54 thoughts on “Weekend reflections”

  1. I’ve been thinking a lot about blog commentators, and I really believe that anonymity or the use of pseudonyms is spineless, regardless of the political persuasion of the comment/commentator.

    Many argue that the use of real name is pointless, as the name has no meaning to people who don’t know them. I argue that there is an inherent honesty associated with using your name, even if no one knows who you are. By putting your name to your comments, you take an intellectual and moral responsibility for them, and to your self, that I don’t believe is possible by using anonymity.

    Another argument I hear involves the need to protect ones employment. My rather blunt response is that if your opinions are incongruent with your job to the point that expressing them may jeopardise your position, I suggest you find another job.

  3. I agree with David too.

    Although I’d have a rather blunt response to:
    “My rather blunt response is that if your opinions are incongruent with your job to the point that expressing them may jeopardise your position, I suggest you find another job.”

  4. In principle we agree with David, however, there is also the practical concern that by posting real names, there is always a backgroud worry that some miscreant will chase you down in person in order to continue to argue the online debate.

  6. The assumption is that it is employees who have to worry.

    I have a business that pays for me and my employees.

    I do not want to lose current or potential clients because of my personal views.

    What do I say to the employee that I have to fire when one of my largest clients leaves because they don’t like my views? David Hiedelberg made me do it?

    Well???

  7. When someone posts under a false pseudonym (or pseudonyms as the case may be), they take the extra cost of their post being treated as less important. Someone posting under (what appears to be) a real name is bound to get more interest in what they’re saying than if they called themselves something really stupid (like “Razor” or even worse, “alpaca”). I think people recognise this and accept the trade between having your comments taken seriously and the potential loss of your job.

  8. A pseudonym or a nom de plume is mroe than acceptable, as long as you are consistent in fornt of the same audience. You knowing my real name means nothing whatsoever unless you want to know me FTF – and you don’t get to do that without my invitation. As long as I am consistently known by the same tag, it’s irrelevant what I choose to call myself.

  9. *I’ve been thinking a lot about blog commentators, and I really believe that anonymity or the use of pseudonyms is spineless

    Using my real name let me just say that I also agree with David.

    Although I think the anonymous options should remain. Even spineless people should be allowed to express their views.

  10. David

    I was just writing about this elsewhere from a different angle. To me, the weak point of blogs is not the pseudonyms themselves, but the general lack of personal candour. Posters are typically willing to issue hairy chested pronouncements and buckets of spleen in a ostensibly courageous manner but how much real courage does that take? What I rarely see is any kind of admission of imperfection, getting it wrong, or even (God forbid) a public shifting of views on a subject. Which is, I would hope, why we’re all here: to learn and move on. I’m yet to see the words “Good point!” on a blog comments page.

    It’s as if these supposedly insightful independent-minded people have totally succumbed to an uninspected (and psychologically-primitive) urge to be “right”. When I read blogs I find myself in the company of some bright people who obviously know a more that me about all sorts of subjects, but there’s a half crazy ring to the proceedings. If a person is unable to adjust their views and admit personal limits, for me at least, it depreciates their other arguments: are they just arguing that position because it resonates with them emotionally and it’s too psychologically difficult to stop?

    Dave Barry said, “I argue very well. Ask any of my remaining friends. I can win an argument on any topic, against any opponent. People know this, and steer clear of me at parties. Often, as a sign of their great respect, they don’t even invite me.” It’s a bit like that for me on blog comments pages, I’m often discouraged from entering a discussion because it seems like I’d be engaging with people who I am unlikely to influence.

  11. Jim, is the blog experience you describe (which is a generally fair characterisation) really terribly different from any other sphere of human interaction?

  12. Jim,

    I have been debating ideas online for 6 or 7 years now. My opinions have changed significantly in that time. It is usually not due to a single encounter but due to a cumulative set of encounters. So individuals may not see the effect that they have on others such as myself.

    * 7 years ago I thought that gun prohibition was generally a good idea.
    * 7 years ago I thought that money was an innovation invented by governments.
    * 7 years ago I thought that inflation targeting was best practice monetary policy.
    * 7 years ago I thought that our high taxes were okay so long as the money was spent carefully.

    Some things have not changed. I still think drug prohibition is mostly counter productive.

