Ok, my 'experement' was initially intended to see how easy it would be to work out some very basic matrix, clicking a button. The idea was to work around familiarity so that the data was usefull for raw UI testing, things like familiarity would then show up as good UI design since they could be compaired to the basic matrix. I intended to extend the software to measure all kinds of UI, learning curves etc.... so that OSS developers would have something they could use as a reference to say my UI is 'better' that X.
I never have managed to get gestures working properly, this is partly because most gesture packages aren't that great at vector / shape processing and partly because I'm a bit of an uncoordinated Muppet at times. I've also found that the keyboard is fast enough and don't often use the mouse except for cut and paste.
'You've completely missed the point. It's not whether it's quicker to hit the corners, it's that it's much easier to hit the corners.'
Well, firstly why isn't it any faster if it's 'easier'?, secondly after reaching the corner I'd expect you to move you mouse somewhere else to do something else so moving to the corner is a distraction. (I didn't get around to testing distractions)
To maximize the window I press F11, which is probably easier and faster than anything that requires a mouse. (again not measured)
For the odd occasion when I have my eyes closed, standing on my head and want to maximize the window I'll think about the corners of the screen, for every other occasion if probably makes no difference.
Well, I'll hopefully release the test software as soon as I get some spare time.
My intent was to produce some stats on the very basics of user interfaces so that they could be used to evaluate more complex interfaces. The first test was designed to look at how long it took people to click on something.
I started out fairly basic, just a box with that appeared randomly on the screen, and then moved up to having boxes that appeared in ordered patterns and at given locations on the screen (including points in the corners), the given location tests where mixed up with random locations to make it a little more realistic.
I was looking to measure a learning curve as the user 'learnt' the location of the fixed boxes but didn't get enough data for proper analysis.
For all tests I recorded the time of the mouse click and the location of the mouse click.
I looked at things like the change in time over time, and looked to any patterns that related to the ordering of the boxes etc...
In the end, after a little practice all the results were showing a straight line (least squares fit) with reasonable t test results for the correlation between the line and the data. Removing points that had exceptionally large times (where the operator had paused) gave an even better fit.
More detailed analysis of the corners of the screen showed that they were no better than anywhere else (for a small box)
So, unless your UI is only made up of points (I never got around to looking at critical sizes of elements) the corners will probably be faster, but if it's made up of anything else then there's no difference.
The tests could have been a bit more scientific, but I was running the project at home with the help of a few friends before looking to take it to a wider audience and expand it further.
I actually wrote an application that timed how long it took to click on a small red box with the word click me written on it (distance / time)
After doing the math you could nicely fit a straight line to the points, I even tried splitting out the results based on the direction of movement and their was very little difference and setup a test to explicitly test the 'corner of the screen' theory.
In the end it was no quicker to reach the corners of the screen than a small box anywhere else on the screen. That it probably why no one utilizes the corners of the screen in the way suggested.
I wrote a few more tests and was going to put together a Java applet so that world + dog could help out. Things like giving your menu entries sensible names and keeping things consistant were far more important for novice and experienced users. I was also looking at things like colour coding, 'vanishing' and growing buttons and other UI elements depending on how often they were used etc...
The main reason for the lack of good user interfaces is that no one ever seems to o solid scientific testing on them, the kind of testing that proves innovations in UI outclass current designs instead of relying on a designers hunch.
That's just it, if you can't use a product in 15 years time you haven't brought it you've just licensed it temporaraly under the delusion that you purchased it. It'd be an easy job being an antiques dealer in a hundred years time, since they won't have to learn anything about the past hundred years where everyting had a lifespan of sub 15 years.
Well if this is the same material that was reported about a week ago everywhere else (and probably/.) it's not strong enough for the space elevator (Aggregated diamond nanorods have a modulus of 491 gigapascals (GPa), compared with 442 GPa for conventional diamond.)
let me add that "forcing on their customers" is a bit like saying that Microsoft is "forcing windows on windows users."
Well, it come pre-installed on most PC's, hell I even have a couple of Windows disks. This is pretty much like DRM coming with the music that you buy and you notice until you try to do something that the DRM prevents, just like you won't notice DRM on an EBook until you sight starts to fail and you need it to be read to you.
As for SUV's just give the people what the marketing folks tell you. It's like wallmart selling crack and then saying that they didn't force anyone to buy it.
One of the key problems with DRM, not mentioned in the article, is that it prevents things from being released into the public domain unless you crack the DRM and the DMCA prevents many people from even attempting that.
