Thanks for sharing this document, which raises some good issues. However, it represents a pretty narrow interpretation of the law, especially in view of the fact that the government itself uses electronic signatures quite extensively these days. You can even sign your tax return electronically! This document assumes a rather high degree of risk to the human subjects, and in reality, there is (or ought to be) a lot more room for negotiation of these matters with the IRB depending on the nature of the study.
there are multiple meanings of "signature" being used here, by various folks. first, we have signature as in "the signature i use to sign the bottom of an informed consent form". handwritten, et cetera. second, we have signature as in a certification that you have read, do wish to submit, etcetera - typically implemented as you typing in a password, clicking a checkbox, or typing in a captcha-style string of characters. this is different (in a lot of ways) than the first meaning of signature. first of all, method 2 is trivially forgeable, and doesn't share many of the properties of method 1 - relatively unique handwriting styles, evidence of coercion or nervousness in people's responses, et cetera. second, we have to consider whether there are good reasons for the utexas.edu-style 'you must get a real signature either f2f or by fax' rule to be as it is. perhaps they have simply not seen anyone present a method of implementing signature-style-number-2 that can pass muster as a unique identifier that certifies a person's intentions. --elijah