Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?

Comment Matlab does not teach math! (Score 1) 333

Hmmm, When I started Undergrad my school had just started a mandatory laptop program for all freshmen so I had a fancy laptop that I felt should be used for class. I proceeded to sign up for the laptop sections of math where I took multivariable calculus and differential equations using Maple. I aced the course and got quite accomplished with the software but I did not LEARN diff. eq, or even really multivariable calc as I later found out during my 3rd year classical mechanics course. I still only know enough diff to realize how to recognize them most of the time. I also really don't think that it was the professor's fault. I think that math requires a certain amount of mechanical crunching in order to become a useful tool. I therefore really think that math should be performed naked, in PEN, if you are studying physics or engineering then by all means teach numerical methods and use simple software to automate the calculations if it makes sense. Otherwise keep in mind that math is a philosophical toolset, and you don't get better at philosophy or logic with wikipedia, you get better by arguing. Programming is a skill that should be taught in school, research is also another useful skill that these days requires internet, (though primary source essays are a better method IMHO), but school is not the place to train white collar workers how to use PWP.

Comment Re:You're framing it wrong (Score 1) 274

Maybe its because I'm doing scientific computing but I think you are lost on your time scales here. You say it takes years to develop an algorithm. I'd say I develope new algorithms every few weeks. Then I spend a long time figuring out which ones actually work well and how to compare them and choose the 'best' for my problem. I publish the resulting software which is protected by copyright and I don't give you the source. Now how are you supposed to have ANY idea how my algorithm works. Its a black box and you have to gain my experience to expand upon it. Therefore my work has value, if you want an expert on solving this sort of question you have to hire me. If you can solve what took me years and layers to develop into a working piece of software quickly, well then I look pretty dumb. In retrospect I've just convinced myself of an underthought advantage to patents. If I do alot of work and am convinced that I know some tricks to solving a problem and you convince me to sell them to you then a patent is a nice way to transfer those rights. Though now that I'm thinking a bit harder I think that a legally binding contract can be just as useful if I just want to share the punchline with you. If you share my punchline without permission or have it stolen then I go through the same (but maybe cleaner, since it is a contract) legal mess to sue you for damages. I do think that formal copyright on software should have source included, but for algorithms I think trade secrets are much better for business and society in the long run.

Slashdot Top Deals

A mathematician is a device for turning coffee into theorems. -- P. Erdos