Considering that at least part of the correct solution was found within 24 hours, I think you’re right, Locke. It might affect accessibility, though—I know I would be sad if I logged on only to find that the discussion had closed already.
Having read through the speculation, I even found most of the chapter quite anticlimactic. Recognizing the correct predictions removed all the tension, since MOR’s tension relies so much on plotting.
That said, though, reading through the discussion gave me a harmless and very insightful lesson into how predictions work. I learned what makes a prediction probable versus plausible, in a way that not only allows me to understand it, but to think about how I would apply it to my life (I hadn’t really internalized that the percents of all possible outcomes have to add to a hundred, even though in hindsight that’s fairly obvious. I also learned about the betting-real-money threshold).
All in all, despite getting in the way of the chapter, it was a nice, closed-environment rationalist lesson. Thank you for prompting the discussion, Eliezer!
Considering that at least part of the correct solution was found within 24 hours,
Dude who came up with the blood debt answer did so in about fifteen minutes, actually. I was one of the ones (in #lesswrong anyhow) who suggested that Harry would destroy the Dementor for the shock-value.
Turns out EY had Harry both go over and under that prediction.
Agreed. I initially felt a lot of tension as to the answer, and it didn’t fade upon a day or two’s speculation, but I did not feel that tension when I read the chapter. I definitely think that a wait over major cliffhangers is indicated, but a long one (even 5 days) cannot sustain the tension.
Five days was perfect in my perspective. To be honest I thought the speculation had the potential to be very fun and mentally stimulating but the way we did it was completely wrong. What ended up happening was everyone proposed own theories left and right and in the end only a few people got some of the answer right, whereas if we collaborated better we could have ended up with an entire community who guessed most of the answer right. Makes for more overall happy.
After considering, I feel it got in the way because people got so much right. It made Harry’s dark side much less awesome.
Instead of being, as usual, impressed, I felt more like “why did he even need a mysterious Dark Side for this? And how did he not come up with it in all the time while trying-to-do-the-impossible before the trial?” Which is unfair, but it’s not clear fair unbiased thinking that decides for us whether or not we enjoy something...
I would say it was, but only because you managed to include elements of the speculation while still thinking of plots and turns that I did not see speculated. With the amount of speculation I participated in, it felt like an excellent emotional roller coaster, which I will try to describe with a few anecdotal sentences.
“Yes, that idea has been referenced!”
“Yes, that idea has been referenced as well! Multiple points, I knew it, I was expecting multiple points to come up, I should have posted that instead of remaining silent.”
“Oh, I should have seen that! In hindsight it feels like could have guessed that.”
“Wait, THAT wasn’t the true answer?”
“Wait, WHAT? I never would have guessed this!”
“He’s surprised, but it made perfect sense!”
“Heehee, he even referenced that as well!”
But it seems like it was only that good because you managed to narrowly outwit the amount of effort I had to put into it and the amount of collective thought I had taken the time to read. I feel like if I had been speculating to an overly large extent that I wouldn’t have been able to consider it as a story, but that if I hadn’t been speculating at all I wouldn’t have gotten the twists and turns it was taking.
So while I personally would vote for this option, I think I can see several reasons why it wouldn’t necessarily work for other people in the same way it did for me, assuming I’m right about my mental pictures of other readers.
It did make it better. But please don’t make people solve a puzzle to get a happy ending. The downside of getting only a sad ending if people fail at that is too high. Not just in terms of how many people will get negative utility from that, but also it will substantially reduce how many people will be willing to recommend the story to others (once it is finished). The potential downside to that is simply too large.
If that really does happen, I’m going to buy 3333 Eliezer Yudkowsky ems and put dust specks in their eyes. Again and again and again, for as long as the universe has usable energy.
(uh.… when I can buy Eliezer Yudkowsky ems, that is.)
EDIT: Apparently LW doesn’t like jokes about eternal torture. Oh well…
I have to say, I enjoyed the process of coming up with and justifying my ‘throw Dumbledore under the bus’ theory a lot more than I actually enjoyed the chapter, which wound up looking like a mish-mosh (a debt and messing with Dementors and Hermione joining the House of Potter and foreshadowing)...
I enjoyed the process of coming up with and justifying my ‘throw Dumbledore under the bus’ theory
That’s a very bad mental habit to get into. As Bryan Caplan explains here.
The key difference between a normal utilitarian and a Leninist: When a normal utilitarian concludes that mass murder would maximize social utility, he checks his work! He goes over his calculations with a fine-tooth comb, hoping to discover a way to implement beneficial policy changes without horrific atrocities. The Leninist, in contrast, reasons backwards from the atrocities that emotionally inspire him to the utilitarian argument that morally justifies his atrocities.