    Regards,
    Terje.

  13. The Currency Lad, implied that my criticism of anonymity was hypocritical, as I only directed it at RWDB. He challenged me here.

    Here’s what I want you to do. Join a thread discussion at three liberal blogs and firmly denounce its participants for posting pseudonymously. I want you to do this for me before 5 pm today. Please obey me on this. (emphasis mine)

    I strongly believe in what I say, and believe it applies to all people, even those whom I agree with.

  14. I don’t wish to tell people how they should dress. However I do ponder peoples motives.

    The Hijab is often worn by Muslim women with the objective being modesty. However in a nation such as Australia where such clothing is out of the norm might it not be that the motive is actually pride rather than modesty?

    Perhaps my characterisation of modesty is misplaced. I am happy if somebody wants to enlighten me.

  15. As one who has been the victim of Internet stalker, I can assure you it is a positively frightening experience especially when that stalker has got to the point where he can tell you what school your children go to.

    Also the use of a pseudonym can enable people to be more honest (yes, and at the same time perhaps abusive and worse).

  16. Lots of legitimate reasons to use pseudonyms online. Mr Lefty because of his job although he brought that on himself by braking client confidentiality. Loosing career opportunities or even there jobs isn’t an alternative to those that do have them David.

  17. Here’s a different topic.

    Some contributors to this site are social scientists who would have experience with factor analysis and its kin. There must be ten thousand researchers using FA every day and there is endless literature on how to determine the important factors and lots of other esoteric stuff. But has anyone ever checked to see if FA works? As far as I can tell, it does not.

    I have generated many sets of artificial data containing 4 clear — sometimes perfect — factors but FA always tells me there are 3 factors. What am I doing wrong?

    Here is an example.

    I constructed an imaginary data set of 100 responses to a questionnaire of 20 imaginary statements (making 20 variables). Each statement is answered on an ordinary 5 point Likert scale, scoring 2, 1, 0, -1, or -2, corresponding to the range “strongly agree� to “strongly disagree.� Of the 100 respondents, the first 25 answered the first five questions either with a 1 or a 2 (randomly assigned) and answered the other 15 questions with anything from -2 to +2 (randomly assigned). The second 25 respondents answered statements 6 to 10 with a 1 or a 2 and of course the other 15 with anything from -2 to +2. Similarly for respondents 51 to 75 with statements 11 to 15, and for respondents 76 to 100 with statements 16 to 20.

    In short: everyone scores unequivocal positive on all statements reflecting their “view� and there are no undecided respondents. A researcher with results like that would be very excited. But factor analysis gives 3 factors, not 4. The alleged 3 each account for 12% of variation and subsequent factors trickle down from 6%. Generating new data and re-running changes nothing.

    The outcome is the same if the data are elevated to fanatic level. I restricted responses to the “other” 15 statements to the range -2, -1, 0. That created four perfect groups each utterly agreeing with its five statements and utterly disagreeing or not agreeing with the other 15. Factor analysis yields 3 factors, each supposedly accounting for 13%.

    If you want to try it, the following Excel formulae will generate the data for import into a statistics program:
    =INT(RAND()+0.5)+1 yields a 1 or a 2
    =INT(RAND()*5+1)-3 yields 2, 1, 0, -1 or -2
    =INT(RAND()*3+1)-3 yields 0, -1 or -2

    I have also generated Q-sort data and found similar outcomes. In various permutations of artificial data containing 4 preposterously unambiguous factors, factor analysis always gives 3 factors.

    You’d think someone, somewhere, in the last 60 years would have checked to see if the process worked. I can’t find anyone.

    Any ideas?

  18. For some reason the formulae have not pasted properly.

    =INT ( RAND ()+0.5)+1 yields a 1 or a 2
    =INT ( RAND ()*5+1)-3 yields 2, 1, 0, -1 or -2
    =INT ( RAND ()*3+1)-3 yields 0, -1 or -2

  20. I agree about the pseudonyms David H, but for myself only with respect to those invertebrate, lily livered, intestinally fortitude deficient practitioners who insult, incite and defame while hiding behind certain well known pseudonyms.

  21. Can I just say, how unhappy, I am ,at the death of Van Nguyen.

    We have a ‘free trade agreement’, with Singapore , where civil rights should have been on the table,as an issue, from the start. Suspect our P.M., quite likes their template for a well ordered society.

