HP Forum Archive 16

[ Return to Index | Top of Index ]

Interesting paper on calculator limitations... Message #1 Posted by David L. Jones on 30 Nov 2006, 6:25 a.m. I found this interesting paper on calculator limitations while surfing around: http://www.uclic.ucl.ac.uk/harold/srf/allcalcs.pdf It mentions the HP20S many times. Dave :)

Re: Interesting paper on calculator limitations... Message #2 Posted by David L. Jones on 30 Nov 2006, 6:29 a.m., in response to message #1 by David L. Jones and another version by the same author: http://www.uclic.ucl.ac.uk/harold/srf/truecalc.pdf Dave :)

Re: Interesting paper on calculator limitations... Message #3 Posted by Antonio Maschio (Italy) on 30 Nov 2006, 11:10 a.m., in response to message #2 by David L. Jones I discovered Thimbleby works long ago, and I totally agree with him. Here's a complete list of his works about calculators: 1. From Henry Briggs to Modern Calculators 2. Calculators are Needlessly Bad 3. A True Calculator 4. A New Calculator and Why it is Necessary Hope this will be delighful. -- Antonio Edited: 30 Nov 2006, 11:12 a.m.

Re: Interesting paper on calculator limitations... Message #4 Posted by Mike Morrow on 30 Nov 2006, 1:50 p.m., in response to message #3 by Antonio Maschio (Italy) Those are interesting articles, made more so by their containing no discussion or consideration at all to RPN logic machines. That would seem to adversely affect the value of the conclusions drawn by the author. Most of the problem areas and his suggestions for improved design are not required RPN machines. Could it be that Prof. Thimbleby is completely unfamiliar with RPN? Mike / Alabama

Re: Interesting paper on calculator limitations... Message #5 Posted by John Limpert on 30 Nov 2006, 2:17 p.m., in response to message #4 by Mike Morrow He does mention them briefly, criticizing the fixed stack of the HP-32SII. While RPN has its advantages, it doesn't solve many of the problems that he addresses.

Re: Interesting paper on calculator limitations... Message #6 Posted by bill platt on 30 Nov 2006, 3:55 p.m., in response to message #4 by Mike Morrow See the top of page (3) of reference (1) above. He briefly mentions RPN and criticizes the 4-level stack.

Re: Interesting paper on calculator limitations... Message #7 Posted by John Limpert on 30 Nov 2006, 2:13 p.m., in response to message #3 by Antonio Maschio (Italy) Those papers help to explain why I find many Japanese (Casio, Sharp) calculators to be very difficult to use. Sometimes I wonder if there is anyone at home in the engineering departments of the major calculator manufacturers. I recently looked at a TI-30X IIS, a common scientific calculator. The secondary key functions were printed in dark blue on a black background. Why not just use invisible ink?

Re: Interesting paper on calculator limitations... Message #8 Posted by bill platt on 30 Nov 2006, 4:09 p.m., in response to message #7 by John Limpert If you've started your calculating life on an HP, the consternation and disgust factor is just too much to bear, switching to a Casio or a Sharp or a kinPo. Having to switch to a "STAT mode" is distasteful and arbitrarily limiting. All that inconsistent crap that Thimbleby so rightfully attacks is just so plain obviously wrong. It isn't the RPN per se that makes HP better, it is that the others are just so stupid! I have a 30S and except for the nice screen and the basic idea of a formula-based entry, I hate it. Sometimes I can suffer it happily, as long as I don't push it to do the "advanced" stuff that it is supposed to do. I hate the fraction method, I hate the modes business, I hate the way the variables and registers are divided into "this one can save a formula, but this one can't" etc. I hate the totally bizarre handling of repeated function calls on cheap "4 function" calculators. Not features, bugs in my opinion. The funny thing about the thimbleby paper is that you've probably reacted viscerally to every one of the flaws he points out. I certainly have, for a long time. But Thibmeby takes those visceral reactions and makes a formal academic presentation. Brilliant stuff (that nobody will read, because it is academic!).

Re: Interesting paper on calculator limitations... Message #9 Posted by Juergen (CH) on 30 Nov 2006, 5:25 p.m., in response to message #3 by Antonio Maschio (Italy) Harold Thimbleby has continued his work. Visit his homepage Harold Thimbleby and click on the link 'Here's some of our latest work, exhibited at the Royal Society Summer Science Exhibition, 2005'. You will find there e.g. a Java version of his calculator and movies showing the calculator in action. Enjoy, Juergen

Re: Interesting paper on calculator limitations... Message #10 Posted by Antonio Maschio (Italy) on 30 Nov 2006, 5:51 p.m., in response to message #9 by Juergen (CH) I know, I know, I also downloaded his (his son's, to tell the truth) software calculator for Mac OS X which is a wonder! -- Antonio P.S. I put his site among my favorites, and I check from time to time if there's something new. I also asked him if they ever will make a true (handheld) calculator. Here's their answer: Quote: I doubt there will be a handheld version calculator soon. Although we are talking with manufacturers about providing a software package for interactive whiteboards. Let's hope for the future!

