An electronics engineer's take on the DU cutoffs
If you don't like signals and systems, you should probably leave now.
(PS : I'm writing this while under medication, so this might make sense to me only till the morn. Damn you fever!
Edit : I read this again when in a better condition. There are unexplored analyses corners. I shall leave this as an exercise to the reader to figure out :-P )
TL;DR : Exams are too easy. Tough to differentiate.
OK, I'm going to model the current examination and selection procedure very simplistically.
The student is an deterministic function who can reproduce the same level of throughput at any given time plus a uniformly distributed random noise. Deterministic part says that the guy knows the answer independent of the situation he is in (like answer to 2+2). The noise accounts for the time variation of the student's ability - he may be sick on a day and not remember anything, or he may take a double dose of red bull and mug the entire text book. (or he may get access to a text book magically in the exam :P cc: +Vikram Kamath )
Next i'll model the exam as an amplifier of this student's output. The gain is inversely proportional to the difficulty of the exam. Lesser the difficulty, more the marks. (I'll get back to this). Then there is the grading. Grading will put your scores exactly in the range 0 - 100.
When you have cutoff of 100%, it can mean a couple of things. The students in the country are geniuses with no par, or the exams are predictable and easy. I will assume it is the latter for the next part of the analysis.
Exam is easy => Gain A is high
Percentage = max ( 100 , A*[f(x) + noise])
Next your selection will be like a DAC measuring this. The resolution of the DAC is dependent on the number of seats available. (which is very less). So, you have large number of students in 99+/- range vying for few seats.
The problem here is that because of A*noise > resolution of DAC , the selection seems like a clouded process. A lot of students will feel "what the... how did that guy get a seat and i didn't"
How would I improve the situation? Reduce A => tougher exams, and change the reference voltage for the DAC ( mid-code value)
Since there are no other knobs. (resolution is DAC can be improved, but this requires infrastructure development and that is a long term thing. Plus d(population)/dt > d(num_of_seats)/dt ).
Now you have lesser aberration, and lesser disappointment. Much like the old JEE-mains type exams. You know you are (not) good enough after answering that. ( NITK people can relate to Herr IR Rao's paper. If you score a mark, then you deserve it)
As always the guys just missing out will be the most disappointed, not the ones at a much lower end of the spectrum. They will just have to deal with it, but they shouldn't blame things like "damn I marked B instead of C even though I calculated it as C".
If you've read this much without cursing me, you deserve a chocolate. I shall give it to you the next time I meet you. (Do tell when you want to meet, so that I can actively avoid you).
(PS : I'm writing this while under medication, so this might make sense to me only till the morn. Damn you fever!
Edit : I read this again when in a better condition. There are unexplored analyses corners. I shall leave this as an exercise to the reader to figure out :-P )
TL;DR : Exams are too easy. Tough to differentiate.
OK, I'm going to model the current examination and selection procedure very simplistically.
Next i'll model the exam as an amplifier of this student's output. The gain is inversely proportional to the difficulty of the exam. Lesser the difficulty, more the marks. (I'll get back to this). Then there is the grading. Grading will put your scores exactly in the range 0 - 100.
When you have cutoff of 100%, it can mean a couple of things. The students in the country are geniuses with no par, or the exams are predictable and easy. I will assume it is the latter for the next part of the analysis.
Exam is easy => Gain A is high
Percentage = max ( 100 , A*[f(x) + noise])
Next your selection will be like a DAC measuring this. The resolution of the DAC is dependent on the number of seats available. (which is very less). So, you have large number of students in 99+/- range vying for few seats.
The problem here is that because of A*noise > resolution of DAC , the selection seems like a clouded process. A lot of students will feel "what the... how did that guy get a seat and i didn't"
How would I improve the situation? Reduce A => tougher exams, and change the reference voltage for the DAC ( mid-code value)
Since there are no other knobs. (resolution is DAC can be improved, but this requires infrastructure development and that is a long term thing. Plus d(population)/dt > d(num_of_seats)/dt ).
Now you have lesser aberration, and lesser disappointment. Much like the old JEE-mains type exams. You know you are (not) good enough after answering that. ( NITK people can relate to Herr IR Rao's paper. If you score a mark, then you deserve it)
As always the guys just missing out will be the most disappointed, not the ones at a much lower end of the spectrum. They will just have to deal with it, but they shouldn't blame things like "damn I marked B instead of C even though I calculated it as C".
If you've read this much without cursing me, you deserve a chocolate. I shall give it to you the next time I meet you. (Do tell when you want to meet, so that I can actively avoid you).
Comments
Post a Comment