If I understand correctly this amounts to voting against the ones you disapproved. Which I also did and was basically fine with. Most of the reviews were good with a few that really weren't.
Definitely concur on how useful it was. I think I ended up as the inverse of you, in that I voted for about 1/3 of them and against the other 2/3. All were engaging, but only a few were well reasoned, well written, and about interesting topics.
Ok. The survey question suggested that the next review contest might be next year, so I guess I'll just post my review (that I was too slow to complete) somewhere else.
Suggestion: As soon as the submission window closes post a listing of the title of the book and review of every entrant so that people can confirm their entry posted
1. There were some amazing submissions that were IMO at least as good as the best finalists. Maybe the readers should be able to choose half the finalists in the next edition?
2. I'd honestly read more reviews by Scott. Hopefully this review contest didn't dis-incentivize him to post more book reviews.
3. Reviewers should somehow be incentivized to cite papers and other sources to back their opinions on the books. Some reviews were flippantly dismissive of the books' opinions, and I didn't know how seriously to take them as they were unsubstantiated.
Thanks for organizing, would love to see this again. Strongly recommend a word/character cap next year to encourage more concise writing - in my opinion it's more skillful to write a 1000 word review than a 5000 word one.
I agree with this, and yet I think I got the most out of the very long reviews, that probed the matter in a lot of detail and nuance. I don't know what a good compromise would be.
You almost certainly got more out of the longer reviews, they had more content than the shorter one. Presumably you'd get even more out of just reading the entire book, which is arguably just a book-length review of its own arguments.
I think there's an example maximum of nuance in a shorter review that can still give enough room to discuss the book, the reviewer's interpretation, and the reviewer's reaction. IMO an idealized review is short enough that a reader can clearly maintain the reviewer's entire argument in mind throughout the full review-reading experience, something not possible when just reading a book and unfeasible in a review of significant length.
But then again you could get that in a book summary found on Wikipedia and elsewhere. The argument I'm making is that the return on investment of time and effort is often maximized on reading a nuanced 5000 word review, as compared to a 1000 word summary or the 100,000 word book itself.
I'd say 5000 words is a reasonable balance. Some of these reviews are upwards of 15,000 words though, which I think is pretty excessive. That's a full hour's reading for a non-speed reader.
"That's a full hour's reading for a non-speed reader."
Dammit, I'm going to have to time myself reading something the next time, because I have no idea how long it takes me to read a piece, and other people tell me I'm a fast reader. So I don't know if I qualify as a speed-reader or not, but somehow "a full hour's reading" doesn't make my heart quail at the prospect.
I didn't feel compelled to finish reviews I was not enjoying. If I lost interest in a review, I would just drop it (and obviously those reviews won't get my vote). I ended up voting mostly for the shorter reviews, since by and large I felt they were higher quality.
Strongly disagree with this suggestion -- one of my favorite reviews, of Progress and Poverty, was one of the longest ones (if not the longest). I also felt no particular compulsion to finish reviews that I was uninterested in (though this was mostly the fault of the subject materials, rather than the reviewers' writing styles which were uniformly quite good -- part of the problem with voting in this contest is that I feel like most of my votes reflect whether I approve of the material, rather than the review).
This is also a celebration of Scott's own tradition of book reviews, some of which get pretty long. No length caps, please.
Same here. Short summaries can be found on other websites, such as Wikipedia. This is ACX, we are not afraid of reading long texts. Heck, even if the review is longer than the original book, as long as it is great, who cares!
Progress and Poverty was my last favorite. I started reading it several times, and got bored each time because it didn't get to the point.
And, to compare the length with scott's own reviews (wordcount by wordcounter.net, not counting the Title):
arabien nights : 4336
a brief history of neoliberalism : 7519
Global Economic History : 2842
why we are polarized : 4573
Origin Of Consciousness In The Breakdown Of The Bicameral Mind : 6800
---- ---- ----
The natural faculties : 7319
progress and poverty : 18343
Scott may write long reviews, but he always gets to the point pretty fast, and if a review of his ends up rather long, that is because it has a number of points, each of which is itself interesting enough on its own.