I don’t think I wanted to get rid of Dumbledore beforehand; but the solution dealt with all the desiderata in one single stroke, as opposed to the actual chapter which was an unsatisfying potpourri of solutions. Cute quote anyway.
No, as I pointed out, one would expect, based on its past performance, the dark side to come up with disastrous yet simple and effective solutions, and this expectation is another desiderata. Which that solution filled as well.
(A token ‘dispel the Patronuses and make the Dementors eat people’ is at least a gesture in the right direction, for all that I find Harry’s belief he can cripple Aurors like that to be risible—if you yell at a pilot ‘actually it doesn’t run on the Bernouilli effect but spiral vortices’ or whatever, does he immediately panic and fly into the ground? Of course not, years of training and experience make him do the right thing, just like the Aurors’ years/centuries of experience conjuring and maintaining Patronuses would override some weird statement by an odd kid.)
It sounds like a modern thing, such retirements usually have loopholes or exceptions, and Patronus casting would start either in Hogwarts or Auror training; all of which could push Aurors well over the century mark.
It was great! It also allowed me to test a couple of thesis of generating solutions. The closest I got was doing something completely different than working directly on the generating of solutions; I can’t remember the name of the theory stating that this should be the case, and while having it strengthened is somewhat disheartening it is nevertheless a useful piece of information.
Now if I weren’t so bad at shaving, I might even remember to use Occam’s razor next time and reduce “Harry marries Hermione” to “Harry makes Hermione part of house Potter”. Still much in the ways of the force have I to learn ;)
Only defence of my marriage theory was that I wasn’t quite sure that people could just be adopted into houses. Even thought it makes perfect sense, having seen it in action.
I suspect/expect you’ll write what power Lucius can legally claim over Harry sometimes soon?
Also I learned to actually look up the experiments you referred to...
All in all thank you for a highly entertaining and inspiring chapter!
I enjoyed every bit of the speculation but then finding out that some of the speculation was correct disappointed me. I would approve of a repeat if you made the puzzle sufficiently hard that nobody figures it out.
Strongly disagree. Puzzles that can’t be solved aren’t puzzles, they are authors being obnoxious. There’s no talent in making an unsolvable puzzle any more than there is in making a Zendo rule that no one can solve. And there’s no fun in it for most people either. We shouldn’t be in a situation where at the last minute we’re informed that no one but a dark wizard would put mustard on top of the sauerkraut (Standard TVTropes warning).
It was disappointing to me because it wasn’t the first time I’d heard the solution. It was like I had a spoiler for the chapter, because I was reasonably confident as to what it was. And while I’ve seen research linked to on LW that says spoilers don’t decrease enjoyment, I definitely find they do, at least for me.
It was redeeming, however, that more complications were added on top of the imperius!debt. If it had simply been Harry winning with it, I think would have found the chapter dull.
In retrospect you’re not supposed to think “well, how was I supposed to know about the sauerkraut” but “oh, that makes sense, I wish I’d thought of that.”
So the key is to make the puzzle only seem obvious in retrospect? This sounds like you want puzzles that actively trigger hindsight bias. Not exactly a promotion of rationality.
Not hindsight bias, just an asymmetrically easy verification. Imagine a large subset sum problem: answers can all be found logically, it’s very hard to find an answer, and it’s very easy to verify an answer. Any such problem can trigger hindsight bias of the form “that clearly would have been easy to solve; I just wasn’t trying”, but that’s a flaw of the biased person not the problem.
Well, not obvious in retrospect, that would be silly. I really don’t understand how you’re arguing with me about the fact that puzzles can be easy or hard without adding sauerkraut.
The issue isn’t that puzzles can be easy are hard. The issue is that a good hard puzzle is still solvable. It takes no talent to make a puzzle that no one solves. The difficulty in making a puzzle that’s worthwhile is making it in the narrow band of puzzles that are tough enough to be interesting but are still solvable.
Right, and what I’m saying is make the puzzle hard enough that nobody figures the whole things out, spoiling the chapter when I actually read it. It’s okay if people think of partial solutions, but when the whole chapter is basically posts A, B, and C glued together then it’s a disappointment.
If part can’t be figured out then it falls into that category, doesn’t it? I’m confused by what you are saying and wonder if there’s some set of terminological differences here. Perhaps we should carefully define our terms?
I don’t think this argument is worth it, it looks fairly silly in retrospect. Really, how to classify the puzzle isn’t the important part; what’s important is the outcome I consider favorable. That outcome involves plenty of discussion and theories (I have no complaints in that regard) and then a solution that is better than any of them.