HTTP has multiple transfer modes, some transfers are sent back in chunks and some are sent back in one stream, this turned out to be a problem when I wrote a spider about 8 years ago... the pronlem was related to the HTTP implementation and not the HTTP protocol itself, maybe servers are better behaved nowadays, RFC aren't any easier to interpret though so I doubt it.
Why ask for a bigger budget to the USPTO? If their aren't enough staff then they obviously need to charge more for applying for or maintaining patents. You could even do this until the USPTO starts to propup some of the bidget deficite, and maybe employ another thousand or so staff just to check for prior art after the patent has been granted.
Just think of the savings as the courts will have less crappy patent disputes and companies will stop wasting the money you spend on crap.
FTP uses two ports, 21 and 20; 21 is for control, 20 is for data transfers at the very least, port 21 has to be forwarded from the firewall to the server? for a NATs client? I run nats at home and have 'NEVER' had to forward a port for FTP, I have had to forwards a port and reconfigure my BT client for BT.
I think two ways of activating a connection is better, since it allows people to work from behind the safety of NATS and a firewall, BT doesn't have this as part of the protocol which is why you have to forward a port through the firewall making you system less secure.
' things can be further complicated; you never know whether active or passive will be the default.'
That's because you generally don't have to know, you didn't know that this is a short coming of BT which makes BT clients just as bad as FTP clients, and the BT protocol worse. Generally so long as the connection works no one is going to care, I've never had to configure my FTP client in any special way and I'm behind NAT's without and FTP port forwarding.
'though you will generally get faster speeds if you have a port forwarded to your computer.'
I think 5k vs 100k deserves a bit more emphasis than 'generally get faster speeds'
'Less than 5. I really doubt that anyone *needs* to edit their registry ever.'
Let me guess, you just format the HDD and reinstall windows instead.
Here's a couple of unavoidable reasons to edit the registry that most if not all people will come accross: 1. On Windows 95 (and probably 98) the size of the registry was limited and apps where a bit crap at using it, the result was you'd run out of registry space and nothing would install properly!! The answer was a manual registry edit.
2. Ahh Drivers!, drivers leave their mucky fingerprints all over the registry it used to be the case (Windows 2000) and possibly may still be that the only way to properly remove any knowledge about a driver from windows was to edit the registry and remove all the crap. Infact it's probably a good idea to do that every time you upgrade a driver or remove something from the system. 3. Remove all the overhead junk that windows uses (like a shadow under the mouse etc...)
Now you may be lucky and have fluked a system where all the drivers allways work and you never had to reinstall anything and you may like Windows eye candy just the way it is but the majority of people seem to have problems that usually result in reinstalling windows when a simple registry edit would do.
BTW, Their are losts of tools for editing most configuration files (some are more of the 'pro' / config file wrapper type) It's a very rare occasion that I have anything that requires editing that isn't accessable by some configuration tool other than vim (webmin's fairly complete) but I usually chose to edit with vim because then I know what's been changed and how to fix it if it becomes a problem.
The only two files I have had to edit by hand were.bashrc and.cvsrc, in the case of.bashrc all I had to do was uncomment one line.
Appart from the problems you've listed one of the largest problems with FTP is that there's no process for sending 'general' metadata about a file, this is probably the fault of most (all?) filing systems that cannot store general metadata about a file.
Oh, and I've never had any problems getting FTP to work when the client is behind a filewall / NATS and getting the server working is just a case of forwarding the port through the firewall and that's defiantly not the fault of FTP but an effect that most people would desire having placed their server behind a firewall.
Also, BitTorrent is no better (infact last time I checked it's worse for NAT clients, since you need to forward a port from to the client if the client wants good download speeds)
I would give up art and move over to being a surgeon with your pixel perfect, shake free hands.
It can be easier if you have a grid on.
I though this was supposed to be about usability, and it's even easier still if you have a type in box to control the points, and what's more you wouldn't have to switch between 'modes' modal interfaces are generally the worst (even for people who use strange pointing devices or their voice to control the PC)
Don't forget you can count every card you see to help with those stats a little, and you can make quite a few assumptions about hand that are folded straight away (unless for some odd reason people are folding very strong hands).
If you can see other peoples hands then counting becomes much easier.
The bottom line is, there is no right move for every situation.
The idea is to maximize of knowing what a right move is and what a wrong move is.
Well, it's probably better to ask the pro player who posted the origional comment but:
If you take a simpler game with similar problems (unknown cards) like Black Jack their are well known counting techniques that even without knowing what the other player has in hand can be used to win at Black Jack.
Poker is just the same, just a bit more complicated.
You can also try to read the other player and try to guess what they have in their hand based on the 'personality' you believe they have.
