You would register your e-mail program as the only application that could
talk SMTP, POP3, etc. If Microsoft Word wanted to send an e-mail, your e-mail
program would pop up, ask you to authenticate yourself and explicitly send
the message.
If someone suggested this on Unix, people would just laugh - 'lose the
ability to script my whole system using my favourite glue language; no way'.
Why it seems any more appealing on Windows, I have no idea.
Simple: The common Windows-users doen't know what a script is. The malicious code
he's living with does. It's like an amputating the legs: it stops the gangrence.
And you won't notice the difference if you never tried to walk.
It may be true that pi contains every possible sequence of digits. So what? To get a message I have to use a code. But when I can choose the code as I like, I can get every message out of any fixed string of digits.
It may be more obvious for you that 1337 means "elite" than "1" meaning the DeCSS-sourcecode, "2" meaning the 2.4.8-source, and so on. But ask someone on the street what 1337 means. You will see that it is not the number that contains the information, but it is your code. "1337" is only the trigger.
Disclaimer: Since I read GEB I'm confused. Not because of the things I didn't understood, but because of the things I understood.
When I read all these articles about intuitive interfaces here, I tried to find out what I especially dislike
about the graphical UIs. I made an interresting psychological discovery (I'm talking about people with a
fair amount of experience, not newbies): when I go through a
directory-tree using a file-manager, my mind is in a permanent state of "not-knowing" - searching for files
and searching for menu-items. I certainly know what I want and how to do it, but first I have to find
it in the interface - *every time*. This induces stress, makes me angry and helpless. This is mostly
unconscious and it comes in small doses which sum up over the day, so it may be not obvious why you are
feeling bad when you come home from work. The worst thing is that I can't do anything about it, because it's
immanent to this way of computing - so I feel I'm doomed. Well, maybe I'm just too sensitive, but that's the
way I feel.
OTOH when I use the CLI, I'm always in a state of "knowing" (assuming I read the documentation). When I know
what I want and how to do it (e.g. change directory to a certain path), I *just type it in*. No searching,
no confusion, no feelings of being lost.
The data were on magnetic tapes, and written in a format so old that the programmers who knew it had died.
I bet they wrote their new reports as Word documents and in 5 years they have to crack it once again. (Don't worry - they have experts for extraterrestrial communication. First they will search for prime-numbers and signs of intelligence, but sooner or later they will manage to open the wicked file...)
1. The Borwein-formula for pi has certain properties that let you compute the n-th digit behind the comma in far less time than by the conventional method of computing all the other n-1 digits. But the computing-time still raises with n.
2. This method works for base 16, but not for base 10. As far as I know, no formula for base 10 has been found, yet.
3. Borwein's formula:
pi = sum[k=0..inf]((1/16^k)*(4/(8k+1)-2/(8k+4)-1/(8k+5) -1/(8k+6))
4. How to compute the n-th digit fast (in base 16):
Multiply the formula with 16^(n-1). Now the n-th digit is directly behind the comma. You get:
And here's the trick: We are only interested in the digits *behind* the comma, so we can calculate each nominator (16^(n-1-k)) modulo its denominator, losing the part before the comma and therefore dealing with shorter numbers which is faster. E.g. for the first term: ( 16^(n-1-k) mod (8k+1) ) / (8k+1). [hint: n mod m is the remainder of n/m]
Once again: if you have a term a/b with a>b then there's a term (b*c+d)=a with d<b [d=a mod b]. Then (b*c+d)/b=a/b and d/b is the part behind the comma and c is the part before the comma (because b/b=1).
Note that each term of the infinite sum has a *positive* value (compared to other formulas, where the terms are alternating positiv and negative). Otherwise this method couldn't work, because you couldn't be sure that the part before the comma is really not relevant.
The pi-algorithm isn't useful for generating random-numbers at all, because the processing time needed for the n-th digit raises with n.
It's much faster than calculating all the n-1 digits, but for a random-generator you need an algorithm that needs a constant time for the next digit.
Slashdot Mirror
Doesn't work for me neither (SuSE 7.2).
Just use http://127.0.0.1
If someone suggested this on Unix, people would just laugh - 'lose the ability to script my whole system using my favourite glue language; no way'. Why it seems any more appealing on Windows, I have no idea.