And really, come to think of it, this chapter did deliver on this in some ways. If only because the events of the chapter consisted of several moves, while the theories only tried to predict Harry’s first move. So I feel more content now having thought things over than I was initially. And you are probably right that most ways to “fix” the problem would end up making things worse: most puzzles that people write are bad because they are too hard, not because they are easy.
Anyway, it strikes me that there are more interesting things to discuss, like wondering whether Mr. Hat-&-Cloak is actually Imperiused Flitwick polyjuiced into Lucius Malfoy. So hopefully there is not too much confusion left over on your end.
Not to worry, I’d expect that both endings will be written, and the game will just determine which one gets labelled “True Ending” in big, friendly letters.
In actuality, I think it made reading the next chapter slightly worse, but made the intervening time much better. And more importantly, I’m pretty sure I learned something about how to solve these sorts of puzzles properly, which is much more relevant.
I read the speculation, glommed onto the right answer when someone else brought it up, and then got amused by progressively more wacky theories for three days. I don’t think the speculation got in the way, per se, but it’s sort of anticlimactic for the answer you regard as obvious to be the correct one. The money bit was a nice twist, though.
Different, but neither noticeably better nor worse. In any case those who would rather not read the speculation can just stay out of the thread, or discussion and speculation could be separated from each other.
When more people are speculating, I am less likely to be the first one with a theory. That takes away from my enthusiasm.
80 was also really sad and caused me bad Fremdschämen. I was really glad the next chapter was so soon and tried not to think about the story much in the meantime. So maybe that was a bigger part of it.
Note that if you voted in the poll, you should also downvote this post. Currently, there are more upvotes in the poll than there are downvotes on that post.
Edit: Whoa, that changed in the time it took me to post!
Poll to see whether the speculation made the chapter reading experience better or worse.
Vote up if you think all the speculation got in the way of the chapter itself.
Five days was too long, IMO. If we only had 24 hours I would have enjoyed it much more.
Considering that at least part of the correct solution was found within 24 hours, I think you’re right, Locke. It might affect accessibility, though—I know I would be sad if I logged on only to find that the discussion had closed already.
Having read through the speculation, I even found most of the chapter quite anticlimactic. Recognizing the correct predictions removed all the tension, since MOR’s tension relies so much on plotting.
That said, though, reading through the discussion gave me a harmless and very insightful lesson into how predictions work. I learned what makes a prediction probable versus plausible, in a way that not only allows me to understand it, but to think about how I would apply it to my life (I hadn’t really internalized that the percents of all possible outcomes have to add to a hundred, even though in hindsight that’s fairly obvious. I also learned about the betting-real-money threshold).
All in all, despite getting in the way of the chapter, it was a nice, closed-environment rationalist lesson. Thank you for prompting the discussion, Eliezer!
Dude who came up with the blood debt answer did so in about fifteen minutes, actually. I was one of the ones (in #lesswrong anyhow) who suggested that Harry would destroy the Dementor for the shock-value.
Turns out EY had Harry both go over and under that prediction.
Agreed. I initially felt a lot of tension as to the answer, and it didn’t fade upon a day or two’s speculation, but I did not feel that tension when I read the chapter. I definitely think that a wait over major cliffhangers is indicated, but a long one (even 5 days) cannot sustain the tension.
Five days was perfect in my perspective. To be honest I thought the speculation had the potential to be very fun and mentally stimulating but the way we did it was completely wrong. What ended up happening was everyone proposed own theories left and right and in the end only a few people got some of the answer right, whereas if we collaborated better we could have ended up with an entire community who guessed most of the answer right. Makes for more overall happy.
After considering, I feel it got in the way because people got so much right. It made Harry’s dark side much less awesome.
Instead of being, as usual, impressed, I felt more like “why did he even need a mysterious Dark Side for this? And how did he not come up with it in all the time while trying-to-do-the-impossible before the trial?” Which is unfair, but it’s not clear fair unbiased thinking that decides for us whether or not we enjoy something...
Vote up if you think that the experience of reading the chapter was better for all the speculation.
I would say it was, but only because you managed to include elements of the speculation while still thinking of plots and turns that I did not see speculated. With the amount of speculation I participated in, it felt like an excellent emotional roller coaster, which I will try to describe with a few anecdotal sentences.
“Yes, that idea has been referenced!” “Yes, that idea has been referenced as well! Multiple points, I knew it, I was expecting multiple points to come up, I should have posted that instead of remaining silent.” “Oh, I should have seen that! In hindsight it feels like could have guessed that.” “Wait, THAT wasn’t the true answer?” “Wait, WHAT? I never would have guessed this!” “He’s surprised, but it made perfect sense!” “Heehee, he even referenced that as well!”