Spam filters are pretty good at distinguishing between a spam email and a genuine one and I should imagine similar techniques could be applies to poker (they've already taught 'spam filters' to play chess!)
In the end it's all down to statistics, if I do X then I have Y chance of winning, and the great thing about statistics is that you don't have to know every case to still be able to make a very good 'guess'.
"Not true in poker. There is unknown information (your opponents hole cards), and thus, no perfect way to play."
That's just statistics, in chess you don't know what move your opponent is going to make do you? so that's also hidden.
In poker you can try to guess what the other player has by their actions and that's just statistics, one thing computers are far better at working with than humans.
In the end a player can't outsmart a computer because it works with pure statistics, chess players have tried to use very odd tactics to out smart the computer and still lost, I expect(know) exactly the same thing will happen in poker. Don't think your game is somehow holy, people are fairly easy read and manipulate.
Please see Linus's posting on LKML for the truth behind this.
I think you should see abstraction filtration comparison along with the fact that the GPL is a copyright license to work out what's a derived work or not and not relly on legal advise from someone who's not a lawyer (Yes and IANAL!!)
'when you GPLv2 some code, you are probably granting the right to dynalink proprietary modules to your work.'
Yep, I agree google for 'Abstraction, Filtration, Comparison' and you should have no doubt that dynamic linking and GPLv2 are ok. This guy is obviously worried that the GPL doesn't allow linking (as many people seem to be) so I advised that he use the GPL with a clause that allows for plugin modules.
The kernel license says...
"NOTE! This copyright does *not* cover user programs that use kernel
services by normal system calls - this is merely considered normal use
of the kernel, and does *not* fall under the heading of "derived work".
I think it would be ok for the person who asked the question to put a similar clause in his license and use GPL.
Slashdot Mirror
Ok, my 'experement' was initially intended to see how easy it would be to work out some very basic matrix, clicking a button. The idea was to work around familiarity so that the data was usefull for raw UI testing, things like familiarity would then show up as good UI design since they could be compaired to the basic matrix. I intended to extend the software to measure all kinds of UI, learning curves etc.... so that OSS developers would have something they could use as a reference to say my UI is 'better' that X.
'If anyone uses mouse gestures for browsing'
I never have managed to get gestures working properly, this is partly because most gesture packages aren't that great at vector / shape processing and partly because I'm a bit of an uncoordinated Muppet at times. I've also found that the keyboard is fast enough and don't often use the mouse except for cut and paste.
'You've completely missed the point. It's not whether it's quicker to hit the corners, it's that it's much easier to hit the corners.'
Well, firstly why isn't it any faster if it's 'easier'?, secondly after reaching the corner I'd expect you to move you mouse somewhere else to do something else so moving to the corner is a distraction. (I didn't get around to testing distractions)
To maximize the window I press F11, which is probably easier and faster than anything that requires a mouse. (again not measured)
For the odd occasion when I have my eyes closed, standing on my head and want to maximize the window I'll think about the corners of the screen, for every other occasion if probably makes no difference.
Well, I'll hopefully release the test software as soon as I get some spare time.
My intent was to produce some stats on the very basics of user interfaces so that they could be used to evaluate more complex interfaces. The first test was designed to look at how long it took people to click on something.
I started out fairly basic, just a box with that appeared randomly on the screen, and then moved up to having boxes that appeared in ordered patterns and at given locations on the screen (including points in the corners), the given location tests where mixed up with random locations to make it a little more realistic.
I was looking to measure a learning curve as the user 'learnt' the location of the fixed boxes but didn't get enough data for proper analysis.
For all tests I recorded the time of the mouse click and the location of the mouse click.
I looked at things like the change in time over time, and looked to any patterns that related to the ordering of the boxes etc...
In the end, after a little practice all the results were showing a straight line (least squares fit) with reasonable t test results for the correlation between the line and the data. Removing points that had exceptionally large times (where the operator had paused) gave an even better fit.
More detailed analysis of the corners of the screen showed that they were no better than anywhere else (for a small box)
So, unless your UI is only made up of points (I never got around to looking at critical sizes of elements) the corners will probably be faster, but if it's made up of anything else then there's no difference.
The tests could have been a bit more scientific, but I was running the project at home with the help of a few friends before looking to take it to a wider audience and expand it further.
Over at the reg
they have a better write up, the cause is to stop people like SCO doing an SCO and not to stop people like IBM doing an IBM.
The crapomiter should have gone straight to red on the 'all patent holders cannot use GPL' bit, that's just stupid.
Chock of shit, well almost.
