Discussion:
[OT] Knuth's rewards [Was: Python syntax in Lisp and Scheme]
(too old to reply)
Jacek Generowicz
2003-10-17 10:39:09 UTC
Permalink
|> Could you afford to pay $327.68 for every bug found in your glorious
|> software?
|Anyone know how Knuth arrives at the amounts he offers ?
|(He also offers $2.56 for errors in his books.)
Surely you recognize powers of two? :-).
Yes, but not, apparently, if you add a little noise, such as a "." (as
I already mentioned elsewhere).
Apparently, so far 8 errors have been found in AoCP, and 15 in TeX....
What? It can't be ... he can't be _that_ confident? That's
insane. Well, I guess that he's under no obligation to keep it up,
should things get out of hand.

(I was thinking that he maybe doubles it after a year in which no bugs
were found, and maybe halves it when one is found ... or similar)

Does he state anywhere what his policy is ?

How do we persuade Microsoft to adopt the same strategy ?
Terry Reedy
2003-10-18 22:00:43 UTC
Permalink
For more about Knuth: his home page is
http://www-cs-faculty.stanford.edu/~knuth/index.html
Jacek Generowicz
2003-10-17 13:37:26 UTC
Permalink
Funny, 32768 and 256 are numbers which ring very loud bells with me
... but inserting a "." in the middle completely suppressed this
otherwise immediate reaction.
OK, so they are powers of 2, in cents ... anyone know any further
motivation ?
A fascination with numbers / mathematics?
I meant "motivation for why those particular powers of two were
chosen, whether they evolve with time, and if so, how?"
unknown
2003-10-17 15:22:30 UTC
Permalink
Could you afford to pay $327.68 for every bug found in your glorious
software?
Anyone know how Knuth arrives at the amounts he offers ?
(He also offers $2.56 for errors in his books.)
For errors in the books, traditionally the reward was $2.00 per error.
Maybe in the more recent ones he increased it to $2.56.

For TeX, I believe he offered $2.56 per bug for the first year, $5.12
the second year, etc. until it reached $327.68 and he prudently
stopped increasing it further. He actually did pay out $327.68 to
someone for some bug a couple of years ago. But prior to that, no
bugs had been found in many years.

A lot of people who find errors in the books and get $2.00 checks from
Knuth save the checks as souvenirs instead of cashing them. A friend
of mine had one hanging on his wall for quite a long time. But Knuth
says there's a guy in San Jose, CA who has received and cashed
something like hundreds of them.
unknown
2003-10-17 20:36:16 UTC
Permalink
I don't think it doubles at every error, but at every version.
So if he pays $2.56 for errors in version 3.141, he pays $5.12
for errors in version 3.1415 (for example). So he gets to control
how fast he is pauperized.
He stopped the increases at 327.68, a number of years ago. There have
been a few releases since then with no doubling.
David Mertz
2003-10-17 08:58:09 UTC
Permalink
Jacek Generowicz <jacek.generowicz at cern.ch> wrote previously:
|> Could you afford to pay $327.68 for every bug found in your glorious
|> software?
|Anyone know how Knuth arrives at the amounts he offers ?
|(He also offers $2.56 for errors in his books.)

Surely you recognize powers of two? :-).

The system is really the ultimate in chutzpah. In takes VERY few errors
before he goes broke... or, say, if the US treasury were at his
disposal, *IT* goes broke.

Apparently, so far 8 errors have been found in AoCP, and 15 in TeX....
there are more than that (known) in my book that's been out for a couple
months (and I share a publisher with him, whatever that's worth). And I
*thought* I worked hard to avoid all the typos, plus put it up for
community review for almost a year, plus had a copyeditor and a
proofreader. It's not wonder that Knuth wrote the bible... he's god.

Yours, David...

--
mertz@ _/_/_/_/ THIS MESSAGE WAS BROUGHT TO YOU BY: \_\_\_\_ n o
gnosis _/_/ Postmodern Enterprises \_\_
.cx _/_/ \_\_ d o
_/_/_/ IN A WORLD W/O WALLS, THERE WOULD BE NO GATES \_\_\_ z e
John Wilson
2003-10-17 10:11:55 UTC
Permalink
Could you afford to pay $327.68 for every bug found in your glorious
software?
Anyone know how Knuth arrives at the amounts he offers ?
(He also offers $2.56 for errors in his books.)
Surely you recognize powers of two? :-).
The system is really the ultimate in chutzpah. In takes VERY few
errors before he goes broke... or, say, if the US treasury were at his
disposal, *IT* goes broke.
I don't think it costs Knuth anything. I was at a talk he gave some years
ago when he said that nobody ever cashed the checks. Everbody has them
framed!

John Wilson
The Wilson Partnership
http://www.wilson.co.uk
David Mertz
2003-10-17 09:05:45 UTC
Permalink
Jacek Generowicz <jacek.generowicz at cern.ch> wrote previously:
|> Could you afford to pay $327.68 for every bug found in your glorious
|> software?
|Anyone know how Knuth arrives at the amounts he offers ?
|(He also offers $2.56 for errors in his books.)