Re: Interesting paper on calculator limitations... Message #11 Posted by Maximilian Hohmann on 1 Dec 2006, 3:18 a.m., in response to message #10 by Antonio Maschio (Italy) Hello! Quote: I know, I know, I also downloaded his (his son's, to tell the truth) software calculator for Mac OS X which is a wonder! I think I am a bit stupid, but I coudn't find this MacOS X version anywhere (I only found the Java version which won't run on my Mac (10.4.6)). Where did you get it from? Quote: Let's hope for the future! Have they contacted (Sir) Clive Sinclair about it yet - he is the only one who comes to mind who might be able to make this come true (or who might have been 30 years ago...). Greetings, Max

Re: Interesting paper on calculator limitations... Message #12 Posted by Pal G. on 2 Dec 2006, 9:07 a.m., in response to message #11 by Maximilian Hohmann I cannot not find this Thimbleby OSX calculator either. After all the nice comments, naturally I want it! I found this webpage and link for an OSX calculator, but the link does not work. http://www.uclic.ucl.ac.uk/harold/calculators/ http://www.heathmoor.plus.com/downloads/TrueCalculatorBetaX.app.sit Further searching for "TrueCalculatorBeta", "truecalculator", etc., on Google does not lead to better results. So... how about a link? Thanks, Pal Edited: 4 Dec 2006, 12:31 a.m. after one or more responses were posted

Re: Interesting paper on calculator limitations... Message #13 Posted by Patrick Jamet on 2 Dec 2006, 11:28 a.m., in response to message #12 by Pal G. Hello The calculator is there (download). Regards

Re: Interesting paper on calculator limitations... Message #14 Posted by Namir on 2 Dec 2006, 12:02 p.m., in response to message #13 by Patrick Jamet Very interesting applet!!!!!!!!!!!! I love the freehand numeric reconition! :-) Namir

Re: Interesting paper on calculator limitations... Message #15 Posted by Pal G. on 4 Dec 2006, 12:32 a.m., in response to message #13 by Patrick Jamet Thank you for the link, Patrick. Cheers, Pal

Re: Interesting paper on calculator limitations... Message #16 Posted by Walter B on 7 Dec 2006, 4:45 p.m., in response to message #13 by Patrick Jamet Thanks, Patrick, for showing us this nice application! I played with it a bit and found it recognizes most handwritten digits and arithmetic symbols, and additionally "e" (the exponential), "pi" and square roots. Kind of undocumented features - at least there is no reference on the help screen. But I do not see a way to get trigonometric functions so far. Anyone has an idea?

Re: Interesting paper on calculator limitations... Message #17 Posted by Antonio Maschio (Italy) on 7 Dec 2006, 2:47 a.m., in response to message #11 by Maximilian Hohmann This is the link Will Thimbleby gave to me. Hope you will enjoy the program (for Mac OS X only!). -- Antonio Edited: 7 Dec 2006, 5:14 a.m.

Re: Interesting paper on calculator limitations... Message #18 Posted by Juergen (CH) on 1 Dec 2006, 3:46 a.m., in response to message #10 by Antonio Maschio (Italy) Porting the software to a PDA (Palm or Windows Mobile) might be a short/middle-term option. But I guess the source code is not available, and developing the calculator from scratch seems to be a big task...

CAS with GUI Message #19 Posted by Mike (Stgt) on 1 Dec 2006, 4:21 a.m., in response to message #9 by Juergen (CH) Thank you for the hint. Looks like the beginning of a CAS with GUI. Ciao.....Mike

My views about Thimbleby's "new calculator" Message #20 Posted by Karl Schneider on 1 Dec 2006, 3:20 a.m., in response to message #3 by Antonio Maschio (Italy) Hello, Antonio -- Quote: I discovered Thimbleby works long ago, and I totally agree with him. Here's a complete list of his works about calculators: More than a year ago, I also had read at least three of the four documents you linked, and even e-mailed Prof. Thimbleby asking if he'd like some comments. Unfortunately, I haven't yet followed through. I agree with most of the criticisms he levied at modern mass-market calculators, which are often counter-intuitive. However, I don't endorse his concept of a new calculator user interface. Fundamentally, the idea presented in the articles is a calculator that unravels and solves problems by combining algebra along with the function of calculation. The user specifies the variable to be solved by simply omitting it from the expression, then the calculator does the rest. The syntax is rather informal and unstructured -- prone to ambiguity and misinterpretation, I believe. I disagree with that approach, primarily on those grounds. I would prefer instead the structure and user-friendliness of the HP-17B/HP-27S equation editor/solver: The user enters an expression with named variables if desired, then the calculator interprets and validates the expression, and subsequently provides an easy means of entering values for unknown variables and solving for the remaining one. That solver can even perform algebra to obtain a "direct" solution if the form of the expression suggests only one primary solution. That capability was incorporated into the HP-33S' solver with a less-capable equation editor, and the solvers on all three models have the unwelcome characteristic that they must be "tricked" in order to find solutions other than the primary one. That suggests my other objection to Thimbleby's concept: An expression might have multiple solutions for the unknown variable that lack a regular pattern, as is found with the roots of a polynomial. Which solution would his calculator find? As mentioned by others, Thimbleby's objection to the RPN-based HP-32SII was its 4-level stack. Of course, that is not a characteristic of RPN per se, but a constraint that could easily be changed. I have not reviewed his latest work. - KS Edited: 2 Dec 2006, 4:15 p.m.

[ Return to Index | Top of Index ]

Go back to the main exhibit hall