On the difference between data mining and data compression
Or, to compress the question, the various differences between data compression. I hate having to answering such school-type questions1, but I guess we're always ready to pawn our soul for 10 points :-). So we defined it as being techniques "for reducing the size of data for transmission/storage etc" (paraphrasing), but were ruled wrong. I'm not sure what the correct answer was, but I think it was to the effect of "extracting only that info from data that is needed" (memory fails me, as I was still reeling from the surprise). Without going deeply into data compression, there are different kinds: lossy (like in MP3s where you say that some data/info is unimportant enough to be thrown away) or lossless (such as in file compression, where you have a different, smaller form of your data to cheaply store or transmit). I'm not sure what was needed here.
The quizmaster did assert that he knew what he was looking for, he knew it would be one of this tricky ones, and he had sources who helped him "nail this one". He also did apologise a couple of times later, saying that he knew it was one of those borderline ones (i.e. for the inherent ambiguity, not that he was wrong about it), which was kind of him, I suppose.
However:
- It's a terrible kind of question to ask - school homework-ish at best. Neither was it entertaining.
- If a quiz-setter knows a question/answer is inherently vague or prone to misinterpretation, then he'd be better off not asking it
- If it's not a domain you are familiar with, then you are dependent on hearing keywords in such answers - not a good idea
- Don't ask for 'layman' interpretations of technical questions - my layman lives in Shivajinagar, yours may live in Salt Lake.
I don't have much of a problem with big quizzes being designed to appeal to non-quizzing audiences, but why must they mess with the fairness of the rounds? The use of D&P over Infinite Rebounds (Infinite Bounce to non-Pune quizzers) is a menace that everyone in the game knows about. Most (and all good) quizzes use the IR system for their regular rounds because it guarantees one thing : between two shots at questions for any team, all other teams will have had a chance to attempt a question. IR does not smooth the fairness of the content or the fortune of receiving an easy question. But unlike D&P, you can be assured that your team will not twiddle its thumbs while the rest get to answer. Think of it as all teams receiving (almost) the same number of deliveries to bat to, at any given interval in time.
Yet another illustration of this from the other day: after 2 rounds of passing (one forward, one reverse), the number of attempts (out of 12 questions): two: A, B, F; three: E; five: C, D. Five vs. Two - that kind of imbalance would never happen in IR.
(This is no reflection on the qualities of the other teams, esp. the winners - just a commentary on why a quiz format shouldn't be so uneven. Let luck of the draw play its role in the content and team placement, but not the number of attempts.)
So why do big quizzes still employ an outdated and inferior system? Makes no sense at all.
1. read up for the future: the difference between ink pens and fountain pens; between MP3s and MP4s; between first slip and third slip (about a feet?)
12 comments:
'Even' I have been guilty of this in the past but I have realized that it's like the Ashes versus IPL. Both are different and cater to different people and sensibilities. The audience at BEQ doesn't care about fairness; D&P is easier for the audience (esp. those who turn up for 1 quiz every year) to understand simply because they haven't seen anything else. IMO, when the express objective is mercenary; all cribs should be held back. It is a show not a quiz;I don't think anyone has any misconceptions about it.
An SRT/ACG doesn't crib that IPL is unfair since they now have to, at the behest of the team owners, endure SKW's idea of post-match evenings :-)
Your HR/Centre head (team owners) won't care about you winning a fair Landmark Open; they will care about you not answering data compression to Derek's satisfaction and missing out :)
Anyways, how many chairs, keyboards and anti-virus software packs this time?
(oh no - sports analogy alert! :-) )
Harish: I know the audience doesn't care - they don't need to. But the QM and quizzers can and do. My view is that the choice of passing format in rounds doesn't affect the audience one way or another. If quiz organisers choose to stick to D&P because it is easier to understand, then that is just being lazy. Derek has done IR at some of the last few Landmarks he was involved in, and I don't think it had any impact on the audience's comprehension.
Perhaps like with IPL audiences, we should stop worrying about the intelligence of BEQ audiences? :-)
If I (or my org) pays for entrance, then I also am a stakeholder, and would like my share of cribs to be made :-) (and I'm not even blaming it on a pre-quiz party)
Finally: 2 anti-virus packs. Want one?
>>It's a terrible kind of question to ask - school homework-ish at best. Neither was it entertaining.
I disagree that this type of question is not entertaining. Just ask anyone who attended the 2005 COEP Tech Quiz hosted by Siddharth N about the entertaining exposition on the difference between turbochargers and superchargers we were all treated to by Siddharth D and Anupam A.
Ok, let me introduce a caveat: such a question is only entertaining if answered for 5 mins by a team consisting of Dani & Anna. :-)
Made me think of becoming a professional quiz-master myself. Then we would have half-decent quizzes, non-brain-dead scoring systems, new ways of catching audience attention (while not resorting to flashing and other cheap tactics normally followed) not to mention a verbal massacre of the innocent (audience) if they tend to get in the way.
By the way, an entire row of aunties wanted a prize at one point in time. They were duly given aunty-virus.
Boys, boys !
When in doubt, run to IR !! is the war cry. IR doesn't work. I have proved that incontrovertibly manyy times over now, most visibly in the last Landmark finals.
Why club HR with centre owners ? We are respectable folks maan !
Congrats to all winners esp Niranjan whose sole contribution seems to have been collect aunty-dotes for Derek's poison pills.
That's IR denialism, that is.
Anand: It is not my argument that IR is the messiah; merely that, compared to systems like D&P, it evens out attempts (and not content or fortune or chocolates). And plus it is easy to execute (there could be complicated systems that aren't easy to run).
Your recent LM analysis only showed, IIRC, that there were v few qns per team and too many easy qns that wouldn't survive for too long in passing.
If you care to elaborate on why IR doesn't work, send me an article - will be happy to put it up here, so that the rest of us can tilt at the windmills :-)
@Anand - Which LM finals are you talking about? It's not perfect but it beats D&P for sure.
JR: Sure, lemme know which parameters to address specifically & I'll work on it.
Anon: Mumbai. Don't espouse D & P either.
Kunal:Nyet.
anand
Anand: if you can summarise your objections to IR and send it, will post. Anything you think is relevant.
Anand- Are you suggesting we use that complicated "sort in ascending order" system you came up with a few years back? Sure you'll end up with more brickbats than Messers Lewis and Duckworth :)
Atul
Post a Comment