wmartin@brl-tgr.ARPA (Will Martin ) (11/18/85)
I just did a "grep surv" of everything in net.announce.newusers and found no evidence of there having been guidance published for net users as to the proper use of the net for surveys, and the proper format for such survey forms. I think this might be an area where something should be published, and added to the new-user documentation. This is inspired by running across a very long survey posted by "kozak@mmm", which was posted to multiple groups as separate independent postings. I will list below what I think is wrong with this particular survey; maybe this will inspire discussion that will shake down to a definition of what is allowable and proper in this area. I know I can just "n" past such surveys, and there is no compulsion that I respond to them. I do, in fact, respond to such surveys as I have an interest in and feel that I can contribute toward. However, I also see no difference between posting shareware on the x.sources.x groups in order to get some money in return, and posting a net-wide survey to gather data for a project or academic research. Both are using the network to gain some profit (monetary or intellectual -- it makes no difference), and we've seen a lot of flamage about this lately... What I found improper about this particular survey: 1) It was posted as separate and distinct messages to many (or at least several) groups, thus eating up more resources and failing to get marked as "read" by the latest version/update of news-reading software after the first encounter. 2) It asked for personal data (date of birth, locations lived over your lifetime, etc.) -- true, it did not ask specifically for a name, but only a wizard type can send truly anonymous replies, so the result does tie personal data to an individual's identifier. 3) It is too long. Of course, this is a subjective evaluation, but it does take 4 or 5 or more screens to display the full survey, so the answers will probably include all that plus at least as much again in response -- so there is a large network mail load if there are a large number of mailed responses. Also, it takes too long to respond to. This means that more of peoples' employers' time will be eaten up with this, and more computer resources. It consists of (after the personal-info section) thirty-five expressions, and the surveyor is looking for the terminology people tend to use in these examples. I would think it would take an half-hour or so to complete, especially if you were fully participating and thinking about it to any extent. The kinds of net surveys I would consider "proper" are short, take little time or thought to respond to, and reveal no personal info (except maybe your opinion in what you respond, or in choosing to respond at all). For example, the recent net.tv "five favorite shows" survey, or one like "what UNIX-based database system does your organization use?" would be fine. Instead of posting longer surveys, I would lean toward posting a notice about them, with a mail address for those interested in participating to send for a copy. (The old argument about unreliable mail doesn't apply here -- if you can't communicate by mail with the surveyor in the first place, you couldn't send back your response to the survey anyhow! So an initial filter to select only those with open mail channels between them would be a good thing!) Well, what do you think? Is this a "non-problem" that I have created out of thin air, or is it something that is worthy of published guidance? Will Martin UUCP/USENET: seismo!brl-bmd!wmartin or ARPA/MILNET: wmartin@almsa-1.ARPA