Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[jotheonah]

It doesn't matter how much fat I use when cooking or how much sugar I eat, my body refuses to put on even a little bit of weight

Switching to whole milk is something doctors often prescribe for underweight kids.
Adding more carbs to your meals might help too.

[silverspawn]

What no my weight is great. It's just that I eat extremely recklessly and won't become overweight

I assume it's great anyway. Don't actually know how much I should weigh given the height.

[silverspawn]

I probably would improve my chess rating by offering more draws. I tried that right now in a position that felt quite bad for me and op accepted. Engine confirms significant advantage, though we were both at ~90 seconds and I had a bit more time and the position was difficult, so who knows.

[silverspawn]

Intelligence is notoriously hard to improve. I feel like one component of it is something like the speed at which you parse concepts, and that's the component that feels completely just straight up impossible to improve. I was thinking about this when I read This exchange with Eliezer Yudkowsky which includes sentences like

Even in that paradigm, except insofar as you expect gradient descent to work very differently from gene-search optimization - which, admittedly, it does - when you optimize really hard on a thing, you get contextual correlates to it, not the thing you optimized on.

I can understand this sentence, and if I wanted to, I could certainly write sentences that are this complex. But it's not how I naturally write because it's too complex. It takes me a few seconds to parse this. And I know he wasn't trying to show off, so I imagine he just has a brain that can parse this at the first try.

The bright side is that I now think this raw processing speed is way less important than I used to think. Holding everything else constant, being better at it will probably make you more efficient at everything, but there's so many other components to being effective at things.

[silverspawn]

While processing speed/short term memory isn't changeable, I believe that something like general problem solving capability can be improved, and I've actually thought a lot about this, but it's not at the point where I can articulate it.

Although I was wondering if thinking really deeply about Zendo questions may be one of the better ways to do it. Those feel like an unusually pure case of hypothesis formation and testing.

[silverspawn]

And like whenever I think about Zendo questions I start thinking about how I think about Zendo questions and always feel like my approach is inefficient, but I've never invested the time to really go into depth about it. Instead my thoughts usually get interrupted by the Zendo question at hand. On net it probably just made me worse at it. I should just commit to thinking about the meta problem for a while.

[MiX]

And like whenever I think about Zendo questions I start thinking about how I think about Zendo questions and always feel like my approach is inefficient, but I've never invested the time to really go into depth about it. Instead my thoughts usually get interrupted by the Zendo question at hand. On net it probably just made me worse at it. I should just commit to thinking about the meta problem for a while.

Have you played competitive zendo? That's much easier to think about better questions, since there's an actual problem you have to solve in terms of what questions to ask.

[silverspawn]

How is that different from the game here?

[infangthief]

I feel like Zendo Questions should be about style, rather than just efficiency.

I like it when the questions are actually questions; you have a hypothesis but to test it you have the extra challenge of coming up with appropriate questions, not just combinations of letters. Sometimes we may have to resort to nonsensical letter combinations if it seems to be the only way to make progress, but I don't like that being the norm.

[MiX]

How is that different from the game here?

At least how I learned it is:

The players have a turn order. On your turn, you can make a single structure (in this case, a sentence), and then say "Master" or "Mondo". If they say Master, the master just says if that sentence passes the rule or not. If they say Mondo, then every player secretly guesses if that sentence passes or not, and then the master reveals it. Whoever guessed correctly gets a guessing stone. After this, the player on that turn can spend a guessing stone to make a guess. Then it's onto the next player.

In this way, you want to control the information gained so that the rule is solved on your turn, but you also want to make questions that you think you know the answer to but others don't, to get morr guessing stones.

Also I feel like I explained myself horribly, but oh well.

[silverspawn]

I feel like Zendo Questions should be about style, rather than just efficiency.

I like it when the questions are actually questions; you have a hypothesis but to test it you have the extra challenge of coming up with appropriate questions, not just combinations of letters. Sometimes we may have to resort to nonsensical letter combinations if it seems to be the only way to make progress, but I don't like that being the norm.

I agree! In fact I've suggested disallowing letter combinations before, but it wasn't picked up. I would still endorse that.

Would you play if i started a thread about Zendo questions where you're only allowed real words?

[silverspawn]

How is that different from the game here?

At least how I learned it is:

The players have a turn order. On your turn, you can make a single structure (in this case, a sentence), and then say "Master" or "Mondo". If they say Master, the master just says if that sentence passes the rule or not. If they say Mondo, then every player secretly guesses if that sentence passes or not, and then the master reveals it. Whoever guessed correctly gets a guessing stone. After this, the player on that turn can spend a guessing stone to make a guess. Then it's onto the next player.

In this way, you want to control the information gained so that the rule is solved on your turn, but you also want to make questions that you think you know the answer to but others don't, to get morr guessing stones.

