Excel 2003: RAND and RANDBETWEEN functions

R

Rita Nikas [MSFT]

Hi all.

I just wanted to give you some information on this concern direct from
Microsoft. In addition, this problem was mentioned in Woody's Watch, and we
wanted to provide some speaking points to that piece.

Microsoft has recently learned of a flaw that affects random number returns
in Microsoft Office Excel 2003's RAND and RANDBETWEEN function. In some
cases, this may result in inaccurate data. Microsoft is currently in the
process of developing and testing a fix for the Excel flaw, which will be
made available to customers worldwide in the near future.

How did such a glaring bug survive the in-house testing that was presumably
done?

Microsoft makes every effort to identify and fix software bugs prior to
product shipment, but these reports are the reality of software
development. Through new technologies such as Watson and the new customer
connection tools, for example, Microsoft is able to better respond to
feedback and fix bugs in a timely manner.

Why were multiple notifications of this problem to Microsoft ignored? The
bug is easily replicated so there would seem to be no good reason for the
issue not to have been escalated promptly. We are still investigating what
reports were actually received by Microsoft. At present it appears that
certainly we weren't aware of this
prior to RTM, which would have enabled us to fix the bug much earlier.

What steps will Microsoft take to fix this bug and advise customers of the
solution? Microsoft is currently in the process of developing and testing a
fix for the Excel flaw, which will be made available to customers worldwide
in the near future.

What steps will Microsoft take to ensure that any future Excel problems are
properly addressed? Microsoft is continually looking for ways to improve
processes and make customers' software experience better. A key part of the
software development process is being able to involve customers in product
testing. Through new technologies such as Watson and the new customer
connection tools, for example, Microsoft is better able to respond to
customer feedback and fix bugs in a timely manner.

Sincerely,

Rita Nikas
Microsoft MVP Lead
Microsoft Corporation

This posting is provided "AS IS" with no warranties, and confers no rights.
 
N

Norman Harker

Hi Rita!

Thanks for your efforts on this one. It's sure appreciated.

--
Regards
Norman Harker MVP (Excel)
Sydney, Australia
(e-mail address removed)
Excel and Word Function Lists (Classifications, Syntax and Arguments)
available free to good homes.
 
H

Harlan Grove

"Rita Nikas [MSFT]" said:
How did such a glaring bug survive the in-house testing that was
presumably done?

Microsoft makes every effort to identify and fix software bugs prior to
product shipment, but these reports are the reality of software
development. Through new technologies such as Watson and the new customer
connection tools, for example, Microsoft is able to better respond to
feedback and fix bugs in a timely manner.
....

Granted bugs happen, but this shouldn't have survived standard regression
testing, if your QA people do regression testing. Unlike mny bugs, the RAND
bug is glaring - it returns negative values after many recalculations. The
only way to have missed it would be not to have run many recalculations.
What steps will Microsoft take to ensure that any future Excel problems
are properly addressed? Microsoft is continually looking for ways to
improve processes and make customers' software experience better. A key
part of the software development process is being able to involve
customers in product testing.
....

http://groups.google.com/[email protected]

http://groups.google.com/[email protected]

Microsoft will let users test, but it won't tell them what they're testing.

OK, it was a mistake, and Microsoft wants to rectify it. Wonderful. So
what's Microsoft's position on the inconsistencies in the SLOPE, INTERCEPT
and FORECAST worksheet functions that have been much discussed over the last
week? And what, exactly, is the bug reporting process that users should
follow? And what are Microsoft's procedures for acknowledging bugs? Is there
a Microsoft web site that lists all recently reported and as yet unfixed
bugs as there is with almost all large open source projects?
 
E

Earl Takasaki

Hi!

I am the one who reported this bug to Woody's Watch. I can tell you that
I only did so after reporting the bug to Microsoft at least 3 times and
receiving absolutely no acknowledgement whatsoever. Zero. Nada. Zilch.

I am not a Microsoft basher, but rather a long-time user and Excel "fan"
as well as a VBA consultant. Please take my criticisms as constructive:

1) It is not easy to figure out how to report a "bug" to Microsoft. I
have always been amazed at how defensive some software companies are
about bugs. As a developer (albeit on a lilliputian scale next to you),
I depend on my customers to find bugs and I make it super easy for them
to report such so that I can take care of the problems ASAP.