Surely you recognize powers of two? :-).

The system is really the ultimate in chutzpah. In takes VERY few errors
before he goes broke... or, say, if the US treasury were at his
disposal, *IT* goes broke.

Apparently, so far 8 errors have been found in AoCP, and 15 in TeX....
there are more than that (known) in my book that's been out for a couple
months (and I share a publisher with him, whatever that's worth). And I
*thought* I worked hard to avoid all the typos, plus put it up for
community review for almost a year, plus had a copyeditor and a
proofreader. It's not wonder that Knuth wrote the bible... he's god.

Yours, David...

--
mertz@ _/_/_/_/ THIS MESSAGE WAS BROUGHT TO YOU BY: \_\_\_\_ n o
gnosis _/_/ Postmodern Enterprises \_\_
.cx _/_/ \_\_ d o
_/_/_/ IN A WORLD W/O WALLS, THERE WOULD BE NO GATES \_\_\_ z e
Daniel Klein
2003-10-18 17:40:58 UTC
Permalink
On Fri, 17 Oct 2003 11:11:55 +0100, "John Wilson" <tug at wilson.co.uk>
Post by John Wilson
Could you afford to pay $327.68 for every bug found in your glorious
software?
Anyone know how Knuth arrives at the amounts he offers ?
(He also offers $2.56 for errors in his books.)
Surely you recognize powers of two? :-).
The system is really the ultimate in chutzpah. In takes VERY few
errors before he goes broke... or, say, if the US treasury were at his
disposal, *IT* goes broke.
I don't think it costs Knuth anything. I was at a talk he gave some years
ago when he said that nobody ever cashed the checks. Everbody has them
framed!
It's got _my_ vote for QOTW ;-)

Damien Wyart
2003-10-17 09:11:26 UTC
Permalink
Could you afford to pay $327.68 for every bug found in your glorious
software?
Anyone know how Knuth arrives at the amounts he offers ?
(He also offers $2.56 for errors in his books.)
Python 2.3.2 (#2, Oct 6 2003, 08:02:06)
[GCC 3.3.2 20030908 (Debian prerelease)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
327.68/2.56
128.0
--
DW
Nick Efford
2003-10-17 11:58:44 UTC
Permalink
Funny, 32768 and 256 are numbers which ring very loud bells with me
... but inserting a "." in the middle completely suppressed this
otherwise immediate reaction.
OK, so they are powers of 2, in cents ... anyone know any further
motivation ?
A fascination with numbers / mathematics?

The version number of TeX is converging to pi, rather than
incrementing in the usual fashion. Currently it is at 3.14159,
I believe...


Nick
Jacek Generowicz
2003-10-17 09:26:23 UTC
Permalink
Post by Damien Wyart
Could you afford to pay $327.68 for every bug found in your glorious
software?
Anyone know how Knuth arrives at the amounts he offers ?
(He also offers $2.56 for errors in his books.)
Python 2.3.2 (#2, Oct 6 2003, 08:02:06)
[GCC 3.3.2 20030908 (Debian prerelease)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
327.68/2.56
128.0
Funny, 32768 and 256 are numbers which ring very loud bells with me
... but inserting a "." in the middle completely suppressed this
otherwise immediate reaction.

OK, so they are powers of 2, in cents ... anyone know any further
motivation ?
Jacek Generowicz
2003-10-17 08:15:21 UTC
Permalink
Could you afford to pay $327.68 for every bug found in your glorious
software?
Anyone know how Knuth arrives at the amounts he offers ?

(He also offers $2.56 for errors in his books.)
Neal Holtz
2003-10-17 14:42:18 UTC
Permalink
|> Could you afford to pay $327.68 for every bug found in your glorious
|> software?
|Anyone know how Knuth arrives at the amounts he offers ?
|(He also offers $2.56 for errors in his books.)
Surely you recognize powers of two? :-).
The system is really the ultimate in chutzpah. In takes VERY few errors
before he goes broke... or, say, if the US treasury were at his
disposal, *IT* goes broke.
I don't think it doubles at every error, but at every version.
So if he pays $2.56 for errors in version 3.141, he pays $5.12
for errors in version 3.1415 (for example). So he gets to control
how fast he is pauperized.
Christian Tanzer
2003-10-17 10:20:43 UTC
Permalink
|> Could you afford to pay $327.68 for every bug found in your glorious
|> software?
|Anyone know how Knuth arrives at the amounts he offers ?
|(He also offers $2.56 for errors in his books.)
Surely you recognize powers of two? :-).
The system is really the ultimate in chutzpah. In takes VERY few errors
before he goes broke... or, say, if the US treasury were at his
disposal, *IT* goes broke.
Apparently, so far 8 errors have been found in AoCP, and 15 in TeX....
There were more errors (but surprisingly few). Knuth capped the amount
after a few doublings.

http://en.wikipedia.org/wiki/TeX has some more information about this.

Knuth's paper ``The Errors of TeX`` is recommended reading.
--
Christian Tanzer http://www.c-tanzer.at/
Loading...