Also I feel like I explained myself horribly, but oh well.

While this sounds both cool and fun, I think I would find it less interesting on the particular axis of being a proxy for hypothesis testing/problem solving

[silverspawn]

Guess: a sentence satisfies the rule iff one word in it has at least 5 more letters than another word

Be honest, to what extent was that solved by the latest set of questions?

[silverspawn]

ferries fairly ferry fairies; fairies fere ferried fairly

[MiX]

Guess: a sentence satisfies the rule iff one word in it has at least 5 more letters than another word

Be honest, to what extent was that solved by the latest set of questions?

That was probably 90% of it. Or more.

[silverspawn]

which reminds me

I hereby claim to have mostly solved the game Worlords Battlecry II, also known as the least balanced game ever made. Out of the 12 races, the strongest ones are either Fey or Wood Elf. The best quick way to realize this is to ask whether any race could conceivable have any chance to beat a fey army. The answer is probably no for everyone except the three Elfs and Fey itself, but Wood Elf is strictly and vastly better than the two other Elfs so they're not in the race.

The proper way to play Wood Elf is to mass produce wisps for the unbounded exponential resource gain and then produce forest spirits to fight and upgrade them. All other units are irrelevant except to put into your starting towers, but towers probably only matter against the computer.

The proper way to play fey is a bit more complex because most units can fight rather than just one. Produce (1) fairy dragons (2) lightning hawks and (3) pixies or forest spirits in a manner that balances resources. Buy the good upgrades. There is no exponential gain so Fey are only in the race insofar as they can win early, which they might well be able to on some maps.

[silverspawn]

Also I've solved Stronghold I.

The proper way to play (at least assuming you have some time) is to set taxes to low, switch between normal and extra rations, ignore any other intervention for popularity and increase your population. How far? Farther. Still? Yes. More people. Still more, really? Yes. STILL more? Did you cover the entire map with food production to the point that feeding more people is impossible because at last  travel distance sets a hard limit and more people won't feed even themselves if they work on more food production? Then you can stop. If not, increase your population.

The basic design flaw is that raising taxes has linear cost but far sublinear returns, so there is no reason to raise taxes above low ever. On the other hand, increasing your population has linear gains until you run into the hard limits of space and travel distance, and also linear cost which is always much lower than the benefit, so there is no reason to ever stop growing your population until travel distance begins to matter.

What units you buy is comparatively irrelevant. Walls and everything related is useless if the game is played against a human. If you play against the computer, you need them for the first attack but after that your economy should be so massive that you can outproduce them on the hardest difficulty. If played properly, anything after the first attack is a total joke.

[silverspawn]

I've sort of made it a mission to optimize all of the games I played when I was young. They all break upon enough optimization pressure, except of course starcraft I. They always break in a way that I'm sure the designers didn't intend.

[silverspawn]

Also in stronghold, the AI of units attacking your base is "search for way to the keep; if it exists go there; if not attack walls; check every couple of seconds to re-decide". This is exploitable by building a single very long winded way as the only path to your keep. Opponent will go through no matter what, which takes far more time than to kill walls. (Stops working once they have catapults which changes the AI.) But don't make it too long or the game crashes at the point where the AI searches for the way in. Also this exploit is unnecessary if you follow the guide above,

[silverspawn]

Just booked Microsoft Word which will cost me 99 euros if I forget to cancel after a month. (Reminder to cancel within a month.)

[silverspawn]

This is better than I remembered it:

[silverspawn]

Guess: a sentence satisfies the rule iff one word in it has at least 5 more letters than another word

Be honest, to what extent was that solved by the latest set of questions?

That was probably 90% of it. Or more.

To make the implicit explicit, I was a little annoyed by this, not because you used that to guess correctly but because I think I expected an acknowledgement that it was because of my post. But idk if this is at all reasonable, also definitely possible that I've done something analogous before without even realizing it and the other person was either lucky/mindful enough not to be annoyed or too shy/polite to say it.

I also made some mistake somewhere because I thought some of the judgments implied that not all letters are equal, which ended up being the case. A first lesson would be to write down hypotheses, which already generalizes to other domains!

[silverspawn]

(I.e., if I had written down -- or even explicitly stated in my mind -- "not all letters are equal because judgings X,Y, the process of doing this would very likely have revealed that the judgments don't in fact imply this.)

[silverspawn]

I was correct about "the rule starts with an existence quantor" and "the rule requires at least two words"

[silverspawn]

Relevantly, reading through the thread and found this:

Guess: the word needs to contain all letters of at least one number (spelled out)

That was going to be my guess!

I, too, felt robbed last round

yuck! This was me being annoyed but too much of a wuss to say it and instead letting it out in a passive-agressive way. I strongly disapprove of this behavior.