Eliezer_Yudkowsky comments on Harry Potter and the Methods of Rationality discussion thread, part 14, chapter 82 - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (790)
When the book is done I'm going for the former, but the rules call for the book to be complete, I believe.
Why would you go for the former - what's the reasoning here? Yes, it's gotten praise from some writers, but not the kind of rapturous praise that would give it the most important prestigious prize they have. MoR is amazing for a fan work... and good for a novel. Is this some sort of satisficing reasoning, where it's better to be nominated for best novel than win best fan work?
I know I'm not an average voter, but HPMOR is literally the best book I have ever heard of. Are there some other books I have missed out on?
Try looking through the Hugo or Nebula best novel awards. Most of them are not didactic like MoR is which makes comparing MoR to them a little unfair since you lose out on the 'I want to base my life on this' effect, and MoR is length-wise at... a trilogy? now, so comparing them to MoR is unfair, and it definitely helps to have read multiple times the heavy influences on MoR like Godel, Escher, Bach or Ender's Game, but I am doing it anyway! Looking through the list, here are ones I've read (~1/3) and would rank as either not much inferior, equal or better than MoR:
And this is far from a complete list; for example, Blindsight is fantastic but was only nominated for a Hugo in 2006 (losing to Spin which I have not read).
I liked Spin a lot-- it's a perfect balance between a mainstreamish novel of character and big ideas science fiction. Gur rnegu vf chg va na nyvra rairybcr juvpu znxrf gvzr cnff zhpu snfgre-- nyy bs gur fhqqra, gur fha orpbzvat n erq tvnag vf n frevbhf ceboyrz.
It's gotten sufficiently rapturous praise. I see no reason not to try.
I found citations for ESR and David Brin (only the latter is a SF writer), who praised it but did not receive it rapturously; are there writers I have missed?
Rachel Aaron (author of the Eli Monpress series, sample of "The Legend of Eli Monpress" currently on my Kindle being read) is I think the most effusive praise I've gotten from a published author, though I've heard rumors of certain others speaking in the halls of SF conventions. But you're probably thinking of the Nebula, which is voted on by members of the SFWA. The Hugo is a reader award, voted on by the attendees of Worldcon. Look at the review page of HPMOR if you're questioning whether the readers have been sufficiently enthusiastic.
I'm honestly a bit nonplussed at the idea that reader reception of HPMOR has been insufficiently enthusiastic to try for a Hugo. It's fairly routine for a review to say that HPMOR is the best thing they've ever read out of all of fiction. If that is insufficient enthusiasm to think, "Hm, I might as well try for Best Novel, or the Gryffindors will look at me funny for my sheer lack of courage", I don't know what level ought to be required.
OK, she's apparently not self-published, so I've added two interviews with her to the article as references. EDIT: And would you believe it, I forgot Hanson is a Notable person and so his recommendation on Overcoming Bias also counts.
I thought it was the other way around, but OK. That will help, but one wonders how much - the awards overlap a fair bit. (A quick count suggests 1/3 winners the same over the past 15ish years.)
The question is which Hugo.
Gee, I wonder how that could possibly be less-impressive-than-it-looks evidence...
Is the prior that low or something? What should the world look like if HPMOR does have a chance of winning Best Novel?
Yes. Can you name a single fanfic that has ever won either prize for best novel? I can't. Keeping in mind that fanfics long predate the Internet, that many noted authors got started as fans or were published in fanzines (Charles Stross being the most recent example I can think of), etc. So Laplace's law gives us a starting point of single percents, and it only gets worse if we take into account nominations as well...
Well, see above. Acceptable substitutes would include reviews and praise by the sort of people whose reviews & praise predict other nominees or winners - Brin is a good start, but we're talking novels that are discussed or reviewed in all the major SF organs and sell into the hundreds of thousands or millions, so the n should be into the dozens for starters.
My guess is there'd need to be more respect for fanfic among the older sf fans, but I could be wrong about that. I'm not the only person for whom HPMOR is the only fanfic I read. (I've read some other good fanfic, but I don't get around to fanfic generally speaking.)
I think a sufficiently high proportion of likely voters are on line, but this is based on a feeling of plausibility rather than actual knowledge.
When you talk about which award you're going for, is this just a matter of which award you encourage people to nominate it for?
In terms of prestige, I think you'd be much better off being nominated for Best Novel and losing than winning the Best Fan Writer award.
I find it hard to imagine you winning, but I also find it hard to imagine specific reasons that would make it extremely unlikely. It's quite possible that I'm just engaging in availability bias (all the other winners have been conventionally published) rather than anything more solid.
