A while back, I switched from using a calculator (the calculator program on Windows, as it happens) to a spreadsheet (Excel). I've been very happy with the results; so has a friend of mine to whom I recommended this. It's probably obvious to most of you, but we didn't do it for many years after we got computers with Excel, so I figure it isn't obvious to some.
The main advantage is accuracy, and security about accuracy: When you're adding a list of numbers using a normal calculator, you can't be sure after the fact that you typed everything right. With a spreadsheet, you can see all the source data right there on your screen, and can check that you entered the right formula just by clicking on the total cell. And if you realize you made a mistake, you can correct the one wrong cell without reentering all the data.
Of course, you can also more easily save the data for the future, copy it into a document or an e-mail, and so on. But even if you know you don't need that, the ability to immediately verify that you've entered all the data right is enough to justify switching to a spreadsheet. And while some features of Excel and other spreadsheets may be quite complex, learning enough just to add a column of numbers is a matter of seconds.
So, again, this might be obvious to some of you (and perhaps useless to a few who don't have a spreadsheet program); but if my and my friend's experience is any guide, it's the sort of obvious thing that many nonetheless overlook for a long time.
www.openoffice.org
It's an open source derivative of MS Office, and can read Excel, Word, PDF, and Powerpoint files (although MS Office won't read files saved in OO format).
I, too, have been using spreadsheets to do calculations for a very long time.
Just never confuse it with a database. It ain't.
Accountants use physical calculators with paper tapes for just this reason. Any calculator program worthy of being called one has a paper tape mode that displays the history of what you've entered.
Judy's TenKey Calculator
Judy's TenKey is a flexible calculator with nearly every feature imaginable, including a scrolling tape that recalculates your totals when you edit an entry. Judy's TenKey saves you time and makes your calculations more reliable ... we guarantee it!
Says the "Dog"
Also, one can paste Excel data or graphs into a Word document (which you can then convert to a PDF, if need be).
This year, for the first time, my Law Review used Excel for compiling our index and for a few other functions. It has worked well, I believe.
My brother was telling me nearly 10 years ago that the smart kids in college used Excel for EVERYTHING, including word processing.
That is nothing. I used to use AutoCAD for word processing because it was easier to use its limited typesetting then to import and maintain vector graphics in Word.
I still have the HP48 I bought in high school for my calculator needs. The 4 line window showing the stack makes a good compromise and the RPN makes it proof against borrowing.
Technically, it's an open source derivative of StarOffice, which was reasonably popular in Germany fifteen years ago. It's merely a competitor to MS Office, like Lotus SmartSuite.
Of course, if you want something small, fast, and basic, you can't beat the original spreadsheet, VisiCalc. PC version available for free and legal download, and runs just fine under Windows XP. (Reference card linked on page. Type /sqy to quit.)
No trouble, it was only the financing for a house we're having built!!! The error was very large and very obvious, but I no longer trust Excel at all. My trusty RPN calculator, on the other hand, is wonderful because you can enter all the numbers at once, then double-check them as you add them up.
I do cuff calculations with 'bc' and 'ghci', though.
Also, the forum software countered my every attempt to use a fixed-width font for those program names; over-aggressive protection measures, I guess.
It is especially apropos in view of EV's later post about ideological diversity in the Federalist Society. Maybe the rest of you get it, but for the life of me I can't figure out how a lawyer's mind can embrace inclusion of blacks, women and gays, say, for diversity's sake while ignoring the fact that there are almost NO scientists or mathematicians on SCOTUS, in Congress or in the Executive, and there probably haven't been since Jefferson and Franklin. Who the hell cares about diversity in the Federalist Society in view of the abysmal lack of diversity and good sense in all branches of our government?
If EV has just found out about the beauties of the spreadsheet, imagine all the scientific mental farts, such as those elucidated in Hofstadter or Paulos, that continue daily to hobble his judgment! I don't know about the rest of you, but I don't like living in a world where my scientist ass is subjected to the 'justice' doled out by science ignoramuses. The situation cries out for us scientists to get together and begin to whack their sorry asses with some our weapons!
My sticking with calculators instead of spreadsheets is not because I'm inept when it comes to science or math, or because I've "just found about the beauties of the spreadsheet." (A spreadsheet is no more scientific or mathematical than a calculator.) A failure of common sense it is. A failure of mathematical and scientific ability it is not. And it most surely is not something that remotely bears on broader questions having to do with the legal profession and science.
