Slashdot Mirror

"premature optimization is the root of all evil"

This is the quote and the relevant word is 'premature.' No one is arguing that we should never optimize. The problem is if we optimize thinking that 'we know best.' It's as if somehow, magically, we know where the slowdowns are without doing actual performance monitoring. "It's a O(n^3) loop, so it must be bad."

Optimizations frequently cause the code to be less maintainable. Optimizations can also take time to implement and document. Also, optimizations are only important in areas that we care about. If an optimization makes code run 5 milliseconds faster in an area where we don't mind waiting 10 seconds, then that optimization is not useful.

An O(n^3) loop is perfectly fine if: 1. it performs fast enough and... 2. it's allows for speed of implementation and code readability. However, if you perform an analysis and find that it is sucking up a bunch of time, then by all means, improve it.

But not until you've got the results of that analysis. And not a 'hunch.'

Optimizaion: Your Worst Enemy
http://www.flounder.com/optimization.htm
Joseph M. Newcomer

Yup, by now it should be well-known that they are unable to design anything worth any crap. That M company somehow managed to spoil and mentally disrupt decades of programmers. See http://www.flounder.com/guiowner.htm etc. for radio button stuff and many other scary things.

There's no excuse for an outright incomplete CS program. Blaming the choice of language doesn't make the program any better. Here at my university (Northern Michigan University), we use Java for the first year of courses, before moving on to C, a nice survey class (introductions to LISP, Prolog, and such), and finally upper-level courses using whatever language is most suitable. Introducing students to programming through Java has harmed nothing, in my opinion. In Java, it's entirely possible to understand the basic concepts of pointers. Granted, you can't do the neat little tricks with pointer math, but those kinds of things are hardly wanted for introductory courses anyway. As for programming with a GUI, all I have to say is "Welcome to 1990." The Average Joe using a computer has been familiar with GUIs for at least two decades now. I personally love the command line (its power, especially for use with scripting, is unmatched), but for some programs, especially in my field (custom programming for small businesses), you need a GUI, just because that's all the user will be able to use. To concentrate on command-line programs throughout the short four years of college is hardly a service to the student. The code behind the GUI is still code, and today's students need to master both aspects. Regarding garbage collectors, they are perfectly acceptable as a programming tool. Yes, a perfect programmer will be able to delete his own objects more efficiently, but that's no reason to remove a useful aid from use when it just doesn't matter. A small program, where most of the processor is wasted waiting for user input, can afford to run a garbage collector, especially if it means the program is ready for use earlier than expected. Use that time for testing, code review, or interface design, instead of wasting it writing unnecessary routines. Reading this article (and re-reading the original), I am constantly reminded of another site I saw, discussing the state of CS programs, as compared to bricks. It's titled "A Degree in Brick Science", and is available at http://www.flounder.com/bricks.htm. This article discusses the same problems as Dewar does, but finds an entirely different conclusion: the underlying problem is how the universities arrange their CS courses. This makes far more sense than blindly blaming the current language of choice for introducing new students to the joys of programming. In my opinion, any CS program worth the name should cover all the topics necessary. What language they use is inconsequential. This is the same rejection of a new tool that has eliminated BASIC from the repertoire of most schools. I learned BASIC first, left it, and have not returned. I took with me the knowledge of how to break up large problems into small ones, and turn those small problems into individual steps to form a final program. I took my experience with boolean logic, my love for string manipulation, and my frustration with stupid mistakes. I moved on. It has not made me less of a programmer at all. Rather, it made me better. I learned quickly my strengths and weaknesses, and have carried them with me. I see Java in the same way: It provides a place for eager students to explore the possibilities without needing to know the details of the theory. Of course, the best programmers will always be the ones who know both. In our push for more theoretical knowledge, we must be careful not to do what Dewar has done. We must not ignore progress in exchange for upholding antiquated traditions. We could take Dewar's argument to an extreme, and claim that we should teach students to set 64 switches at a time to program a modern processor. This would quickly be rejected. Yet it is a more pure way to program, to actually have to think about the individual wires. Assembly language is a tool to make machine code easier to write and understand. C++ is a tool to make assembly easier to write and understand. Why must we stop there? -Sarten-X

Though I withhold judgement until some confirmation comes.

The point isn't "wow, I can scan any document and paste it into Microsoft Word as a gif image". The point is

1: "wow, if I write this into Microsoft Word 2004 with default settings without adjustment, and print to file, and place it on top of the CBS memo, they match. What's the chance of that?"
(remember aliasing and faxing causing jagged letters)

And

2: "Wow, we can't find a single typewriter in existence in the early 1970s that has all the typographical characteristics that the documents have"

I don't use "right" or "left" blogs as my final judges on news. For example, "left" blogs (in this case the Daily Kos) correctly pointed out that the government paying somebody to secretly shill for them was different than a political campaign paying bloggers displaying notices that may or may not have been prominent enough. And the "right" bloggers (Little Green Footballs etc.) were correct about the memo story.

And please reference your claim "even experienced typographic experts who are Democrats and Kerry supports have analyzed it to death and concluded they were obvious fakes made in Word."

He may have been referring to Joseph M. Newcomer, Ph.D., the most prominent example, who isn't a document examiner, but a computer typesetting expert. And with all certified document experts that have examined the documents either concluding that they are forgeries or being unable to reach a conclusion, what are the odds that all those certified experts are evil Republicans? Even the CBS-hired panel expert concluded that the documents were forgeries.

Going on harping on this story is like saying that Saddam had massive stockpiles of French WMD at the day of the invasion, and that extraterrestials lifted them out of the country before they could be found. Is it POSSIBLE? Yes, maybe. But it's rather ridiculous.

Joseph Newcomer doesn't go as far as the blogs did. He won't accuse Hailey of fraud and call for his firing, but he does accuse Hailey of doing bad research.

The Bush "Guard memos" are forgeries!: The Columbia Journalism Review

Personally, I was one of the people who did a directory listing of Hailey's web folder and found unfinished photoshop files before and after he put in superscripts, and the report before he altered it. Saving to several files for safety can bite your arse if you're committing a fraud. Phooey.

A little-known feature of the U.S.S.R. under Communism was that when someone purchased a typewriter, it was delivered to the local police office. The people there took a razor blade and nicked various characters, then registered the owner, the serial number of the typewriter, and a complete sample of the typewritten output. Since the characters exhibited consistent errors, if a samizdat appeared, all that was necessary would be to compare the characters in the document in question with known samples from the registered typewriters, and the offending typewriter could be identified.

He does it so well... I can't add to it.

the second link was incorrect. Here is the right one.