Skip to content

Latest commit

 

History

History
45 lines (31 loc) · 3.7 KB

45._Sharing_knowledge_and_resources.md

File metadata and controls

45 lines (31 loc) · 3.7 KB
Raw

45. Sharing knowledge and resources

Created Saturday 08 May 2021

Context

I find it very exhilarating to honestly share my knowledge and strategies. This is sometimes seen as naivety, but I'll try to demonstrate that this is not the case. In fact, we should share as much as possible, because it makes life fun.

Problem breakdown

Consider the scenario of a standardized test, that is highly competitive. Only a very small fraction of the people get selected. Let the materials be the data D. Let the models(strategies) be M. There are 4 scenarios for here, that model student-student relations:

  1. Show both(my wish) - everyone is eased, energy, resources, behavior are all well regulated. "Failure" depends on making the model better. Everyone learns a lot, enjoys it and retains it. Fearless exploration is the norm here. The best part: failure is unavoidable(limited seats is a fact), but the person knows exactly why they failed, it can be used further in life. There are no regrets. If there was a doubt about data or model not being shared, it'll lead to heartache, helplessness, compulsion, loss of ambition, positivity which is dangerous in general and affects other parts of life like children/love-life/empathy in general.

But, doesn't your individual chance of passing get affected? Let's compare P(win/sharing) and P(win/~sharing). If we can prove these are independent, we can do whatever we like. P(win/sharing) * P(win/~sharing) = TODO: see this

  1. Hide D, Show M - data itself is hidden. So the people not accessing it have no way to test models. This is pretty obvious for an IGUS and will result in data being collected by the 'victims'. They are likely to share it:
    1. They are an older generation
    2. They want more data, so reciprocation is key. And data acquistion is an expensive task, and data sources are not static, they may(😂️ may? obsoletion is at an all time high) change with the environment and there's a pressing need for regular updates.
    3. Empathy. **Anyways, data cannot be hidden, especially with the Internet, and data is static and replicable. **Hiding data is a total waste of time, only a fool will use such a 'strategy'.
  2. Show D, Hide M - this is the most likely to happen and has a significant chance of success. But it has flaws:
    1. No feedback for bettering the model.
    2. Speed of model accuracy is slow: data is related, and we're actually solving the same problems again and again, without realizing it soon. Redundant work.
    3. Self-confidence and doubt: because the model is doubtful and slow, the whole endeavor is endangered. Work feels like a chore. Chance of quitting can increase. Because there's too much blind confidence involved, and you can't fool yourself.
    4. Fear: because everyone is making their own model, there's absolutely no way to know who's the best. This can be a cause of great discomfort, every test is a 'doomsday' scenario. There's lot of perception filtering involved, it's tiring.
    5. Future use: you'll not be confident to reuse the knowledge for other things in life: you never tested it for generality.

Of course, this has the probability to work very very amazingly, but it's small and sporadic, nobody would want to bet on this. And of course, people try to lie and feed wrong facts.

  1. Hide both - This has the problems of both 2 and 3, so it's worse than the two. It is both foolish and too risky.

Related

Heck, this is getting better(or worse). We'll simulate to see, anyway. http://eugeniacheng.com/wp-content/uploads/2017/02/cheng-morality.pdf

Conclusion

There's no choice but to share data and models as fast/quick as possible. That'll increase chance of success for all and lessen heartache, plus its reusable in other parts of life. Maximum returns.