I do love the HP RPN.
So much so that I'm a FORTH type person when it comes to software.
And yes, it would be good to have a physicists or an engineer in Congress. However, they are not as good at lying as lawyers, so it is no wonder they don't make the cut. Insufficient practice dissembling.
But still I'd appreciate it if you were to comment, one of these days, on the probable consequences of the legal profession's relative ignorance of science. Doesn't that lack of 'diversity' bother you? You can afford to be frank, since your superior science abilities have already disqualified you from SCOTUS!
(Before Excel there was Multicalc ims. I got it free with a Mac in 1984.)
physicists and engineers don't lie as often as lawyers (at least not as they perceive the truth), because they become so convinced of the "truth" of their position and reasoning that they believe what they say whether it is accurate, rational, or useful. Of course this generalization doesn't apply to all science/engineer types, but it applies to a fair number of them. In my experience engineers can be the most irrational of professionals in real life non-engineering matters.
I've encountered engineers who calculate the value of a business to 4 decimal places and truly believe their answer is the "truth". Never mind that the reality was the business in question couldn't be valued within -+ $50,000 accurately and had no "exact" answer as its value.
A quick test would be to find out if one is dealing with the anal retentive tending to irrational behavior engineers ask them this. If they want to hang a picture in the center of wall do they measure the wall with a tape measure to find the exact spot where to put the nail, and if they were to find the nail were off exact center by an 1/8th of an inch would they feel compelled to move the nail or they wouldn't be able to stop thinking about it, or do they just eye ball it, walk up to the wall and put the nail in and hang the picture step back and look the picture to decide if they need to make any adjustments.
Another one would be to tie the engineer up to a chair and then take a brand new gas station road map out and unfold it and then fold it back up "wrong". Then just lay it folder up "wrong" in front of them and see if they go into convulsions, twitching uncontrollably while trying to escape their bonds so they could refold the map "correctly".
OK, maybe not the last test, but the first one has definite accuracy in sizing up the psychology of a person on your side of things or on the other side of things.
Says the "Dog"
I pretty much agree with what the "Dog" said. I'd wish for more economists.
One problem I've never been able to solve is this...
Frequently, I'll create a spreadsheet with numerous calculations, none of which are rounded. But then I realize I need the caluclations rounded to zero, two, six, etc. decimal places. I've never found a way to round the results of all calculations to a certain digit. You can round all the numbers in a spreadsheet fairly easily, but this doesn't get the same result.
E.g.
1 / 3 x 5.38 = 1.793333
but I want 1 / 3 = 3.33 x 5.38 = 1.78
Then I have to go back and insert =round() functions in every damn calculation.
As far as obsession on excessive digits, I've noticed this with alot of people (engineering and science) who have that problem, mostly those types are in universities or labs where they dont do any real "hands on" work.
Was your ex-wife an engineer obsessed with precision in trivial matters? You seem to harbor a lot of hostiliy....
For the record, as an engineer and lawyer, I would measure for the picture, and wouldn't fold the map properly. I learned at least something in law school.
DeezRightWingNutz(7.9.2007 5:20pm) wrote:
Then I have to go back and insert =round() functions in every damn calculation.
I just checked the help for the Open Office spreadsheet program and found that it can easily be configured to use displayed precision. I checked Excel and could not quickly determine whether it has this feature or not. If this is a feature that would be helpful, you might consider the OO spreadsheet -- it's somewhat different from Excel, but is easily learned.
As far as rounding off in general, all floating point calculations (i.e. not integer) performed with spreadsheets and calculators will have limited precision. This can be a problem if your calculations contain both very large and very small numbers, as mentioned earlier about regression.
Full Version of Published Paper), which showed both the existence of a number of errors and the poor performance of Microsoft in fixing them. One of McCullough's papers compares the GNU spreadsheet favorably to Excel.
The free program R is terrific for real statistics and graphics but can easily be used as a calculator too.
Microsoft did not fix the problems in Excel 97 in the 2000 or 2002 version as pointed out by McCullough.
Excel is not even programmed properly for a variance calculation. It uses the one-pass formula, which requires the subtraction of two large numbers. This formula will sometimes breakdown and give you a negative answer because of the finite precision of digital computers. But more fundamentally it shows Microsoft’s amateurish approach to mathematical calculations. No doubt MS give the variance problem to some H1-B who grabbed a statistics book and copied the one-pass formula in ignorance of the consequences for numerical accuracy.