But it seems like it was only that good because you managed to narrowly outwit the amount of effort I had to put into it and the amount of collective thought I had taken the time to read. I feel like if I had been speculating to an overly large extent that I wouldn’t have been able to consider it as a story, but that if I hadn’t been speculating at all I wouldn’t have gotten the twists and turns it was taking.
So while I personally would vote for this option, I think I can see several reasons why it wouldn’t necessarily work for other people in the same way it did for me, assuming I’m right about my mental pictures of other readers.
It did make it better. But please don’t make people solve a puzzle to get a happy ending. The downside of getting only a sad ending if people fail at that is too high. Not just in terms of how many people will get negative utility from that, but also it will substantially reduce how many people will be willing to recommend the story to others (once it is finished). The potential downside to that is simply too large.
Nah. We can do this.
And if we don’t would you then want him to go through with it?
It’s not a real challenge, otherwise.
It is a challenge whether or not we get rewarded/punished for success/failure.
Emphasis on “real”. That was meant to evoke Ishtar’s way of thinking in Just another day in utopia.
Yes. We can earn our happy ending.
If that really does happen, I’m going to buy 3333 Eliezer Yudkowsky ems and put dust specks in their eyes. Again and again and again, for as long as the universe has usable energy.
(uh.… when I can buy Eliezer Yudkowsky ems, that is.)
EDIT: Apparently LW doesn’t like jokes about eternal torture. Oh well…
Where is the vote that “all the speculation was a better than the chapter itself”?
That’s no slight on the chapter, mind. The discussion was both entertaining and useful.
I have to say, I enjoyed the process of coming up with and justifying my ‘throw Dumbledore under the bus’ theory a lot more than I actually enjoyed the chapter, which wound up looking like a mish-mosh (a debt and messing with Dementors and Hermione joining the House of Potter and foreshadowing)...
That’s a very bad mental habit to get into. As Bryan Caplan explains here.
I don’t think I wanted to get rid of Dumbledore beforehand; but the solution dealt with all the desiderata in one single stroke, as opposed to the actual chapter which was an unsatisfying potpourri of solutions. Cute quote anyway.
Except for the whole “not throwing anyone under the bus” thing.
No, as I pointed out, one would expect, based on its past performance, the dark side to come up with disastrous yet simple and effective solutions, and this expectation is another desiderata. Which that solution filled as well.
(A token ‘dispel the Patronuses and make the Dementors eat people’ is at least a gesture in the right direction, for all that I find Harry’s belief he can cripple Aurors like that to be risible—if you yell at a pilot ‘actually it doesn’t run on the Bernouilli effect but spiral vortices’ or whatever, does he immediately panic and fly into the ground? Of course not, years of training and experience make him do the right thing, just like the Aurors’ years/centuries of experience conjuring and maintaining Patronuses would override some weird statement by an odd kid.)
I thought mandatory retirement was set at one century?
It sounds like a modern thing, such retirements usually have loopholes or exceptions, and Patronus casting would start either in Hogwarts or Auror training; all of which could push Aurors well over the century mark.
It was great! It also allowed me to test a couple of thesis of generating solutions. The closest I got was doing something completely different than working directly on the generating of solutions; I can’t remember the name of the theory stating that this should be the case, and while having it strengthened is somewhat disheartening it is nevertheless a useful piece of information.
Now if I weren’t so bad at shaving, I might even remember to use Occam’s razor next time and reduce “Harry marries Hermione” to “Harry makes Hermione part of house Potter”. Still much in the ways of the force have I to learn ;)
Only defence of my marriage theory was that I wasn’t quite sure that people could just be adopted into houses. Even thought it makes perfect sense, having seen it in action.
I suspect/expect you’ll write what power Lucius can legally claim over Harry sometimes soon?
Also I learned to actually look up the experiments you referred to...
All in all thank you for a highly entertaining and inspiring chapter!
I enjoyed every bit of the speculation but then finding out that some of the speculation was correct disappointed me. I would approve of a repeat if you made the puzzle sufficiently hard that nobody figures it out.
Strongly disagree. Puzzles that can’t be solved aren’t puzzles, they are authors being obnoxious. There’s no talent in making an unsolvable puzzle any more than there is in making a Zendo rule that no one can solve. And there’s no fun in it for most people either. We shouldn’t be in a situation where at the last minute we’re informed that no one but a dark wizard would put mustard on top of the sauerkraut (Standard TVTropes warning).
