20060730

Poor to Average

My email is somewhat broken at the moment. It's always interesting for me to hear people talk about five 9's reliability, i.e. 99.999%, since this equates to about five minutes of down time in a year. My experience is that any system in real life usually has at best three 9's reliability (about 8 hours down time per year), and I'm reasonably happy with two 9's: three days of downtime over the year, hopefully spread out in small chunks and taking into account various midnight -- 2am fixes that sys admins seem to enjoy so much. Anyway, my remote Unix system went down sometime yesterday around 4pm, coincidentally when I was reuploading my website, but hopefully not caused by this, and the sysadmin has left a message promising its return sometime on monday, which would equate to around 1.5 days of solid downtime, so well on course for two 9's. If we can stay at this level, and not degrade to one 9, then I'll be reasonably happy. Of course, this is as nothing to the pain caused to sensitive teenagers by the Great MySpace Outage of 2006. Meantime, if anyone has sent me email, it seems to be still there (since my inbox is on a different disk to the one containing all my data, which is currently missing, presumed safely backed up on tape somewhere*). But, it will take a while before I get around to answering it. Yeah, that's why I haven't replied to you, computer ate my .pinerc, that's the reason [+].

[* Of course, all my important data is backed up over about six disk drives and three locations; however, I belatedly realize that my sent-mail is not as regularly duplicated; this will change when things get back to normal].
[+ Yes, I still use Pine as my primary email client. You got a problem with that? It's not as if I'm using mail or elm]

Anyway, in lieu of wasting time on email, I've instead been exercising some of my latent aggression on reviewing bad data mining papers. They are all pretty awful. At the moment, I am particularly riled by one entry on the review form I have to fill in. It reads:

How is the presentation?
5 (Excellent)
4 (Good)
3 (Above average)
2 (Below average)
1 (Fair)
0 (Poor)

Firstly, there is something naggingly ungrammatical about the question, but I can't quite say why. But mostly, the presentation of most of the papers I'm reading is pretty bad. They aren't well explained, figures are squeezed so small as to be illegible, they appear to have been written by translating to Portuguese and then to English via Babelfish, and so on. So the average is pretty bad. I would be delighted if I saw many papers whose presentation quality was "fair". "fair" to me means about average, i.e. the expected level. So it irks me that below average in this list is above fair, because based on the papers I have to review, fair would be significantly above the average. It's generally stupid to have "above average" and "below average" as options here. So here is the list that I would like to see:

5 Good
4 Acceptable
3 Poor
2 Bloody awful
1 Absolutely incomprehensible
0 Riddled with spelling mistakes, grammatical errors, illegible diagrams, made-up words, references that don't exist, badly translated idioms, sentences that make no sense whatsoever after repeated attempts to decrypt, and clear evidence that the paper was written in Microsoft word.
-1 Average

No comments: