Comment by WindyLakeReturn | Hacker News Reader

Comment by WindyLakeReturn | Hacker News Reader

WindyLakeReturn

I think it does, but I also think it sharpens it.

I have been helping one kid with a programming class and one thing I've noticed is that they often lack a certain approach to problem solving that I personally developed by playing video games. They also played video games but far fewer and definitely less of the harder problem solving ones. It is entirely possible I liked the games because I already had the talent and the games didn't build the talent, but until we get a definite study on the correlation and causation my bet is going to be on them sharpening the skill.

After a certain point it does begin to sap. This is the nature of all entertainment. A few good books can enlighten a mind but too many books can result in someone who reads but never does anything with the knowledge. Even one doing math for fun can end up spending all their time solving math puzzles that are created just to be math puzzles and which don't tie into any greater discoveries.

I think this is fine. First, the system is far too hard to control so I don't see any way to improve it that wouldn't wreck the system. Maybe minor suggestions to diversify entertainment a bit and to be sure to share your experiences with a like minded community, but these are small tweaks and not massive overhauls.

I also don't think we can predict when something will produce the best results. Even if 9 of every 10 kids who get heavily drawn into math problem solving end up just wasting time solving math problems that exist just for the sake of math problems, that last 1 might get into applying the problem solving to novel fields of math that expand the total knowledge of humanity. If one is too heavy handed in optimizing that they'll risk destroying the ability for that 1 who goes further. I doubt we can be sure that trade off is worth while.

> Even one doing math for fun can end up spending all their time solving math puzzles that are created just to be math puzzles and which don't tie into any greater discoveries.

Numberphile had a good take on 'recreational maths' with their video on the loops and strings in pi. Dr. Crawford has a good bit near the end (~12:30 mark) on how these techniques lead to bigger discoveries. The 'playing' around and the methods generated often lead to much deeper truths in math, and serve as very good practice for new mathematicians.

https://www.youtube.com/watch?v=W20aT14t8Pw

> I've noticed is that they often lack a certain approach to problem solving that I personally developed by playing video games

What does that consist of?

Ability to divide a problem in smaller problems recursively until you have problems that are more manageable while avoiding being overwhelmed by the overall problem. Anyone who has played a new 4x game blind likely has experienced a sort of information overload that can deter someone from learning the game at all. With experience, I learned to break down the new information and UI into parts that closely match existing experience, with verifying that the match does hold, and then slowly expand my knowledge of each area until I've figured out how to play the game. Often this involves playing against the easy AIs until I have the basics and then working my way up the chain until the level of optimization and understanding ceases to be fun.

The same applies to handling a programming problem. Larger problems can involve many smaller problems that are mostly unrelated, but students new to programming can be overwhelmed and mentally shut down because they can't solve the entire problem at once.

Once a student learns to break it down into much smaller problems, they can find the ones they know or can easily solve and separate it from the parts of the problem that requires googling around. If a teacher tells them to use a certain library, knowing how to break down the problem into smaller ones helps arrive at the problems that need the library and the ones that don't. Maybe you need to visualize the data, so knowing if it is possible generate and store the data while ignoring the entire visualization until you are ready, knowing how to handle visualizing sample data you create on the spot while ignoring how it should be created, and knowing how to merge these two partial solutions into a larger one can make a daunting problem much more manageable.

This is a skill you can learn many ways. Just doing sudoku puzzles can teach a methodology approach to problem solving of solving the bits you know and seeing how that opens up more options. Board games are another great tool that can teach similar approaches.

Even a game like Final Fantasy 12 where you can build basic programs your allies follow when fighting can help reinforce concepts like handling edge cases and the computer doing exactly what you specified. Do you keep wiping because you set heals to only happen <20% hp when the boss does 25% hp damage an attack?

In other words, if you obsess about efficiency, you might be chasing a local optima and making your system more fragile.

I tend to frame this in terms of the Multi-Armed Bandit Problem. It's a trade-off between exploration vs exploitation. It's often impossible to predict beforehand which strategy is optimal for a given environment. Which, I suspect, is why progressive temperments and conservative temperments coexist.

Yes, but even worse it will be a local optima of the efficiency function you've chosen and people often choose very poor efficiency functions that ignore second order and higher effects.

Reply on news.ycombinator.com