Even Scott's famously wordy "Meditations on Moloch" is only 14,250 words.
I also didn't like the Progress and Poverty review. Having a quick look at it again, I think that what I didn't like about it is that it's more of a book summary than a review; in all its eighteen thousand words the review finds very little time to actually critically engage with George's ideas instead of simply presenting them.
(Of course I'm biased because I don't like Georgism at all due to being one of those evil landowners whom it is designed to fuck over; I have a sneaking suspicion that most of the people who _do_ like it just happen to be tenants. The odd thing about taxation policy is that everybody just happens to think that the fairest possible system is one that just happens to benefit them. Personally I support a "flat fee" model of taxation, where nations are more like country clubs and every member pays exactly the same and those who can't pay aren't allowed in.)
The editor part of me fully agrees with and loves your suggestion. The SSC/ACX fan part of me is here in part due to the wonderfully protracted posts -- if I wanted concise analyses I'd go elsewhere -- and hence opposes your suggestion; the reviews are "in character" for Scott's blog. Unifying these parts, I find myself a fence-sitter...
I strongly disagree. The point of this contest, to some extent, is a pastiche of the kind of Book Reviews that Scott Alexander himself writes, and those aren't known for being short.
Given that a kind person has done word counts on the reviews above, it looks like a 5,000 word cap would be more in style for the kind of reviews on here. While I'd be tentatively in favour of "don't make it so long you bore people to death", I don't know how long that would be. If someone can write a Walter Pateresque hard, gemlike flame of a review in 1,000 words I certainly wouldn't discourage them, but neither do I want a bed of Procrustes where a good review gets chopped down to fit an arbitrary limit.
Maybe a suggestion (not hard rule) that people try for around 5,000 words but if they fall short or overshoot, not to worry, unless they end up writing 100,000 words in which case they probably should scrap the review and make it a post on their own blog 😁
Reviewers aren't getting paid, and cutting to a prescribed length is hard and unpleasant work, so I'd recommend letting them ramble on to their heart's content.
Speaking as the author of one of the longer reviews (~14k I think?), there's a decent chance I could have cut it down somewhat. Maybe to 10k, with some effort and sadness on my part. And quite possibly that would have improved it. But I don't think I could have gone down to 5k without just writing something substantially different that wasn't what I want to write. And at 1k I'm thinking what's even the point.
So I think a word count limit would do two things. One, sure, it would probably improve the quality of the submitted reviews, which is your stated goal. But two, it would impose a filter on the sorts of reviews that got submitted.
It may be that the thing I wanted to write isn't a thing you particularly wanted to read, and that filter would be fine in your eyes, but eh. Given that Scott is capable of judging the skill and quality of writing without imposing a word cap, I think a word cap would not have improved matters.
If you think several of the reviews were too long, I'd suggest instead saying something like: "hey Scott, some of those reviews were an awful lot to read. Presumably you already used that as part of your quality control when deciding which to shortlist. But for me personally I think it would have been better if you had used it as a larger part, only accepting the longer reviews if they stood out more than they already did." And then maybe Scott can try to take that suggestion in future, if he thinks there's a vague consensus in favor among readers.
(I think I'd be broadly in favor of it, despite that I wouldn't be surprised if my own would have not made the shortlist in that case.)
(This comment is probably also too long, but I lack the time etc. etc.)
Heads-up: Your form presently allows people to vote multiple times. (I grant this is true regardless what setting you use (e.g. if you limit it to one by Google account, people could just make several Google accounts), but there's not even an attempt to prevent it at the moment.)
I think the lack of engagement on this reflects the fact that people in this stack generally believe that we're all trying to get closer to truth at the expense of personal defense mechanisms, and that cheating on something as banal as a book review contest with a grand or so on the line is detestable and ignorable.