I actually wrote an application that timed how long it took to click on a small red box with the word click me written on it (distance / time)
After doing the math you could nicely fit a straight line to the points, I even tried splitting out the results based on the direction of movement and their was very little difference and setup a test to explicitly test the 'corner of the screen' theory.
In the end it was no quicker to reach the corners of the screen than a small box anywhere else on the screen. That it probably why no one utilizes the corners of the screen in the way suggested.
I wrote a few more tests and was going to put together a Java applet so that world + dog could help out.
Things like giving your menu entries sensible names and keeping things consistant were far more important for novice and experienced users. I was also looking at things like colour coding, 'vanishing' and growing buttons and other UI elements depending on how often they were used etc...
The main reason for the lack of good user interfaces is that no one ever seems to o solid scientific testing on them, the kind of testing that proves innovations in UI outclass current designs instead of relying on a designers hunch.
That's just it, if you can't use a product in 15 years time you haven't brought it you've just licensed it temporaraly under the delusion that you purchased it. It'd be an easy job being an antiques dealer in a hundred years time, since they won't have to learn anything about the past hundred years where everyting had a lifespan of sub 15 years.
PowerPoint documents aren't presentations there slide shows, anyone who's seen a good presentation should know that.
Well if this is the same material that was reported about a week ago everywhere else (and probably /.) it's not strong enough for the space elevator (Aggregated diamond nanorods have a modulus of 491 gigapascals (GPa), compared with 442 GPa for conventional diamond.)
Well, it starts up fine but the input is buggered in a really weird way.
let me add that "forcing on their customers" is a bit like saying that Microsoft is "forcing windows on windows users."
Well, it come pre-installed on most PC's, hell I even have a couple of Windows disks. This is pretty much like DRM coming with the music that you buy and you notice until you try to do something that the DRM prevents, just like you won't notice DRM on an EBook until you sight starts to fail and you need it to be read to you.
As for SUV's just give the people what the marketing folks tell you. It's like wallmart selling crack and then saying that they didn't force anyone to buy it.
One of the key problems with DRM, not mentioned in the article, is that it prevents things from being released into the public domain unless you crack the DRM and the DMCA prevents many people from even attempting that.
You can download the demo from here a review the game for yourself.
Hopefully I can get it working under wine!
No multiple transfer modes,
HTTP has multiple transfer modes, some transfers are sent back in chunks and some are sent back in one stream, this turned out to be a problem when I wrote a spider about 8 years ago... the pronlem was related to the HTTP implementation and not the HTTP protocol itself, maybe servers are better behaved nowadays, RFC aren't any easier to interpret though so I doubt it.
Why ask for a bigger budget to the USPTO?
If their aren't enough staff then they obviously need to charge more for applying for or maintaining patents. You could even do this until the USPTO starts to propup some of the bidget deficite, and maybe employ another thousand or so staff just to check for prior art after the patent has been granted.
Just think of the savings as the courts will have less crappy patent disputes and companies will stop wasting the money you spend on crap.
'Perpetuating the "hide things from the stupid user" UI philosophy only makes people less willing to learn'
Why do people always assume that everyone wants to know, or even should know how their computer works.
Do you know how a modern car works? And more to the point could you fix or improve upon one?
Aside from the world of ricers, jane double chin couldn't give too shits about how her car works, so why should she give two shits about her PC?
FTP uses two ports, 21 and 20; 21 is for control, 20 is for data transfers at the very least, port 21 has to be forwarded from the firewall to the server? for a NATs client? I run nats at home and have 'NEVER' had to forward a port for FTP, I have had to forwards a port and reconfigure my BT client for BT.
I think two ways of activating a connection is better, since it allows people to work from behind the safety of NATS and a firewall, BT doesn't have this as part of the protocol which is why you have to forward a port through the firewall making you system less secure.
' things can be further complicated; you never know whether active or passive will be the default.'
That's because you generally don't have to know, you didn't know that this is a short coming of BT which makes BT clients just as bad as FTP clients, and the BT protocol worse. Generally so long as the connection works no one is going to care, I've never had to configure my FTP client in any special way and I'm behind NAT's without and FTP port forwarding.
'though you will generally get faster speeds if you have a port forwarded to your computer.'
I think 5k vs 100k deserves a bit more emphasis than 'generally get faster speeds'
'Less than 5. I really doubt that anyone *needs* to edit their registry ever.'
.bashrc and .cvsrc, in the case of .bashrc all I had to do was uncomment one line.
Let me guess, you just format the HDD and reinstall windows instead.