2) Unless one has some "pull" (i.e., one is a journalist or VIP or
insider) one is ignored completely by MSFT. This is counterproductive.
A simple acknowledgement would be nice. When the problem is a severe
bug, some follow through should be forthcoming. Stonewalling only causes
frustration and MORE not less negative publicity, and futhers the
conception that you guys don't care. Quite frankly, if Woody's watch and
others Excel VIPS that I personally emailed did not make a fuss, it is
evident that you would have probably ignored this particular problem
until a BIG problem occurred or someone of sufficient stature (other
than little ol' me) reported it independently.

3) Those people who will be affected by the bug are probably those who
are running sophisticated models. While it is nice that you have
acknowledged the bug here in a forum, it is inexcusable that you have
not publically warned your power users that their calculations may be at
risk, nor have you offered a wordaround. This is especially insensitive
because you have publically touted the advantages of the "new and
improved" RAND() function in your public literature. I hope a bridge
doesn't collapse....

Nuff said. I hope that this problem will be fixed soon.

Best,

Earl Takasaki
EXCELent Consulting
 
H

Harlan Grove

Earl Takasaki said:
1) It is not easy to figure out how to report a "bug" to Microsoft. I
have always been amazed at how defensive some software companies are
about bugs. . . .

Standard procedure seems to be to call MSFT customer support and PAY for the
priviledge of reporting bugs. That fits in well with Microsoft's corporate
vision: make the customer pay, and pay, and pay, and pay, . . .
3) Those people who will be affected by the bug are probably those who
are running sophisticated models. While it is nice that you have
acknowledged the bug here in a forum, it is inexcusable that you have
not publically warned your power users that their calculations may be at
risk, nor have you offered a wordaround. . . .

Disclosure shouldn't be limited to power users. Compare MSFT's modus
operendi with respect to bugs with that of most if not all open source
software projects. MSFT discloses next to nothing. See if there's any
mention of this problem with RAND in XL2003 in the KnowledgeBase yet (there
isn't as of 14:40 PST 27 Dec 2003), even though Rita acknowledged this as a
problem over a week ago. Rita's posting is likely all MSFT plans on
releasing.

Look at the web sites of any open source project. Current bugs are listed
fairly prominently. For example, OpenOffice,

http://www.openoffice.org

in which the link to Bugs & Issues is easily found in the upper-left corner
of the page. Then look at Office's main page,

http://office.microsoft.com/home/default.aspx

and see if you find a link to anything resembling a bug or problem listing.
It doesn't exist because MSFT won't admit errors of any kind until they're
blatantly obvious and thoroughly reported.
 
R

Rita Nikas [MSFT]

Hi Harlan.

I apologize for not responding sooner as I was on vacation for the
holidays. I thought I'd posted another message related to
SLOPE/INTERCEPT/FORECAST to the public NGs - I apologize for not doing so
earlier.

BTW, Jerry W. Lewis has also alerted me directly that STYEX is also
impacted.

Microsoft has recently learned of a flaw that affects results returned in
Microsoft Office Excel 2003's SLOPE, INTERCEPT, FORECAST, and STYEX
functions.
In some cases, this may result in inaccurate data. Microsoft is currently
in the process of developing and testing a fix for the Excel flaw, which
will be
made available to customers worldwide in the near future.

Unofficially, we are currently in the testing cycle of a possible fix. A KB
article should be forthcoming but we are not sure exactly when. Microsoft
recommends not using the SLOPE, INTERCEPT, or FORECASTfunctions in
Microsoft Office Excel 2003 until a patch is available.

Related to your questions:
What is the bug reporting process that users should follow? And what are
Microsoft's procedures for acknowledging bugs? Is there a Microsoft web
site that lists all recently reported and as yet unfixed bugs as there is
with almost all large open source projects?

The bug reporting process users should follow? Probably the fastest is to
contact Product Support Services by telephone for assistance. As far as
acknowledging bugs, this post does just that. And once some type of KB is
available, that will also serve to document the problem completely. We
don't have a web site (other than KB articles within the Microsoft KB) that
lists all recently reported and unfixed bugs. It is a good idea that I do
think we as a company should consider, so I will be sure to mention it to
my management.

