"if they make large numbers of errors or have difficulties with the command language when these could be reduced by changes to the program, then this must be regarded as a failing of the program and not of its users"
This is an excellent point, ahead of its time, and a controversial position as late as the mid 90s. ("No, it's a training problem.")
> "if they make large numbers of errors or have difficulties with the command language when these could be reduced by changes to the program, then this must be regarded as a failing of the program and not of its users"
By this standard, automobiles have an absolutely terrible design.
How come? The operations of an automobile can and often are learned by even the most stupid ones of all people. Further, all automobiles from the past several decades mostly work the same way: the controls are quite simple and to some extent even intuitive.
The operators of automobiles do however "make large numbers of errors" of varying levels of severity. The linked paper isn't looking just at ability to learn to use the tool in the first place, but also at which kinds of errors crop up during normal use. By the standard being looked at, "user collided with other vehicle while changing lanes" would be an error, and if such errors are common, it would suggest a problem with the design.
Also have to address "when these could be reduced by changes to the program." Though I have heard automobile controls are not optimized- but what became standard first.
Yes and no, I think. There is also a "memoryless" effect at work that causes new users to favour more complex and presumably powerful products, which they then invest in and are reluctant to switch away from. I'm not sure the resulting network effects in the market are an example of path dependence. There is an article by Don Norman on the complexity phenomenon here: http://www.jnd.org/dn.mss/simplicity_is_highly.html
Not really. The path history isn't as relevant as the fact that a very small minority of diehards control the path.
Emacs is a good example. The directions of Emacs are controlled by the small minority who take the time to understand how to create elisp extensions or the even smaler minority who actually contribute code to the main codebase.
What about the transition from manual to automatic transmissions? It's also interesting to note the continued prevalence of manual transmissions in Europe and elsewhere. This is a major way that automobile control can vary.
I am in Ireland and learned to drive a manual transmission in the UK. I didn't know anything about the history of the introduction of automatic transmissions in the US until now, but I just found this interesting history:
http://ateupwithmotor.com/terms-technology-definitions/hydra...
This is an excellent point, ahead of its time, and a controversial position as late as the mid 90s. ("No, it's a training problem.")