Perhaps what makes me uneasy in the above is the overloaded term, “information”. In my book, computers don’t do information. They do data. Data is a difference - information is a difference that makes a difference. Since it makes no difference to a computer what it’s processing, it’s not information.

My gut is also disagreeing with the bit about quality attributes, but not telling my head anything.

For now, I want to stay with “information.” I agree that the computer doesn’t care what it’s processing. I’m assuming that if somebody spent the time and effort to automate the processing, there must be people somewhere for whom the data makes a difference.

One aspect of the post that bugs me is the absoluteness of “always and entirely.” I generally avoid absolutes like that. I wrote it as an absolute anyway, partly because I wanted to experiment with absoluteness, and partly because I can’t think of a single counterexample.

If you discover more clearly what you’re objecting to, I’d love to hear about it.

If I substitute “data” for “information”, your first sentence still bugs me. Here’s why in a nutshell: you say that a computer can do only three things. Mine does more than three things - it can send email for me, run a game for me, draw pictures at my bidding, play music, and so on. So I can’t take your statement literally.

Perhaps my gut is only signaling confusion. But I suspect that my head read your statement as “anything a computer does can be reduced to one of three things”. That makes more sense of my gut reaction - “reduce to” is a problematic concept. For instance, what *people* do also reduces to things that are relatively trivial. That tells us no more about the purpose of humanity than the deep reduction of computer things to “moving data around” tells us about the purpose of automation.

I think that the term “purpose” at the end is also one that my gut identifies as problematic. I don’t know that *my* purposes in automating something are, always and entirely, to improve the quality attributes of a process. Yet I spend a lot of time automating things (less time now than I used to). Sometimes my purpose in doing so is to have fun, or to learn stuff. I don’t know that these have to do with quality attributes of a process.

We have SQL itself and it’s formal sublanguages (DML, DCL and DDL). Programming languages (Java, Python), Data definition languages (HTML, XML). Metadata languages: DTD, XSDL, XSLT, etc. Don’t forget analysis and design pattern languages to simplify and standardize discourse.

I think that there are cases where the formalism of application software trumps speed. We need application software to do a job that can be done in Excel. Why? Excel is always ad-hoc. We want something more formal.

The problem with your statement is that it does
define ‘what computers, via programs, do’ but
it does not distinguish between the many and
varied ways in which computers are used. A system
for stock control could cause an order to be
sent to a supplier without any human intervention.
An alternative system might display a possible
purchase to the user, who confirms it, and the
order is sent automatically to the supplier.

The first is automation; I am not what label to
apply to the second solution- it is not TOTAL
automation, it is not Decision Support.