Granted, it's a pain to have to wait on this, and it would have been nice
had we not had these problems to begin with. However, we are aware of them
and working as rapidly as we can to provide a solid fix to resolve this
situation - we just don't want to be too hasty so I appreciate your
patience until such time as a resolution is announced.

Sincerely,

Rita Nikas
Microsoft MVP Lead
Microsoft Corporation

This posting is provided "AS IS" with no warranties, and confers no rights.


--------------------
| Subject: Re: Excel 2003: RAND and RANDBETWEEN functions
| Date: Sat, 20 Dec 2003 01:35:06 GMT
|
| ...
| >How did such a glaring bug survive the in-house testing that was
| >presumably done?
| >
| >Microsoft makes every effort to identify and fix software bugs prior to
| >product shipment, but these reports are the reality of software
| >development. Through new technologies such as Watson and the new customer
| >connection tools, for example, Microsoft is able to better respond to
| >feedback and fix bugs in a timely manner.
| ...
|
| Granted bugs happen, but this shouldn't have survived standard regression
| testing, if your QA people do regression testing. Unlike mny bugs, the
RAND
| bug is glaring - it returns negative values after many recalculations. The
| only way to have missed it would be not to have run many recalculations.
|
| >What steps will Microsoft take to ensure that any future Excel problems
| >are properly addressed? Microsoft is continually looking for ways to
| >improve processes and make customers' software experience better. A key
| >part of the software development process is being able to involve
| >customers in product testing.
| ...
|
|
http://groups.google.com/[email protected]
|
|
http://groups.google.com/[email protected]
bl
|
| Microsoft will let users test, but it won't tell them what they're
testing.
|
| OK, it was a mistake, and Microsoft wants to rectify it. Wonderful. So
| what's Microsoft's position on the inconsistencies in the SLOPE, INTERCEPT
| and FORECAST worksheet functions that have been much discussed over the
last
| week? And what, exactly, is the bug reporting process that users should
| follow? And what are Microsoft's procedures for acknowledging bugs? Is
there
| a Microsoft web site that lists all recently reported and as yet unfixed
| bugs as there is with almost all large open source projects?
|
|
|
 
R

Rita Nikas [MSFT]

Hi Earl.

As per my e-mail to you, I'd like to get complete information from you on
when and how you reported this problem to Microsoft. I don't find this
acceptable at all and will investigate so we can avoid problems like this
in the future

Sincerely,

Rita Nikas
Microsoft MVP Lead
Microsoft Corporation

This posting is provided "AS IS" with no warranties, and confers no rights.