On the practical side, I'd be surprised if any of the finalists cared all that much about the prize. The clear thought that allows one to do well in this sort of contest translates pretty directly into life opportunities that render a grand or so much less meaningful than avoiding the impression of foul play among your peers.
It's hard to untangle liking the content of the book vs the content of the review. And without reading all the books it's hard to tell if the review is accurate. Any advice for voting more rationally?
Yeah, picking a good book to review does seem to be part of it. Except that it's hard to know if the book is good before you're read it. I suppose a good reviewer might read many books and only review the ones worth writing about.
My own advice is not to think overthink it. If you like it (for whatever reason), vote for it.
Alternatively, you could ask yourself: would you want the person who wrote that review to write another book review? If yes, then vote for it, otherwise no.
My thought is that the point of the contest is to encourage reviews that the community finds interesting/useful/thought-provoking, not as a competition of sheer writing prowess, so I'm not attempting to correct at all for "well-written review of a book that isn't interesting to me".
(But some correcting against "this book says things I agree with" vs. "this book says things I disagree with" is probably merited)
I'd say, choosing a good book to review is important part of writing a good long-form book review. Ideally a review imparts worthwhile information, thus worthy books make worthy reviews.
While a skillful and knowledgeable writer can write an engaging and informative review of a book that would not stand as tall on its own merits, reviewing a book with a good content and relating its high points is more reliable way to disseminate good information. Thus not something I would penalize while assessing a book review (as long as the reviewer's own contributions are not bad).
However, this is a point of view of judging reviews as source of information. One could choose choose to reward reviewers writing skills and such, like a teacher writing comments on students' papers. Such educational motive is not bad, but IMO not best suited in competitions.
Yeah, is the goal of the reader reading the review to answer the question, "Should I read this book?" or "What's the summary of this book?". The contest is a bit artificial as my goal in reading isn't really either of those, but more like, "Is there something interesting or useful I can learn from this review?"
Yes, there were many of the reviews i liked not because of the criticism/review, but because of the book content (and vice versa); but I assume for this contest it is impossible to untangle, so am just voting on overall enjoyment.
Thumbs up for the book review! As others are saying, it was great and thanks for all the effort - from scott and all the contributors! If this ever gets repeated, here are couple of books that a budding reviewer might enjoy reviewing (I'd do it myself, but my writing style is bland. I also dont have the guts to compete so openly...)
* Ted Chiang's short story collection. it is very much sci-fi for 'clever people', for lack of better term. there are some stories where i notice things that a layman couldnt possibly notice (e.g. one story is just one big analogy to entropy); makes me wonder how much I am not noticing in other stories!
* Nothing is True and Everything is Possible, by Peter Pomerantsev. I have not read it yet, but his podcast interview on 'Subject to Change' podcast is packed with some fun and intriging stories of life in Russia.
So, how do I vote tactically with approval voting? Assuming it's _very important_ to me that my favourite review wins, and, if it doesn't, it's _very important_ that my second-favourite wins and so on. How many should I vote for?
(I'm mostly joking; I know it doesn't matter that much. But I am interested in the theory here...)
Plainly we need Proportional Representation via Single Transferrable Vote, then a rake of tallymen, vote-counters, returning officers, and hire a count centre. Crowds of cheering supporters hoisting the winner aloft, tight-lipped losers proffering congratulations.
And naturally, pundits to forecast the way the vote is going! And graphs!
I had initially thought this election was using Irish-style STV, and put considerable thought into ranking the 17 reviews. I'm a little peeved since approval voting is a lot easier (but also makes it difficult for me to differentiate reviews I loved from those I merely liked).
Approval voting is actually mathematically better. STV has a lot of weird possibilities for example it's possible that voting your honest preference can change the result to an outcome you like less than what the outcome would have been had you not voted at all
Approval voting is not "mathematically better" than either IRV or STV; they satisfy different criteria (notoriously, approval voting fails later-no-harm in an obvious way, while IRV and STV satisfy it).