Here's a couple of unavoidable reasons to edit the registry that most if not all people will come accross:
1. On Windows 95 (and probably 98) the size of the registry was limited and apps where a bit crap at using it, the result was you'd run out of registry space and nothing would install properly!! The answer was a manual registry edit.
2. Ahh Drivers!, drivers leave their mucky fingerprints all over the registry it used to be the case (Windows 2000) and possibly may still be that the only way to properly remove any knowledge about a driver from windows was to edit the registry and remove all the crap. Infact it's probably a good idea to do that every time you upgrade a driver or remove something from the system.
3. Remove all the overhead junk that windows uses (like a shadow under the mouse etc...)
Now you may be lucky and have fluked a system where all the drivers allways work and you never had to reinstall anything and you may like Windows eye candy just the way it is but the majority of people seem to have problems that usually result in reinstalling windows when a simple registry edit would do.
BTW, Their are losts of tools for editing most configuration files (some are more of the 'pro' / config file wrapper type) It's a very rare occasion that I have anything that requires editing that isn't accessable by some configuration tool other than vim (webmin's fairly complete) but I usually chose to edit with vim because then I know what's been changed and how to fix it if it becomes a problem.
The only two files I have had to edit by hand were
Appart from the problems you've listed one of the largest problems with FTP is that there's no process for sending 'general' metadata about a file, this is probably the fault of most (all?) filing systems that cannot store general metadata about a file.
Oh, and I've never had any problems getting FTP to work when the client is behind a filewall / NATS and getting the server working is just a case of forwarding the port through the firewall and that's defiantly not the fault of FTP but an effect that most people would desire having placed their server behind a firewall.
Also, BitTorrent is no better (infact last time I checked it's worse for NAT clients, since you need to forward a port from to the client if the client wants good download speeds)
Move your mouse to 10:30. Click.
I would give up art and move over to being a surgeon with your pixel perfect, shake free hands.
It can be easier if you have a grid on.
I though this was supposed to be about usability, and it's even easier still if you have a type in box to control the points, and what's more you wouldn't have to switch between 'modes' modal interfaces are generally the worst (even for people who use strange pointing devices or their voice to control the PC)
where's the singularity? surly that must count as a state of matter?
Don't forget you can count every card you see to help with those stats a little, and you can make quite a few assumptions about hand that are folded straight away (unless for some odd reason people are folding very strong hands).
If you can see other peoples hands then counting becomes much easier.
The bottom line is, there is no right move for every situation.
The idea is to maximize of knowing what a right move is and what a wrong move is.
Well, it's probably better to ask the pro player who posted the origional comment but:
If you take a simpler game with similar problems (unknown cards) like Black Jack their are well known counting techniques that even without knowing what the other player has in hand can be used to win at Black Jack.
Poker is just the same, just a bit more complicated.
You can also try to read the other player and try to guess what they have in their hand based on the 'personality' you believe they have.
Spam filters are pretty good at distinguishing between a spam email and a genuine one and I should imagine similar techniques could be applies to poker (they've already taught 'spam filters' to play chess!)
In the end it's all down to statistics, if I do X then I have Y chance of winning, and the great thing about statistics is that you don't have to know every case to still be able to make a very good 'guess'.
"Not true in poker. There is unknown information (your opponents hole cards), and thus, no perfect way to play."
That's just statistics, in chess you don't know what move your opponent is going to make do you? so that's also hidden.
In poker you can try to guess what the other player has by their actions and that's just statistics, one thing computers are far better at working with than humans.
In the end a player can't outsmart a computer because it works with pure statistics, chess players have tried to use very odd tactics to out smart the computer and still lost, I expect(know) exactly the same thing will happen in poker. Don't think your game is somehow holy, people are fairly easy read and manipulate.
Ok, I actually read /usr/src/linux/COPYING....
Please see Linus's posting on LKML for the truth behind this.
I think you should see abstraction filtration comparison along with the fact that the GPL is a copyright license to work out what's a derived work or not and not relly on legal advise from someone who's not a lawyer (Yes and IANAL!!)
'when you GPLv2 some code, you are probably granting the right to dynalink proprietary modules to your work.'
Yep, I agree google for 'Abstraction, Filtration, Comparison' and you should have no doubt that dynamic linking and GPLv2 are ok. This guy is obviously worried that the GPL doesn't allow linking (as many people seem to be) so I advised that he use the GPL with a clause that allows for plugin modules.
The kernel license says...
"NOTE! This copyright does *not* cover user programs that use kernel
services by normal system calls - this is merely considered normal use
of the kernel, and does *not* fall under the heading of "derived work".
I think it would be ok for the person who asked the question to put a similar clause in his license and use GPL.