--------------------
| Subject: Re: Excel 2003: RAND and RANDBETWEEN functions
| Date: Sat, 27 Dec 2003 09:08:04 GMT
|
| Hi!
|
| I am the one who reported this bug to Woody's Watch. I can tell you that
| I only did so after reporting the bug to Microsoft at least 3 times and
| receiving absolutely no acknowledgement whatsoever. Zero. Nada. Zilch.
|
| I am not a Microsoft basher, but rather a long-time user and Excel "fan"
| as well as a VBA consultant. Please take my criticisms as constructive:
|
| 1) It is not easy to figure out how to report a "bug" to Microsoft. I
| have always been amazed at how defensive some software companies are
| about bugs. As a developer (albeit on a lilliputian scale next to you),
| I depend on my customers to find bugs and I make it super easy for them
| to report such so that I can take care of the problems ASAP.
|
| 2) Unless one has some "pull" (i.e., one is a journalist or VIP or
| insider) one is ignored completely by MSFT. This is counterproductive.
| A simple acknowledgement would be nice. When the problem is a severe
| bug, some follow through should be forthcoming. Stonewalling only causes
| frustration and MORE not less negative publicity, and futhers the
| conception that you guys don't care. Quite frankly, if Woody's watch and
| others Excel VIPS that I personally emailed did not make a fuss, it is
| evident that you would have probably ignored this particular problem
| until a BIG problem occurred or someone of sufficient stature (other
| than little ol' me) reported it independently.
|
| 3) Those people who will be affected by the bug are probably those who
| are running sophisticated models. While it is nice that you have
| acknowledged the bug here in a forum, it is inexcusable that you have
| not publically warned your power users that their calculations may be at
| risk, nor have you offered a wordaround. This is especially insensitive
| because you have publically touted the advantages of the "new and
| improved" RAND() function in your public literature. I hope a bridge
| doesn't collapse....
|
| Nuff said. I hope that this problem will be fixed soon.
|
| Best,
|
| Earl Takasaki
| EXCELent Consulting
|
|
|
|
| Rita Nikas [MSFT] wrote:
| > Hi all.
| >
| > I just wanted to give you some information on this concern direct from
| > Microsoft. In addition, this problem was mentioned in Woody's Watch,
and we
| > wanted to provide some speaking points to that piece.
| >
| > Microsoft has recently learned of a flaw that affects random number
returns
| > in Microsoft Office Excel 2003's RAND and RANDBETWEEN function. In
some
| > cases, this may result in inaccurate data. Microsoft is currently in
the
| > process of developing and testing a fix for the Excel flaw, which will
be
| > made available to customers worldwide in the near future.
| >
| > How did such a glaring bug survive the in-house testing that was
presumably
| > done?
| >
| > Microsoft makes every effort to identify and fix software bugs prior to
| > product shipment, but these reports are the reality of software
| > development. Through new technologies such as Watson and the new
customer
| > connection tools, for example, Microsoft is able to better respond to
| > feedback and fix bugs in a timely manner.
| >
| > Why were multiple notifications of this problem to Microsoft ignored?
The
| > bug is easily replicated so there would seem to be no good reason for
the
| > issue not to have been escalated promptly. We are still investigating
what
| > reports were actually received by Microsoft. At present it appears that
| > certainly we weren't aware of this
| > prior to RTM, which would have enabled us to fix the bug much earlier.
| >
| > What steps will Microsoft take to fix this bug and advise customers of
the
| > solution? Microsoft is currently in the process of developing and
testing a
| > fix for the Excel flaw, which will be made available to customers
worldwide
| > in the near future.
| >
| > What steps will Microsoft take to ensure that any future Excel problems
are
| > properly addressed? Microsoft is continually looking for ways to
improve
| > processes and make customers' software experience better. A key part of
the
| > software development process is being able to involve customers in
product
| > testing. Through new technologies such as Watson and the new customer
| > connection tools, for example, Microsoft is better able to respond to
| > customer feedback and fix bugs in a timely manner.
| >
| > Sincerely,
| >
| > Rita Nikas
| > Microsoft MVP Lead
| > Microsoft Corporation
| >
| > This posting is provided "AS IS" with no warranties, and confers no
rights.
| >
| >
|
 
R

Rita Nikas [MSFT]

Hi Harlan and Earl.

It is Microsoft policy to initially charge for support incidents. However,
if an issue is found to be a bug of some kind, the customer is entitled to
a full refund for the support incident. Other Microsoft support options are
outlined here:

http://support.microsoft.com/default.aspx?scid=sz;[ln];top

Online support is initially much less expensive as compared to telephone
support, so that might be more viable to more people, especially in a NG
environment where people are concerned that what they're seeing is a new
issue.

If using one of the above support options doesn't work or isn't available,
then another option might be to post a message, asking the MVPs to take a
look at a concern (if you suspect it's a bug). I'm in constant contact with
the MVPs, and they frequently let me know if there are NG issues needing
Microsoft investigation.

In all the communications I've had (since this issue has cropped up), I've
been in communication with development and others. We are definitely
planning on releasing an update and a Microsoft Knowledge Base article when
the adequately tested fix is available.

Sincerely,

Rita Nikas
Microsoft MVP Lead
Microsoft Corporation

This posting is provided "AS IS" with no warranties, and confers no rights.