  22. I am sorry that Van is dead.

    I have a lot of stressful things going on in my life at the moment. Thinking of Van Nguyen gives me perspective and makes me feel happy. In relative terms my week has been just fantastic.

  24. Wilful asked: Is the blog experience you describe really terribly different from any other sphere of human interaction?

    I know what you mean. It’s not very different from your typical argument, but quite different from better conversations. Lots of human activities are not framed as combat, and they are usually more enjoyable and more productive. It’s a mind set, isn’t it.

  25. They didn’t even let his mum give him a hug.

    Thinking about the things I have and done said online in the last two years, I don’t wish I had used a pseudonym, although I haven’t had eny bricks through eny windows either.

    Somehow I think it would be too late to start using a pseudonym without being found out and humilated. The on line equivilent of fruit throwing and defrocking.

  28. In an ideal world we wouldn’t need anonymity. But in the real world in which I (and everyone else) live there can be very nasty and unjustified consequences for a person publicly expressing an honest opinion on a sensitive and important issue, in a way that allows others to easily identify the commentator, and this can seriously limit the range of views we need to hear. Anonymity allows a much more frank, robust and diverse exchange of views, without fear of retribution from a range of possible sources.

    Another legitimate advantage of anonymity on discussion sites is that it allows people to more readily evolve their thinking about an issue and change their minds, without losing as much face as they would if they used their real name.

    Furthermore, if you knew just how much info about your online activities is being increasingly stored in databases around the world, often permanently (by Google for example, and very often accessible by various governments), and just how easy it is to use this info to profile people’s online and offline activities, you might think twice about be so ready to publicly identify yourself on discussion sites. You may believe that you have nothing to fear from that practice, but I have a less sanguine view, and I still have a right to express my honestly held opinions.

    I agree that a few commentators abuse the privilege of anonymity, but so what? As long as anonymous comments aren’t defamatory or particularly abusive, and most aren’t, and as long as anonymous commentators always use the same pseudonym on a given site, then what is the problem? If you don’t like anonymous comments then just ignore them. Alternatively, only visit sites like Webdiary that don’t allow anonymity (except in special circumstances). Or set up your own discussion site, don’t allow anonymity, and see how many comments and commentators you get.

    My experience is that discussion-based sites tend to be fairly self-regulating and that abusive or ill-considered posts generally get jumped on pretty quickly and called to account, or removed and the commentator banned by the site owner. Furthermore, the kind of anonymity provided by most online discussion sites is not true anonymity, because you generally have to provide at least an e-mail address to be allowed to post a comment on the site, so if you really are out of line then the site owner can report you to the authorities.

    What about anonymous or protected sources used by the conventional media? And the anonymous comments on talkback radio? Are they spineless too?

    But the main issue is: What does providing your real name ultimately have to do with the quality of the evidence and argument you put up? Is 1+1=2 any more true if David Heidelberg says it, or less true if Seeker says it?

    I strongly disagree that commenting anonymously is “spineless”, and I find that comment ill-considered, naive and even somewhat abusive.

    On a completely different and much more important topic. I disapprove of van Nguyen’s drug smuggling, but I disapprove far more of this insanely costly and unwinnable ‘war on drugs’, and I am totally opposed to the death penalty. He should have been given a stiff jail term, mostly for the crime of gross stupidity. My deepest sympathy to van Nguyen’s mother, especially for not being able to hold her son one last time, a particularly offensive and inhumane rule on the part of the Singapore government. I hope she eventually finds some peace after all this.

  29. I use my own name because for me it has no disadvantages. It won’t cause problems at work, and I’m most unlikely to be a target for a stalker. The advantage is that people who know me might bump into me on the blog and say hello. It’s the same principle that governs my decision not to wear a mask when I walk the streets.

    But I can well imagine that for others the disadvanteges outweigh the advantages. In any case, once you’re an established blog identity, and people get to know you, your identity is as real as any other, with its own history and reputation. You’re as accountable as anyone else for the quality of your arguments. I don’t think of Nabakov or Katz as any less real a person than David Heidelberg, who could be a collective of female con-artists for all I know.