Ranked preference voting, though. I'd expect a significant numbers of voters to rank "no award" ahead of any fanfic just on general principles. If it was just a single round of single preference voting the odds would look much better.
Not only the attendees. People with supporting memberships can vote as well.
I am curious about your concern.
Do you want to save EY from a hubris born smackdown?
Do you want to keep public attention off fanfiction so you aren't tempted to publicly defend it and publicly identify as a fanfic fan?
Do you fear a loss of face for the Singularity Institute?
Forgive me for visiting your intentions, if that is unwelcome.
You only get one shot at the awards. My best calibrated guess is that MoR has a chance at best novel somewhere between 'not a chance in hell' and 5%, while best fanwork moves the odds of victory to 50+%. Which one do you think is better?
(I'm also a little concerned that EY is casually making what looks to me like an incredibly obvious mistake.)
I expect you meant that as a rhetorical question, but I'm not sure it is. I generally agree with your confidence estimates, but of course it's also worth looking at the payoffs. It's also worth comparing the payoffs for being a Hugo nominee for best novel and a Hugo award winner for best fanwork.
I don't have an informed opinion about those payoffs, just saying that it's not as simple as "50+% chance of winning an award" vs "<5% chance of winning an award."
Which is precisely why I am asking these questions, because there are many ways Eliezer could conclude it's a good idea:
he has non-public information
the rules favor MoR in some way I am unaware of
etc.
My theory is that Eliezer is overestimating his chances of winning best novel.
5) is not a believable option, since the legalities of fanfic prohibit any conventional sort of publication, and just slapping "Book 1"/"Book 2"/etc. on top of chapter headings does very little.
The rest is good analysis, though.
No, they don't. They just mean it takes a publisher with a little guts, willing to defend it under fair use grounds (in MoR's case, parody).
All you need to vote is a supporting membership, cost $60 or so. You don't have to attend.
As soon as HPMOR is finished (hopefully not soon), I will buy a supporting membership to the next year's worldcon. On that note, let me urge Eliezer to finish HPMOR in the summer of some year, so enough supporting memberships can nominate it by January 1.
I'm not sure that materially increases the number of votes one could expect. Gee, only $60...
I also think this is a good idea, and hereby vow to buy a membership when HPMoR is finished for this purpose of voting it for Best Novel. As pointed out, even being nominated would get it a lot more attention.
I'm hoping for something like Neil Gaiman had when he won and then they banned comics/graphic novels afterwards.
I think the main thing you're missing is that nothing bad happens to me if I don't win. This could serve as a mantra for a whole lot of things in life that are worth trying.
'Nothing bad happens to' two-boxers either. Do I really need to explain that loss of a gain is as bad as a gain of a loss?
Why, so you did. Careless reading of me... my apologies.
You're assuming that Best Fan Writer and Best Novel are equally valuable awards. This is not even close to true.
There are people who make a practice of reading all the Best Novel nominees, for years after the award is given, and I think people are much more likely to read Best Novel nominees before voting than to get around to the fan writing awards.
If anyone can track down number of votes in the different categories, it would be a way of checking on my opinion.
I have twice already pointed out that possibility.
It would be good information, yes. As far as I can tell, there may be some very recent records of recommendations by the Nebula members (and also claims of considerable politicking, so an appeal by Eliezer to fans would be far from unprecedented nor especially effective).
The Hugo voting system is pretty complicated, but apparently they've released detailed statistics since the 1980s - one of the first quotes in that 1997 analysis is
The most recent data is for 2011. Best Novel took 1813 ballots, with 779 vs 753 for #1 vs #2 (as I said, complicated - it's IRV).
The fan categories are fanzine, fan artist, and fan writer; the last had 323 ballots, with 70 vs 40. So in other words, just to get within spitting distance of Best Novel #2, you would need almost twice as many ballots as were cast for the entire category of Best Fan Writer.
You see why I would give MoR for Best Novel 0-5%, and EY for Best Fan Writer 50+%?
I can't find your mentions that the Best Novel Hugo might be worth more than the Best Fan Writer. This doesn't mean I think you're lying or mistaken, but where are they? Were they in some other thread or sub-thread?
I'm not sure if my point that just getting nominated for Best Novel is a huge win in terms of readership has registered. For these purposes, there is no difference I can see between coming in second and coming in last.
I agree that HPMOR doesn't have a great chance of winning Best Novel. It depends tremendously on what the competition is, and I'm assuming we're talking about at least two years from now, so what novels it would be up against are hard to predict. I don't think there are any predictable blockbusters in the immediate pipeline.
Subthread.