--------------------
| Subject: Re: Excel 2003: RAND and RANDBETWEEN functions
| Date: Sat, 27 Dec 2003 22:53:44 GMT
|
| ...
| >1) It is not easy to figure out how to report a "bug" to Microsoft. I
| >have always been amazed at how defensive some software companies are
| >about bugs. . . .
|
| Standard procedure seems to be to call MSFT customer support and PAY for
the
| priviledge of reporting bugs. That fits in well with Microsoft's corporate
| vision: make the customer pay, and pay, and pay, and pay, . . .
|
| >3) Those people who will be affected by the bug are probably those who
| >are running sophisticated models. While it is nice that you have
| >acknowledged the bug here in a forum, it is inexcusable that you have
| >not publically warned your power users that their calculations may be at
| >risk, nor have you offered a wordaround. . . .
|
| Disclosure shouldn't be limited to power users. Compare MSFT's modus
| operendi with respect to bugs with that of most if not all open source
| software projects. MSFT discloses next to nothing. See if there's any
| mention of this problem with RAND in XL2003 in the KnowledgeBase yet
(there
| isn't as of 14:40 PST 27 Dec 2003), even though Rita acknowledged this as
a
| problem over a week ago. Rita's posting is likely all MSFT plans on
| releasing.
|
| Look at the web sites of any open source project. Current bugs are listed
| fairly prominently. For example, OpenOffice,
|
| http://www.openoffice.org
|
| in which the link to Bugs & Issues is easily found in the upper-left
corner
| of the page. Then look at Office's main page,
|
| http://office.microsoft.com/home/default.aspx
|
| and see if you find a link to anything resembling a bug or problem
listing.
| It doesn't exist because MSFT won't admit errors of any kind until they're
| blatantly obvious and thoroughly reported.
|
|
|
 
H

Harlan Grove

It is Microsoft policy to initially charge for support incidents. However,
if an issue is found to be a bug of some kind, the customer is entitled to
a full refund for the support incident. Other Microsoft support options are
outlined here:

http://support.microsoft.com/default.aspx?scid=sz;[ln];top
...

Bug reporting is support?!

To be fair, the first support call may be free, but what happens on the second
bug found? Does Microsoft just get credit card authorizations when users make
the first call to report a bug, or do you post the charge initially and credit
it later, thus taking advantage of the float either at the user's or their
credit card issuer's expense? [Rhetorical - I'd be very surprised indeed if you
were at liberty to make that public.]

That there is no direct way to report bugs in Microsoft software to Microsoft
free of *initial* charges other than posting to newsgroups and hoping & praying
that MVPs pass along the news to you or other individuals (who seem to take
holidays occasionally, so aren't always available) speaks for itself just how
much Microsoft wants to receive bug reports. No news is good news.
 
N

Norman Harker

Hi Rita!

I'm sure that your actions on this are fully appreciated by all.

As far as the history goes, I think I'm right in saying that
**patent** action on this was only really forthcoming after the
problem was raised directly with you. You "kicked ass" to get things
moving along and also gave almost immediate public feedback, which you
have now updated.

That's not saying that nothing was being done before. But feedback and
updates soothe the savage beasts who suffer from these things.

--
Regards
Norman Harker MVP (Excel)
Sydney, Australia
(e-mail address removed)
Excel and Word Function Lists (Classifications, Syntax and Arguments)
available free to good homes.
 
D

David Byrne

Hi All,

Back after a few years away.

Experimented with this new feature the other night and found that the
proportion of negative numbers is almost invariably 16-17 % OR 83-84
%.

When plotted, give a beautiful (half) Normal Distribution graph.

When 16-17%, approx 10% of all results 0 to +0.1.

When 83-84%, approx 10% of all results 0 to -0.1.

Some unusual transition effects when reverts to all positive.

I can confirm that ABS(RAND()) is in no way Random.

In the 16-17% cases described, 0 to .1 become around 14.5% of the
output, while 0.9 to 1 is around 5.5%.

More food for thought
 
H

Harlan Grove

...
...
Experimented with this new feature the other night and found that the
proportion of negative numbers is almost invariably 16-17 % OR 83-84
%.

When plotted, give a beautiful (half) Normal Distribution graph.

When 16-17%, approx 10% of all results 0 to +0.1.

When 83-84%, approx 10% of all results 0 to -0.1.

Some unusual transition effects when reverts to all positive.

And never goes negative again?
I can confirm that ABS(RAND()) is in no way Random.