It was disappointing to me because it wasn’t the first time I’d heard the solution. It was like I had a spoiler for the chapter, because I was reasonably confident as to what it was. And while I’ve seen research linked to on LW that says spoilers don’t decrease enjoyment, I definitely find they do, at least for me.
It was redeeming, however, that more complications were added on top of the imperius!debt. If it had simply been Harry winning with it, I think would have found the chapter dull.
So it should be a completely fair puzzle that nobody solves. If Harry Potter can do the impossible, then why not.
How do you determine that a puzzle is completely fair and isn’t solved? Is that a meaningful category?
In retrospect you’re not supposed to think “well, how was I supposed to know about the sauerkraut” but “oh, that makes sense, I wish I’d thought of that.”
So the key is to make the puzzle only seem obvious in retrospect? This sounds like you want puzzles that actively trigger hindsight bias. Not exactly a promotion of rationality.
Not hindsight bias, just an asymmetrically easy verification. Imagine a large subset sum problem: answers can all be found logically, it’s very hard to find an answer, and it’s very easy to verify an answer. Any such problem can trigger hindsight bias of the form “that clearly would have been easy to solve; I just wasn’t trying”, but that’s a flaw of the biased person not the problem.
Well, not obvious in retrospect, that would be silly. I really don’t understand how you’re arguing with me about the fact that puzzles can be easy or hard without adding sauerkraut.
The issue isn’t that puzzles can be easy are hard. The issue is that a good hard puzzle is still solvable. It takes no talent to make a puzzle that no one solves. The difficulty in making a puzzle that’s worthwhile is making it in the narrow band of puzzles that are tough enough to be interesting but are still solvable.
Right, and what I’m saying is make the puzzle hard enough that nobody figures the whole things out, spoiling the chapter when I actually read it. It’s okay if people think of partial solutions, but when the whole chapter is basically posts A, B, and C glued together then it’s a disappointment.
If part can’t be figured out then it falls into that category, doesn’t it? I’m confused by what you are saying and wonder if there’s some set of terminological differences here. Perhaps we should carefully define our terms?
I don’t think this argument is worth it, it looks fairly silly in retrospect. Really, how to classify the puzzle isn’t the important part; what’s important is the outcome I consider favorable. That outcome involves plenty of discussion and theories (I have no complaints in that regard) and then a solution that is better than any of them.
And really, come to think of it, this chapter did deliver on this in some ways. If only because the events of the chapter consisted of several moves, while the theories only tried to predict Harry’s first move. So I feel more content now having thought things over than I was initially. And you are probably right that most ways to “fix” the problem would end up making things worse: most puzzles that people write are bad because they are too hard, not because they are easy.
Anyway, it strikes me that there are more interesting things to discuss, like wondering whether Mr. Hat-&-Cloak is actually Imperiused Flitwick polyjuiced into Lucius Malfoy. So hopefully there is not too much confusion left over on your end.
Next to the thumbs up and thumbs down karma buttons, should be placed a snapping finger icon.
What could it do that could possibly be of great enough import?
I hope you write the ending you want, rather than playing games to see which ending we will earn.
Not to worry, I’d expect that both endings will be written, and the game will just determine which one gets labelled “True Ending” in big, friendly letters.
That somehow doesn’t stop me from worrying.
In actuality, I think it made reading the next chapter slightly worse, but made the intervening time much better. And more importantly, I’m pretty sure I learned something about how to solve these sorts of puzzles properly, which is much more relevant.
I read the speculation, glommed onto the right answer when someone else brought it up, and then got amused by progressively more wacky theories for three days. I don’t think the speculation got in the way, per se, but it’s sort of anticlimactic for the answer you regard as obvious to be the correct one. The money bit was a nice twist, though.
Different, but neither noticeably better nor worse. In any case those who would rather not read the speculation can just stay out of the thread, or discussion and speculation could be separated from each other.
When more people are speculating, I am less likely to be the first one with a theory. That takes away from my enthusiasm.
80 was also really sad and caused me bad Fremdschämen. I was really glad the next chapter was so soon and tried not to think about the story much in the meantime. So maybe that was a bigger part of it.
Note that if you voted in the poll, you should also downvote this post. Currently, there are more upvotes in the poll than there are downvotes on that post.
Edit: Whoa, that changed in the time it took me to post!
Karma balance; vote this down.
Funny how karma never adds up in those polls.
There should be a house rule about always linking to the karma sink in the poll choices.
Not that it matters that EY or anyone else gets a few points of extra karma, it’s just my “defective mechanism detected” brain lobe giving me OCD.