As I suggested, it may be a good idea - but Eliezer did not present that as a reason, has not presented it as a reason, and has not justified it. What makes you think the nomination is better than winning a fan Hugo, besides anecdotes about peoples' reading lists including nominees?
Wait, 800 votes is sufficient to win Best Novel? I think I'm with Eliezer on this now. That may actually be attainable with this fan base(if only because it's vastly easier to mobilize for online works than for paper ones, due to logistical issues)
Edit: I was unclear on the voting process, and am retracting the above.
What kind of expectations about the voting for a Hugo did you have that when told it's 800 (rather than 8, 80, 8000, 80000...) strikes you as fantastic? And do you think that there are 800 dedicated MoR readers who would go to Worldcon and rank it #1 in the vote? Remember the infamous 1% rule: for every reader, there's 1% who will leave a comment or review, and out of those, 1% will do something additional. Going to Worldcon and voting is quite something additional.
(Well, one way to answer this would be to put up a poll on hpmor.com asking how many readers attended Worldcon 2011, Worldcon 2012, etc. Liars will inflate the numbers, but it would help get an upper bound.)
Keep in mind that "this fan base" would probably vote Methods (or EY, I suppose) in for Best Fan Writer pretty much uniformly, but is likely to be fragmented for Best Novel.
I'm curious as to what would have been your original probability estimate for "Given that Eliezer writes a Harry Potter fanfiction, it becomes the most popular Harry Potter fanfiction on the Internet." Or, for that matter, getting out of the AI Box. Not everything impossible that I try, works - nobody coughed up $1.6 million to get faster HPMOR updates this time around, which I tried because, hey, why not - but to me, not trying for Best Novel, given feedback so far, just seems horribly, horribly non-Gryffindor. To me it seems like you're the one making this obvious, horrible mistake whose reference class of timidity errors could put a shadow over someone's entire life. Don't ask her out, don't interview at the hedge fund, don't try for the scholarship, go for Best Fan Work instead of Best Novel...
Offer 20-to-1 odds against HPMOR winning Best Novel and I'll buy in. Hm, now I'm curious as to which side of the bet Zvi or Kevin would take.
I would have assumed that the proper reference class was all your other fiction, which as much as I enjoyed them, were all short compared to MoR even the popular ones like 'Three Worlds Collide'. (One of my favorites, the Haruhi fanfic, was, what, 2 pages on FF.net?) Short fanfics cannot become the most popular, so I would have assigned it a very low probability. Had you asked an estimate for 'wrote a full trilogy of HP fanfic novels', my estimate would be quite different. I won't pretend to know what it would have been. (I remain surprised and amused that it has become as large and popular as it has been, and that you now procrastinate on both your rationality book(s) and the Center by writing MoR. Truly, fortune passes everywhere!)
Oh, I would've bet for you back on SL4. You already had succeeded in getting SIAI running, after all. (Although here again the counterfactual gets hard - I barely remember what I was like back in 2003-4 when I joined SL4 and IIRC you had already won an AI Box by then.)
Sure. I am poorer than a church mouse, so I can risk no more than ~$100. How's this:
"For any Worldcon 2013-2017, MoR will win the Hugo 'Best Novel' award. The stakes will be $100 against $5 (non-inflation adjusted). Payment by either Paypal or Bitcoin (at that day's exchange rate on the largest exchange eg. Mt.gox) to the winner or a charity of the winner's choice. In case of any dispute, the verdict will be judged by Carl Shulman or another person mutually acceptable to Eliezer Yudkowsky and gwern."
I've pinged them.
Eliezer has conceded the bet & paid me $5.
Okay, it's on.
Can anybody get in on this bet? I would take it as well.
I think the odds of HPMOR winning the Hugo (edited to say: for Best Novel) are far less than one percent. This has nothing to do with my estimation of the literary merits of the fic, and everything to do with my understanding of the literary establishment and its unwillingness to tangle with copyright law.
I've added it to the registry & PredictionBook.
So 2+ years on, has your opinion changed any? I think the probability has dropped a fair bit*, but I'm trying to simplify my finances, so I'm going to make you an offer: I'll sell you my side of the bet for 20% off, or $4. (Offer good for the next month or so.)
* to expand: I think the decrease in post tempo has destroyed much of the virality & reduced fan ardour, and more specifically, I haven't seen any prominent endorsements of MoR by elites like David Brin over the past year or so, which are necessary to legitimize MoR and make it a contender. You can't win a Hugo just by appealing to some people online, you also have to win over the niche of voters at the convention.
I'd take the Eliezer side of the bet if I was confident that HPMOR could compete against an unusually weak slate of nominees.
Very low. It struck me as a jawdroppingly stupid idea. Then I read it and it was rather good.