Well, no pseudorandom number generator is random, and LCGs show deterministic
patterns. Do you mean ABS(RAND()) is clearly nonuniform? If so, do you mean
there's nontrivial skewness, kurtosis or both?
In the 16-17% cases described, 0 to .1 become around 14.5% of the
output, while 0.9 to 1 is around 5.5%.

OK, skewed. Gotta wonder whether the positive RAND() results are still uniform
but the negative results are skewed.
 
R

Rita Nikas [MSFT]

Hi all.

I just wanted to let you know that we do have Microsoft Knowledge Base (KB)
articles related to the RAND problem and a **released fix** for the RAND
issue (it actually addresses a couple of other concerns, too) is now
available.

KB 834520 specifically talks about RAND returning negative numbers and
refers to the fix available as part of 833618. If you'd like to review
these articles for more information, please see:

http://support.microsoft.com/?kbid=834520 (for the RAND problem
specifically)
http://support.microsoft.com/?kbid=833618 (for the fix)

Please also know that you can contact Product Support Services through
several different support options:

http://support.microsoft.com/default.aspx?scid=sz;[ln];top

My understanding is that, as long as you are experiencing at least *one* of
the problems outlined in 833618, you can call Product Support Services and
receive the fix free of charge. Please be aware that, if the fix doesn't
resolve your concern or you need further assistance, that can be a charged
incident.

Sincerely,

Rita Nikas
Microsoft MVP Lead
Microsoft Corporation

This posting is provided "AS IS" with no warranties, and confers no rights.


--------------------
| Date: Sat, 20 Dec 2003 00:19:01 GMT
| Subject: Excel 2003: RAND and RANDBETWEEN functions
|
| Hi all.
|
| I just wanted to give you some information on this concern direct from
| Microsoft. In addition, this problem was mentioned in Woody's Watch, and
we
| wanted to provide some speaking points to that piece.
|
| Microsoft has recently learned of a flaw that affects random number
returns
| in Microsoft Office Excel 2003's RAND and RANDBETWEEN function. In some
| cases, this may result in inaccurate data. Microsoft is currently in the
| process of developing and testing a fix for the Excel flaw, which will be
| made available to customers worldwide in the near future.
|
| How did such a glaring bug survive the in-house testing that was
presumably
| done?
|
| Microsoft makes every effort to identify and fix software bugs prior to
| product shipment, but these reports are the reality of software
| development. Through new technologies such as Watson and the new customer
| connection tools, for example, Microsoft is able to better respond to
| feedback and fix bugs in a timely manner.
|
| Why were multiple notifications of this problem to Microsoft ignored?
The
| bug is easily replicated so there would seem to be no good reason for the
| issue not to have been escalated promptly. We are still investigating
what
| reports were actually received by Microsoft. At present it appears that
| certainly we weren't aware of this
| prior to RTM, which would have enabled us to fix the bug much earlier.
|
| What steps will Microsoft take to fix this bug and advise customers of
the
| solution? Microsoft is currently in the process of developing and testing
a
| fix for the Excel flaw, which will be made available to customers
worldwide
| in the near future.
|
| What steps will Microsoft take to ensure that any future Excel problems
are
| properly addressed? Microsoft is continually looking for ways to improve
| processes and make customers' software experience better. A key part of
the
| software development process is being able to involve customers in
product
| testing. Through new technologies such as Watson and the new customer
| connection tools, for example, Microsoft is better able to respond to
| customer feedback and fix bugs in a timely manner.
|
| Sincerely,
|
| Rita Nikas
| Microsoft MVP Lead
| Microsoft Corporation
|
| This posting is provided "AS IS" with no warranties, and confers no
rights.
|
|
|
 
D

David Byrne

Hi All,

I know the Hot Fix is out, but this is a Provable Workaround that may
prove useful, and give some insight into the nature of the bug.

The following is based on a VERY large battery of tests using up to
130,000 cells at a time.

1. The percentage of negative numbers generated by Rand() varies
from less than 1 to over 90.

2. This is a bimodal distribution, with peak frequencies of around
16% and 84%.

3. Across around a hundred samples of 100,000 cells, the effect of
adding 1 where the number generated is negative, produces a
distribution indistinguishable from ones where all values are
positive.