Simple: The common Windows-users doen't know what a script is. The malicious code he's living with does. It's like an amputating the legs: it stops the gangrence. And you won't notice the difference if you never tried to walk.
Microsoft: $0 (spent no MS-tax!)
Linux: $150 (Debian and SuSE-boxes)
It seems like the proprietary business-model is basically flawed. No money here.
It may be true that pi contains every possible sequence of digits. So what? To get a message I have to use a code. But when I can choose the code as I like, I can get every message out of any fixed string of digits.
It may be more obvious for you that 1337 means "elite" than "1" meaning the DeCSS-sourcecode, "2" meaning the 2.4.8-source, and so on. But ask someone on the street what 1337 means. You will see that it is not the number that contains the information, but it is your code. "1337" is only the trigger.
Disclaimer: Since I read GEB I'm confused. Not because of the things I didn't understood, but because of the things I understood.
There's a download-mirror at http://sourceforge.net/projects/atheos/
Don't donwload from www.atheos.cx -- Kurt's machine can't handle the load.
When I read all these articles about intuitive interfaces here, I tried to find out what I especially dislike about the graphical UIs. I made an interresting psychological discovery (I'm talking about people with a fair amount of experience, not newbies): when I go through a directory-tree using a file-manager, my mind is in a permanent state of "not-knowing" - searching for files and searching for menu-items. I certainly know what I want and how to do it, but first I have to find it in the interface - *every time*. This induces stress, makes me angry and helpless. This is mostly unconscious and it comes in small doses which sum up over the day, so it may be not obvious why you are feeling bad when you come home from work. The worst thing is that I can't do anything about it, because it's immanent to this way of computing - so I feel I'm doomed. Well, maybe I'm just too sensitive, but that's the way I feel.
OTOH when I use the CLI, I'm always in a state of "knowing" (assuming I read the documentation). When I know what I want and how to do it (e.g. change directory to a certain path), I *just type it in*. No searching, no confusion, no feelings of being lost.
Happy End.
Bah! Programmers never die, they just go sub and don't return().
How materialistic! According to my religion, old programmers don't die, they just cast to void.
The data were on magnetic tapes, and written in a format so old that the programmers who knew it had died.
I bet they wrote their new reports as Word documents and in 5 years they have to crack it once again. (Don't worry - they have experts for extraterrestrial communication. First they will search for prime-numbers and signs of intelligence, but sooner or later they will manage to open the wicked file...)
1. The Borwein-formula for pi has certain properties that let you compute the n-th digit behind the comma in far less time than by the conventional method of computing all the other n-1 digits. But the computing-time still raises with n.
) -1/(8k+6))
2. This method works for base 16, but not for base 10. As far as I know, no formula for base 10 has been found, yet.
3. Borwein's formula:
pi = sum[k=0..inf]((1/16^k)*(4/(8k+1)-2/(8k+4)-1/(8k+5
4. How to compute the n-th digit fast (in base 16):
Multiply the formula with 16^(n-1). Now the n-th digit is directly behind the comma. You get:
16^(n-1)*pi=sum[k=0..inf]( (16^(n-1-k))/(8k+1) - (2*(16^(n-1-k)/....
And here's the trick: We are only interested in the digits *behind* the comma, so we can calculate each nominator (16^(n-1-k)) modulo its denominator, losing the part before the comma and therefore dealing with shorter numbers which is faster. E.g. for the first term: ( 16^(n-1-k) mod (8k+1) ) / (8k+1). [hint: n mod m is the remainder of n/m]
Once again: if you have a term a/b with a>b then there's a term (b*c+d)=a with d<b [d=a mod b]. Then (b*c+d)/b=a/b and d/b is the part behind the comma and c is the part before the comma (because b/b=1).
Note that each term of the infinite sum has a *positive* value (compared to other formulas, where the terms are alternating positiv and negative). Otherwise this method couldn't work, because you couldn't be sure that the part before the comma is really not relevant.
The pi-algorithm isn't useful for generating random-numbers at all, because the processing time needed for the n-th digit raises with n. It's much faster than calculating all the n-1 digits, but for a random-generator you need an algorithm that needs a constant time for the next digit.