  31. Mike P :

    I hardly every use factor analysis, so this might be incorrect, but if I remember correctly, factor analysis picks up groups of variables that are correlated.

    Your functions are misleading because whilst they appear to look correlated, actually, the correlations between variables they produce are quite small. (you can look with =correl(A1:A…; b1:B…) )

    Thus factor analysis shouldn’t pick them up, because they are not related and factor analysis is looking for groups of correlated variables, not variables with similar absolute values (which your function generates).

    Try using a function like

    a1 = rand()

    then in the next column

    b1 = a1+ 0.01 * rand()

    which will give you highly correlated variables and then see what happens.

  32. To John Quiggin –
    I use SPSS and I am no expert but the progam wouldn’t make a difference. I have not done it with these data but I have seen in classes data run through various programs and though they vary in mode of use, they give identical results. Anything else would be a mistake in the basic maths.

    To conrad –
    I am bound to get correlations if I induce them. I can’t really go making one answer depend on another. The data have to be digits between -2 and +2. That’s all a person can respond with. Those functions merely provide a way of simulating that.
    SPSS tabulates all the correlations (in effect applies correl to everything) and you are right – they are all over the place. Still – I have four utterly clear factors in stock standard questionnaire form. FA is what people do with such data and here it fails though the data are far clearer than any real data could ever be.

  35. Mike :

    Your function simulates independent events taken from a given distribution. This means that with infinite cases your variables will be completely uncorrelated, even if they are taken from exactly the same distribution.

    Factor analysis is designed to pick up groups of variables that are correlated — a bit like doing a cluster analysis on a cross-correlation matrix. If there is no correlation, which is the case with your variables with infinite samples, then factor analysis should not be able to pick anything up but noise.

  37. They didn’t even let his mum give him a hug.

    However, they have given the family and friends such consolation as can be derived from such things by allowing them to take the poor bugger’s remains and give it a decent farewell.

    British and Australian practice on such occasions was to cut the body down (after the hour dangling as prescribed by law), carry out an inquest and once the coronor handed down his verdict remove it to a yard usually within the gaol’s grounds and bury it in a pit into which was thrown quicklime to aid decomposition. The grave was usually unmarked and was of course in unconcecrated ground.

    Such was the vengence of the majestic law of our civilised Anglo nations.

    Having worked in a Borstal adjacent to a gaol where an execution was carried out one bleak November morning I am here to tell you that the effect on all who carried out their duty was that they were all firm opponents of the death penalty by about 20 past eight that morning.

  38. Now Ugly Dave has just added an important part of the blogo-de-plume discussion. I have no knowledge of who he is, except that he is more beautiful than me. But he did add a detail from experience, which was vivid for me and I really enjoy reading that kind of encapsulating event. Metaphor and all.

    So, as a non-nom-de-plume person, I am really in favour of a system of anonymity which enables people to say stuff which they couldn’t say otherwise. I gain from it.

    But I do generally assume that the plumesters have a reason for it because it is second best. I like to have a rounded sense of the people I am reading and sometimes engaging with, and it helps to have a name. Aside from anything else, I can google it and get a little bit more of a person’s public presence.

    However, I have wondered about using my own name. For good or ill, I am stuck with a presence sprayed across the internet which pretty clearly shows who I am, particularly late at night, fur brained and disinhibited.

  39. Ugly Dave, that wasn’t part of the vengeance of the law. It was to stop people trying to revive the victim and/or claiming he had survived. These things had happened before, as had a corpse falling into the hands of surgeons (who were sometimes startled to see that their dissections achieved resuscitation!).

    So it’s a survival of a practice that had a practical reason (just like banning close contact family visits in case of Tale of Two Cities style scenarios).

  40. Conrad –
    “Your function simulates independent events taken from a given distribution.�

    I don’t understand this. What is this “given distribution�?

    Maybe this is getting somewhere. You are saying that factor analysis is unsuitable for analysing these data. If I can just get the reasons straight, then that would be pretty much what I need to know.

    With infinite cases (as with 100 cases), a quarter of them are in almost complete agreement on a quarter of the variables. That is, those five variables are indicating something precious to a quarter of the respondents. To three quarters of the respondents these five statements have no systematic meaning so they will be poorly correlated. And indeed they are, typically 0.1 or 0.2. Looking at the correlation matrix, you can just barely detect the pattern of the four groups.

