About Rules
An interesting excerpt from "... Essays on Graduate Education" by David Merrill:
"... most rules were created because some problem occurred that required a difficult decision. To avoid the problem in the future a regulation is created. However, the same problem never seems to occur again, but the rule is then applied to a number of other situations that were not problems previously but because of the rule they are now problems."
Rules, nevertheless, cannot be done away with, so it is important to frame them right. Most rules are broad and hard constraints, and that is what hurts. For example, a minimum of 60% marks or 2 years experience as the eligibility criteria has many loopholes. Among others,
- 60% at universities A and B may not be comparable
- 60.1 in a certain combination of courses could be worse than 59.9 in some other combination
- Even across two years, the value of 60% could be vastly different. Most universities do not have any concept of a standardised score that is even roughly comparable across years
Similar objections can be raised against the time criterion.
Relative to the importance they place on depth and rigour in academic matters, even the topmost institutes are very careless when it comes to rules and regulations. They use bulldozers where small hammers are required, losing a lot of good stuff in the debris.
Rulebooks are typically a sequence of "if (condition) eliminate/pick" rules. Careful consideration would lead to much more nesting in the rules, but the system would become complicated and brittle. This sounds familiar. Probabilistic systems vs. Rule-based systems in AI?
So what we really need is a system where no criterion eliminates or picks absolutely -- it only diminishes or boosts the probability of being chosen.
There are difficulties with this system also: a) where many people from varied backgrounds are involved, it is too much work to find reasonable parameters and weights, and b) the choice of parameters and weights would still be subjective and some people (fewer hopefully) would still have reason to complain.
But it is when even small organizations/institutes/departments fall into the trap of 'broad and hard' rules that the situation becomes so ridiculous that we have to call it bureaucracy!
"... most rules were created because some problem occurred that required a difficult decision. To avoid the problem in the future a regulation is created. However, the same problem never seems to occur again, but the rule is then applied to a number of other situations that were not problems previously but because of the rule they are now problems."
Rules, nevertheless, cannot be done away with, so it is important to frame them right. Most rules are broad and hard constraints, and that is what hurts. For example, a minimum of 60% marks or 2 years experience as the eligibility criteria has many loopholes. Among others,
- 60% at universities A and B may not be comparable
- 60.1 in a certain combination of courses could be worse than 59.9 in some other combination
- Even across two years, the value of 60% could be vastly different. Most universities do not have any concept of a standardised score that is even roughly comparable across years
Similar objections can be raised against the time criterion.
Relative to the importance they place on depth and rigour in academic matters, even the topmost institutes are very careless when it comes to rules and regulations. They use bulldozers where small hammers are required, losing a lot of good stuff in the debris.
Rulebooks are typically a sequence of "if (condition) eliminate/pick" rules. Careful consideration would lead to much more nesting in the rules, but the system would become complicated and brittle. This sounds familiar. Probabilistic systems vs. Rule-based systems in AI?
So what we really need is a system where no criterion eliminates or picks absolutely -- it only diminishes or boosts the probability of being chosen.
There are difficulties with this system also: a) where many people from varied backgrounds are involved, it is too much work to find reasonable parameters and weights, and b) the choice of parameters and weights would still be subjective and some people (fewer hopefully) would still have reason to complain.
But it is when even small organizations/institutes/departments fall into the trap of 'broad and hard' rules that the situation becomes so ridiculous that we have to call it bureaucracy!