Good software criteria

So what do I want from my software?

• Ease-of-use, productivity and efficiency: I want to prototype new ideas quickly. If successful on a small example, I want to apply the newly developed approach to large data sets.
• Mobility and portability: I want to organize my research for future access by me and others. I want to have access to my virtual laboratory wherever I am. I want to run it on any machine I might come across.

Furthermore, Easy things should be easy, hard things can be hard. (Mathematica and its book by Wolfram are a great example on making simple things simple, and difficult things possible.)

I like to use any tool that helps me to get my work done. But I shy away from tools that seem to make it difficult to switch to a superior tool if it becomes available. Similarly, I shy away from tools that do not collaborate well with other tools I use. For example, a L^ATEX editor that does not allow me access to the standard underlying L^ATEX ASCII text violates both criteria.

Furthermore, I like software tools that are widely used since they tend to be reliable (so called pseudo-standards) and ease the exchange of information and data. ( The masses are great at avoiding bad software. But don't expect silver bullets in mass software. ) Furthermore, new competing software will probably attempt to smoothen the transition for users of standard alternatives.

Example for good standard software solutions to common tasks are L^ATEX for text typesetting, GNU make makefile implementations, or RCS for version control. Only if the software directly relates to my research area might I forego standard software for potential silver bullets.