    It still baffles me: if you had a set of Likert data with four utterly clear factors, what would it look like if not as I have described?

    Out of interest I just ran a K means cluster analysis specifying four clusters and it correctly allocated all but four of the hundred respondents.

  41. I use a semi-anonymous handle for a simple reason. I wish to post as an individual. However, when I say something as an individual it is not uncommon for me to find that people conclude that it is the official position of organisations I am prominently involved in. Sometimes they do this out of malice, in other cases out of stupidity.

    People may be able to work out who I am from some of my posts, but they can’t prove it. They also can’t google my surname to find whether I have said anything stupid.

    I want the right to be able to say what I believe, sometimes to put out first reactions which may alter with further debate. If I have to use my full name I’d almost never post at all – I’ve had too many things I’ve said used against not only myself but innocent people in the past.

  42. David Heidelberg said:

    I’ve been thinking a lot about blog commentators, and I really believe that anonymity or the use of pseudonyms is spineless, regardless of the political persuasion of the comment/commentator.

    I don’t know, David. It’s not my habit to use a pseudonym. I’ve only been using one since I’ve been blogging from Việt Nam, the last year or so. But for that, I make no apology. One may not intend to criticize the government of the day, but one may be percieved to criticize. I think I’ll spare me the thrills of doorknocks in the night, thank you.

  43. I got to thinking about this whole pseudonymity issue that’s come up yet again. I got to writing about it too, but I decided that I wasn’t going to waste the results in any comments threads. So there.

  44. Mike :

    Your distribution is just the function you create with the little excel command. Your data is independent because variable A doesn’t rely on variable B at all — i.e., I can’t predict A from B.

    Factor analysis, or for that matter, any statistical technique that works based on picking up correlations amongst variables won’t be any good — It won’t pick up things like differences in distributions.

    An example of this might be a survey with lots of questions about soccer and lots of questions about AFL. The questions also need to be about some of the things that people have an opinion on if they care, but where their opinion is basically random from one person to the next.

    i.e., where a lot of people will not care at all about one but will care about the other (I like soccer, but couldn’t care less about AFL).

    If you asked me a pile of questions about the uniforms of both (what colour should they be etc), I might give a different distribution of answers to the soccer questions, and random noise to the AFL question. However, the SD of my random noise might be less than the answers to the soccer questions and my answers might not be at all correlated with the next person (you think the uniforms should be red with great passion, I think they should be blue…). You might be able to find another group of people that are like me except they don’t care about soccer but do about AFL.

    In this case you will get variables that have the some predictability, but the predictability is in the SD and not the mean or their relationship to other variables. i.e., variables with one pole, where the “neutral” end = I don’t care, and questions where people need not be at the “I don’t care” end of the pole to answer, but where the answers are polarized but independent.

    So you might have to obvious patterns in you data that won’t be picked up by a factor analysis.

    An example data set where might look like this (which yours does with more variables). First two columns = soccer, second two = AFL. Both have the same means an SDs and don’t correlate across variables, but its obvious there are two variables.

    7 1 3 4
    1 7 4 3
    2 6 4 3
    6 2 3 4
    3 4 7 1
    4 3 1 7
    4 3 2 6
    3 4 6 2

  45. Sorry, Ithink i typed in a perfect negative correlation.

    7 7 3 3
    1 7 4 4
    7 1 3 4
    1 1 4 3
    3 3 7 7
    4 4 1 1
    3 4 1 7
    4 3 7 1

    is a better example

  46. I have used electronic messaging since the late 1980’s, usually with my full name attached in the signature, with postal address, home work & mobile numbers, home and work e-mail addresses, and have NEVER been stalked. Perhaps my posts are too dull and uninteresting for anyone to persue!

  48. With Phil Ruddock ,at the helm, as attorney-general, I reckon, the anonymous comment should be invited,glorified, rather than considered spineless. Bring it on as a counter to sedition law.

    Spinelessly yours,etc

  49. Doug, I notice that you did not reveal your full name this time, let alone your various phone numbers, or physical or electronic addresses.

    I don’t have a problem with that, indeed I think it is a good idea not to. But I am genuinely curious as to why you didn’t do it this time when you so readily did so all the other times?

Comments are closed.