STV does have its issues, though; while IRV's failure of no-favourite-betrayal is highly non-trivial to use, STV's is quite blatant.
IIRC there's a theorem that no electoral system can completely eliminate tactical voting *except for* random ballot (i.e. select a ballot at random, whoever that person voted for wins).
(For prize distribution where cents in prize >> #voters, you can get the advantages of random ballot without the randomness; simply break up the prize according to the distribution of votes.)
Uh, do you mean instant runoff voting, like in Australia? That's not a mathematically stable system. It fails the monotonicity criterion, and in simulations it tends to produce weirder, less predictable results than other systems. e.g. http://zesty.ca/voting/sim
Also, ranked-choice ballots are harder to fill out than Approval or Score ballots. I would have preferred Score voting (e.g. an 8-point scale from "Just O.K." to "magnificent") over Approval (the neutered binary version of Score voting).
Also nice is that Score voting lets you express *how much* you prefer A over B. And, Score and Approval voting produce clear numeric results (an average score for each candidate) whereas a collection of ranked-choice ballots can be interpreted in a variety of ways according to one's desired "spin".
We also need a prediction market that trades in probability futures, which can be organized into tranches and then re-sold as a meta-predicitive instrument. I mean, obviously, right ?
Approval voting is not an ordinal voting system. You are asking how to vote strategically in a cardinal voting method by giving your ordinal preferences; this is what makes your question so difficult to answer. With no information on the strength of your preferences, there is no way to tell where you should place your cutoff.
The optimal threshold strategy for score voting (including approval voting, which is just score voting on a binary 0-1 scale), is to just approve everyone you like more than your expected utility.
It almost certainly wouldn't make sense to use "Irish-style STV" for a list, since that's a proportional voting method designed for emulating the diversity of the electorate within a legislature, whereas you'd expect the point here to be more "order this list by total popularity". Approval voting is better for that, and simpler. The optimal compromise would be score voting—just rate each option on a 0-5 scale.
If I understand this right the cutoff threshold is the expected utility of the winner.
So if you expect the winner to be a book review you approve of, you'd approve of every book review you approve of, and if you expect the winner to be a book review you don't approve of, you'd approve of every book review you approve of.
Bwah-hah-hah! Cast my votes, so now waiting with evil glee the final results and the resulting "what? no way!" disagreements.
This was great fun, I'd love to do it again. Thanks so much for giving us all this chance to read and vote on a "selected by personal interest of the reviewer, not a pre-selected slate of reviews" assortment, and thanks to the entrants for their time, effort, and willingness to put these before us for criticism and assessment.
I approve of approval voting, but it would be nice if there was a way to participate in the voting even if we haven't read all of the reviews—like a Yes / No / Didn't Read, where the latter would not affect the score of that review relative to the others.
I don't think there's a way to do that without fundamentally 'breaking' the simplicity of approval voting.
It basically becomes a reddit-style upvote downvote system, where you've got to decide whether "approved by X% of Y% voters" beats out "approved by (X + N)% of (Y + M) voters".
I think if you just haven't read all the reviews because you haven't had time, you've got through June to find time or else skip voting. If you skipped a review because the topic just wasn't interesting to you, I do actually think that's probably a valid reason to disapprove.
There's no reason Approval can't have a "no opinion" option other than "it's not the standard formulation of Approval". Yes, Approval is usually proposed without this feature because simplicity is its big selling point - if you aren't dead-set on simplicity, you go with Score voting instead.
Now, if there's a big imbalance in the *kind* of people who abstain from voting on certain reviews, there could be a problematic bias, but I'm not sure if this is ever worse with a "no opinion" option.
Suppose review of book B is featured on subreddit /r/B where they all love B, causing hordes of voters to read and vote for this particular review, ignoring all the others. With a "no opinion" option, you could get lucky and most voters say "no opinion", so B gets a big bump from these many extra votes — and that's all. Without "no opinion", the hordes arrive and vote "No" on all other reviews, or, if they're honest, abstain from voting entirely. I expect this second case to produce a much stronger pro-B bias than the first case.
range voting (https://rangevoting.org/) has this 'no opinion' feature. Basically, rank each review 1-10, or 'No Opinion' if you didn't read it. The highest average wins.