Excel is also unfit for even simple things like calendar calculations. My version of Excel 2000 thought 1900 was a leap year.
In summary don’t use Excel for anything important and check the answer. Perhaps MS has fixed the problems in the latest version, but don't bet on it.
Excel is also unfit for engineering and physics calculations as well. In fact most packages fail some of the NIST benchmark tests, including Matlab, which is popular with engineers. The only program that worked every time was Mathematica. The subroutine libraries from the National Algorithms Group (NAG) are also an excellent source of high quality mathematical software.
Now in may cases Excel will provide an acceptable answer, but you don’t know when it might fail. For example it will do a least squares fit without checking for ill conditioning, and it can sometimes give you a completely wrong answer without any warnings.
Again my comments are limited to older versions of Excel, but in the past MS has not demonstrated a commitment to providing numerically reliable software.
And when you litigate, I'm sure the guy in robes needs a lot of help from folks like Special Masters who had a class beyond baby calculus, right?
It's probably okay to use Excel for the design of fusion reactors: they don't work yet anyhow. :)
Congress created the Court of Appeals for the Federal Circuit for many of the reasons you give. You may wish to look at the bios of the judges on that court. On cursory inspection, I believe Judges Gajarsa, Linn and Moore all have backgrounds in electrical engineering, and Judge Newman has a doctorate in chemistry.
I wonder if part of the problem is that the Court of Appeals for the Federal Circuit has such narrow jurisdiction that the judges appointed to it, however well regarded within their specialties, are not generally thought of as having the kind of relatively broad knowledge desirable on the Supreme Court? By the same token, how often do patent lawyers become judges, especially at the appellate level?
First, I'm not sure that it's a "problem" that the Federal Circuit is not a stepping stone to the Supreme Court, the way that the D.C. Circuit is.
As for the Federal Circuit's narrow focus, the court handles patent cases from all over the country, and that is probably the bulk of its caseload. However, the Federal Circuit also hears cases involving veteran's affairs, international trade, federal taxes and government contracts (the last two in appeals from the Court of Claims).
There was a proposal back in early 2006 (if I recall correctly) to transfer appellate jurisdiction for all immigration appeals to the Federal Circuit, but that proposal died a quiet death.
The Federal Circuit docket does not include a criminal law component (they are the only cicuit court not flooded with habeas petitions, I believe), and that could be seen as a gap in experience. On the other hand, many trial and appellate judges come to the bench from civil practice with almost no exposure to criminal law, even though criminal law occupies much the time for most federal judges. Still, you may have a point: most presidents would probably be wary of appointing a judge to SCOTUS who had no track record on criminal cases.
I have no idea how represented patent lawyers are among the judiciary compared to their representation in the legal profession; but I know that both percentages are very small.
Yes, Excel has accuracy problems in the nether regions. Heck, I have written patents on increasing the accuracy of arithmetic functions (esp. trig functions). But in real life, that isn't the sort of thing that I deal with. Rather, I deal with the sort of problems that don't stress Excel's accuracy.
In the case of having to round a bunch of numbers to a certain accuracy, I will typically round one of them in another column (or row) to the proscribed accuracy. And then, I will duplicate that cell through the rest of the column or row, as needed. The default of cut and paste in Excel is to paste formulae, adjusting the coordinates of the operands, as needed. You can use this technique any time you need to do anything repetitive.
Also, Excel is being taught in a lot of schools now for just this sort of all-in-one tool. I know of a lot of kids who were adept at it by the time they hit high school.
Note that the only reason that I became used to using Excel over its competitors is that I found myself working for companies and a law firm that mandated its use. After a couple of years, I found myself more adept at using it than, for example, Quattro Pro, and thus began to prefer it. But other spreadsheet programs can do essentially the same sorts of things with about the same amount of work, once you gain familiarity with them.
Finally, my background is somewhat similar to EVs, and overlapped his a bit. I worked as a software engineer from 1975 through 1990 specializing in operating systems and data communications. So, not surprisingly, my patent practice since then has concentrated on software and computer architectures.
The answer to your rounding question is under:
Tools—>Options—>Calculation—>check the box "Precision as displayed"
then set the display to show only two digits.