I Want Comment Triage Software

Nobody comments here, so it’s not a personal need, but I want to see comments on blogs and articles organized differently. First I’ll say what I want to see, and then I’ll explain why.

Each comment will be directed to one of four triage categories. These will not be the traditional “Doesn’t need treatment now”, “Needs help”, and “Too hard to help–let him die” categories. Rather, they will be:

1. This author is on a blogger-created list of “I benefited from reading this” commenters, so the comment gets posted on a list that appears directly after the post.

2. This author is not on the list. The comment is posted after all the Category 1 comments.

3. This post contains bad language. It is bounced back to the commenter with a form message saying, “This comment is not civil. Please revise and resubmit.”

4. This author is banned. Bounce the comment back with a form message saying, “You are banned from this site. Don’t bother to comment.”

The blogger would also be able to easily switch the way a comment is directed. In particular, he may demote an individual “I benefited from reading this” comment, elevate a Category 2 comment, demote or elevate or ban a commenter, and bounce back any message with the incivility comment and possibly a personal message.

The reason for these categories is that comments are potentially the best part of a web publication. Crowdsourcing works. I can imagine a blog where the blogger himself is merely an editor, good at selecting a topic and selecting which comments are best— the old newspaper “Letters to the Editor” talent. But it is essential to get rid of the space-filling mediocre, low-quality, or even good-but-not-good-enough comments, or it is too much trouble for the reader to sift the sand to get the gold. The key thing on the Web is Too Much Information.

A separate problem is that the Web breeds corruption. It is too private. The temptation to pornography is the biggest example; Facebook exhibitionism is another. (Exhibitionism may seem the opposite of private, but it is not— the problem is that the exhibitionist does not have to see and be shamed by the bored or scary reactions of the viewers.) Incivility in comments and message boards is another temptation of privacy. We need to rebuke those people, and in some cases they just need basic instruction in good manners. We have two aims: to cleanse the website of their odor, and to teach them to smell better so they will (a) stink up the rest of the Web less, and (b) start to be good company for our own websites. It may not work for burglars, but I think rehabilitation can work for web churls. Indeed, one sign that it will is that they always hide under anonymity, a sign of insecurity that shows how even verbal criticism stings them.

This post is inspired by Instanpundit’s new comment policy. I see that that site puts “most popular” comments at the top, based on reader votes. That’s a first step, but not a very good one. When I say “Crowdsourcing works”, I do not mean that the crowd gets things right on average. In fact, we can depend on the crowd to reliably get things wrong. The average person does not have the talent or taste to detect a good comment. This should be obvious. If the average person ran a blog, would it be any good? No. That’s why Instapundit is a success. We rely not on ourselves, but on Prof. Reynolds for spotting interesting articles and his self-restraint in not discussing them at length (I would guess that his talent for discussing isn’t as great as his talent for spotting; if it is, then I should condemn his self-restraint rather than praising it). If we use a vote of 10,000 readers on what is a good comment, we will almost surely get a wrong answer. If we had a vote of just 10 readers, we would have a somewhat better chance, because maybe the 10 would be an unrepresentative sample and we’d have happened to get 6 smart people, but we can be sure we don’t have 6,000 smart people in a random sample. The Law of Large Numbers works against us, just as in Gambler’s Ruin.

No, Crowdsourcing works because it relies not on the Mean or Median or Mode, but on the Max. Suppose you draw 12 numbers from 0 to 100 randomly (where each number has equal probability), and then draw another 3 numbers. Probably the maximum of the 12 will be bigger than the maximum of the 3. If the quality of the blogger is 90, the Max of the 12 will probably be greater than 90, but the Max of the 3 will not. The median of the 3 and the 12 will almost certainly both be less than 90, but you would not be entirely shocked if the median of 3 were bigger than 90. Here are some random draws to illustrate: (from crowdsourcing)


com 1 com 2 com 3 com 4 com 5 com 6 com 7 com 8 com 9 com 10 com 11 com 12 Max Median Mean

54

41

72

91

56

43

87

41

88

75

67

51

91

61.5

64

13

18

54

77

57

39

60

74

2

11

3

6

77

28.5

35

79

18

12

59

20

73

100

72

32

36

16

53

100

44.5

48

57

9

55

6

45

44

31

38

8

94

11

50

94

41

37

13

7

65

62

96

13

71

43

12

59

65

89

96

60.5

50

47

41

18

76

3

72

52

58

7

58

37

5

76

44

40

8

51

4

80

78

52

51

4

12

56

91

20

91

51

42

43

90

56

26

32

5

40

54

51

36

6

57

90

41.5

41

9

47

49

                 

49

47

35

7

82

67

                 

82

67

52

2

95

51

                 

95

51

49

The talent needed is that of the person who can rank numbers— who knows that comments 1, 2, and 3 have values 9, 47, and 49.

This has gotten digressive, but I’ll add one more digression: does this mean democracy is bad? Yes– if you’re trying to decide questions of fact or taste. No– if you’re trying to decide political questions. Aristocrats and technocrats are smarter and more honorable, but they’re not honorable enough, and they’re still biased. Moreover, they’re not as smart as they think they are. Thus, we must be careful what kind of questions we leave to them.

I think I’ll use this example to teach some stats to my kids when they’re homeschooled this coming fall.

  1. No comments yet.
  1. No trackbacks yet.


Warning: Illegal string offset 'solo_subscribe' in /home/erasmuse/rasmusen.dreamhosters.com/b/wp-content/plugins/subscribe-to-comments.php on line 304

Subscribe without commenting