4. In the "16%" and "84%" examples, "creative" plotting of the
frequencies at 0.1 intervals from -1 to + 1 produces a graph not
unlike a Normal Distribution. From these graphs, and the similar ones
for other percentages, the result (3) above is almost intuitive.

5. RANDBETWEEN(a,b) produces distributions of a basically similar
form. That is, for a given percentage of results outside the expected
range, there is a predictable set of "out of range" values, and
predictable frequencies for all values within and outside the range.

6. Once again, a (simple) addition to any values outside the
expected range, produces a distribution of values indistinguishable
from one where all values produced are within the expected range.

(email me direct for copies of RAND() results)

Some personal observations

Having formerly been a frequent contributor to these forums, I was
somewhat saddened to see the change in tone that seems to have crept
in more recently.

One regular correspondent appears renowned for public fault-picking.

Just this once I wish to do the same in an attempt to redress the
issue, and will not debate the matter further publicly.

I find this sort of behaviour irritating, distracting, inappropriate,
unprofessional, and demeaning to all concerned.

Those who give freely of their own time deserve better than this.

Perhaps more importantly, on the RAND() issue, we have recently seen a
sweeping "proof" that if the use of ABS(RAND()) produces a skewed
distribution, so "MUST" the use of x+1. This was presumably based on
some ill-founded assumption.

I find it incomprehensible that someone of that author's background,
experience, qualifications and personal beliefs could make such a
statement in a global forum without checking the facts.

For this "foul" alone I would like to wish that contributor well on
his (hopefully) self imposed "time out", and trust that if/when he
returns it will be with a new spirit of goodwill and co-operation. I
feel sure that in the next month or two, problems will continue to be
answered in a timely manner.



"Discovery is seeing what everyone else has seen
And thinking what no-one else has thought"



David Byrne
Melbourne Australia
 
N

Norman Harker

Hi All!

I've installed the hotfix and tested for generation of negative random
numbers.

Multiple tests (about 50) of which 10 were for more than 1 hour have
produced no cases where negatives were generated.

Now to test if the distributions are flat!

All we need now is a Service Release or Patch but I'm sure we should
all thank Earl and others who discovered the problem and provoked the
solution.

--
Regards
Norman Harker MVP (Excel)
Sydney, Australia
(e-mail address removed)
Excel and Word Function Lists (Classifications, Syntax and Arguments)
available free to good homes.
 
N

Norman Harker

Hi Don!

I'll chase this up and come back to you.

From your earlier post. Yes! there certainly is a bug in the RAND
function in Excel 2003. This has been fixed by the hotfix. You should
still be able to obtain it if you call Product Support Services for
your area.

--
Regards
Norman Harker MVP (Excel)
Sydney, Australia
(e-mail address removed)
Excel and Word Function Lists (Classifications, Syntax and Arguments)
available free to good homes.
 
G

Guest

----- Norman Harker wrote: ----

Hi Don

I'll chase this up and come back to you

From your earlier post. Yes! there certainly is a bug in the RAN
function in Excel 2003.

Yes, I know. I discovered it, and announced it to the workd right here in September Last Year..

http://www.microsoft.com/office/com...tions&mid=4366cb3a-5651-4ca7-9790-73557bfade9

This has been fixed by the hotfix.

I heard, but I still have to see this Hotfix for myself ;-

You should still be able to obtain it if you call Product Support Services for your area

Think I'll decline your kind invitation to partake of Micros??t Music on Hold for two hours at my expense, to be asked for my Credit Cazrd Number, for them to tell me what I alredy know

ABS(RAND()) rules


--
Regard
Norman Harker MVP (Excel
Sydney, Australi
(e-mail address removed)
Excel and Word Function Lists (Classifications, Syntax and Arguments
available free to good homes
 
P

PatFinegan

Before paying $35 to speak to Microsoft Tech Support (albeit with th
chance of a refund), does anyone know if the RAND() "hotfix" include
the VBA Rnd() function? Most of the derivative pricing models I've see
on Wall Street use the VBA Rnd() function rather than the spreadshee
=RAND() function
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top