While the article raises a few good points, I got distracted by the statements about Scheme at the end of the article. Yes r5rs might be overly minimal, but all real-world scheme implementations contain the usual basic tool functions like string parsing. The performance quoted of a minute for parsing a megabyte sized file can only mean two things: either a defunct scheme was used or the program used contained major flaws - this can happen if you are new to a language, but this is not a useful measure of the language. Even garden-variety byte code interpreters should do this in seconds.