere device.
[1] Google Chrome is another example of UI fail. When you middle click on a link to open a new tab, Google Chrome should always open the new tab _adjacent_ to the source tab. Currently the positioning of the new tab might as well be random (it's not random, but it's ridiculous). When you finally find an interesting link you want to read NOW, instead of later (like the previous new-tab links), it should be in an easily predictable spot. Google Chrome fails this. In the normal world, people find lost items in the last spot they look. In the google chrome world, people apparently keep searching for lost items after they find them.
If it doesn't involve proxying, it involves NAT.
I don't see how I'm being dishonest. I've been telling the truth. If you can't handle the truth, that's not really my fault.
As for me being a tosser, while it may be true, it should not be relevant to this discussion.
You may not know how much you'd miss the benefits of a car or other vehicle till you've actually used one. Then you might feel almost crippled when you have no access to them.
If you've seen skilled "dumb terminal" or cash register workers operating the corresponding Application Specific UIs, you'll see that very many people can learn to operate UIs very effectively and efficiently.
With features like my proposal, a Desktop UI would allow such workers to operate multiple applications and multiple windows as quickly as if they were one application specific UI. You would not require a monolithic application designed for user-efficiency.
As it is, the popular Desktop UIs often get in the way with flashy "cutscenes" between actions. And you are required to click multiple times, or alt-tab, or winkey-<number> many times just to get to the window you want.
Try it yourself: switch amongst 4 different windows quickly, and try to immediately work with them the split second you switch to them. The four windows could be: email, documentation, editor, ssh to remote machine #1 (and even more related windows).
With most "popular" GUIs you can't even do that quickly - you have to move you hand from the keyboard and click. Windows 7 allows you to do that if they are different application windows with winkey+<number> (but the switch only happens on key-release not key-press which means it is slower).
You can do that with "screen" but that's a CLI interface. Then there are some tiling window managers that might allow you to do that, but this functionality can be in a "normal" GUI without affecting the beginner users.
I'd rather waste time on Slashdot than waste it on my Desktop GUI.
> And as soon as there are more than 20 files/directories in a directory, I am on the shell, anyway.
That's more an indication of how crap the Desktop GUI or file manager is. Seriously.
Using a CLI may be better over high latency low bandwidth links, or when you are scripting stuff.
But it is a really terrible GUI if it's better to use the shell just because you need to deal with more than 20 files.
I bet gamers will still find it easier to manage hundreds of "RTS game" units with a GUI than a CLI.
Perhaps game GUI designers have more clue than Desktop GUI designers.