I want to use this as an opportunity to kindly request you host more user generated content contests in the future.
I believe there are many great writers in this community with fascinating things to say. Unfortunately, many of these people don't write (or share their content) because they don't have an audience. I can attest to how discouraging this can feel. By offering the possibility of an audience (even in the form of the non-successful entries linked to in the Google Doc), it will help promote lots of great writing.
To be specific - I think you could very easily run a "best SSC-like essay competition" amongst the userbase and generate absolute gold and help people discover lots of wonderful new writiers. To make your life easier, you could get a team of volunteers to read the initial submissions and run the process similar to the book review one.
I do think broadening the scope to allow essays that are interesting but not book reviews might lead to some good content.
Would probably have to be restricted against culture war topics, though - at least if it were me, I wouldn't want to deal with the potential fallout from someone *else*'s contentious culture war views being spotlighted on my platform.
+1, and also, love Scott's recent 'peer' review post. If Scott does more peer review, he will end up with fewer essays but higher quality; he could fill in the gaps with community content.
Personally, I was a bit disappointed by the theme of the reviews, as I was voting I honestly had to remember "wait, which one of the 5 books that make analogies between the decline of the Roman empire and the decadence of the US is this one?"
It does feel like the selection could have used more books that are just fun factual rides (e.g. the one about Galen, or the one about the digitalization of books) or at least fun investigation/science/life-experience, etc. Politics is such a silly subject and it kills the mind. Pundit predicting the fall of America because:
1. Berton Woods
2. Waves at China and Roman Empire
3. I'm not a racist/evangelist/anti-equalitarian <BUT>
I'm going to list off word counts, in case anyone wants to take them into account when voting.
Order Without Law review: 14,595 + 5,191 on another site
On the Natural Faculties review: 7,319
Progress and Poverty review: 18,343
Are We Smart Enough to Know How Smart Animals Are review: 5,330
Why Buddhism Is True review: 4,341
Double Fold review: 6,817
The Wizard and the Prophet review: 5,669
Through the Eye of a Needle review: 5,412
The Years of Lyndon Johnson review: 2,138
Addiction by Design review: 5,064
The Accidental Superpower review: 9,017
Humankind review: 7,256
The Collapse of Complex Societies review: 5,554
Where's My Flying Car review: 2,811
Down and Out in Paris and London review: 13,445
How Children Fail review: 4,509
Plagues and Peoples review: 4,200
Quotes of the source are included because you have to read them and they were included by the reviewer; Scott's notes at the top (and in one case bottom) are not.
I'm amazed to see how long my favourite one was. In my rankings I penalised a couple for being "too long" in the sense that I got a bit bored reading them, but I must've enjoyed my favourite one so much that the words just flew by. :)
(Just to clarify, I'm aware it's approval voting so there's no such thing as ranking, but as I was reading them I rated them out of ten and took a few notes to help me remember which was which and what I thought.)
I did the same, but also tried to order them best-to-worst, deciding where each new one fit as I read it. Which was sometimes difficult, because it might have been months since I read the others. But I don't think my ratings out of ten were very reliable either, because my lowest was 5 and my highest was 8 and I'm pretty sure that's not because the quality range was only 3/10.
(The two systems were mostly consistent, but not entirely. Best-to-worst, my ratings out of ten were 8, 8, 8, 7-8?, 7, 7, 7, 7, 7?, 7, 6, 6, 6-7, 6, 5, 6.)
I think I voted to approve of the top 5, somewhat arbitrarily.
If there's one place where "wow, that's a lot of words" should not be a barrier, it's here 😀
I do think that if you're enjoying a piece and find it interesting and amusing and educational, then you're not going to think it too long. There were a few reviews I felt were too wordy, could be more concise, but that was because they struck me as "you're not making your point, or not making it clearly, say what you mean and tighten this up".
It was amazing! A big thanks to all reviewers and to Scott!
One thing that was not working properly: in the runners-up file, links were generally broken. I know this is hard to fix in hindsight. But if you host a contest again, Scott, (I so much hope you do!), you could ask for a format that allows you to keep the links working without much effort.
Is it possible to see what the votes for the non finalists looked like? I voted for a few and would be interested in seeing how my thoughts match the consensus. Additionally the non finalists were hit or miss enough that I don't want to read all of them but would probably be interested in ones that just missed the cut.
I voted for Progress And Poverty (interesting topic), Double Fold (the librarians flooding the comments to explain how it was "all wrong" was rather revealing) and Addiction By Design (important topic).
Dear Scott, would it be possible to extend the voting time by, say, two weeks? It seems hard to match reading the final entries with a busy schedule. Also, thanks a lot for organizing this, I think it was a great endeavour!
On the Natural Faculties was probably my favorite because it 1) changed my views about ancient medical science, 2) concerned a text we all weren't already reading, 3) and most importantly was written fairly well.
I really enjoyed that one too. In particular I thought it was good as a book review because it's a review of such an ancient text that needs so much proper contextualization just to get started, as opposed to reviewing, say, "How to win friends and influence people" totally doesn't.
I can't imagine why you decided to gather less information in the main contest (Approval Voting) than you did in the auxiliary book review contest (Score Voting).
Also, Score Voting is something that ought to work well with a separate "ballot" on the bottom of each review (it's always easier for people to rate the thing they just read).
Funny story. I thought my favorite review of the contest, by a longshot, was the Arabian Nights one. I was just waiting to vote for it. I also figured it was the one Scott was pretty sure would win in those predictions he posted a while back. Only after I reviewed and quantified all 17 in order to vote did I wonder where it was and go find it, only to find out it was his review and not in the contest at all! Doh! No wonder if was so fitting with the themes of the blog.
Was on the fence a bit about recommending the Down and Out review, as it mostly summarises highlights from the book in a digestible format and praises Orwell, rather than being especially strong as criticism per se. However I pondered on this and realised that "summarise highlights of a book into a format digestible over about 20-30 mins" is actually also a good thing which I in fact come to SSC/ACX for, so I don't feel bad about voting for it.
Also voted for the review of The Wizard and the Prophet, but felt it might have done without the bit about the tattoos, however neat a framing device it is.
I understand how difficult it can be to cope with the academic workload, especially when it comes to complex topics. There are several platforms where you can find student freelancers or professional writers ready to help with essay writing. For example, sites like Upwork or Fiverr offer a wide range of freelancers with different experience and prices. Or you can find online sites where they offer to buy essays.
If I understand correctly this amounts to voting against the ones you disapproved. Which I also did and was basically fine with. Most of the reviews were good with a few that really weren't.
Definitely concur on how useful it was. I think I ended up as the inverse of you, in that I voted for about 1/3 of them and against the other 2/3. All were engaging, but only a few were well reasoned, well written, and about interesting topics.
Agreed; this was a lot of fun to participate in from the audience perspective. Hopefully the writers also enjoyed themselves.
Scott: I emailed you several times from a couple of email addresses saying that I sent a book review in, never heard back from you.
Too late now I figure…
If you're not reserving the review for a future contest, you could at least link to it in an open thread.
I don't think I got any of these. Next time I'm going to be more careful about this sort of thing.
Is there already a plan about this "next time" thing?
I mean, he's done a lot of contests (like the ACC ones); presumably he intends to continue doing them occasionally.
Ok. The survey question suggested that the next review contest might be next year, so I guess I'll just post my review (that I was too slow to complete) somewhere else.
Suggestion: As soon as the submission window closes post a listing of the title of the book and review of every entrant so that people can confirm their entry posted
+1
Putting my suggestions as a comment here:
1. There were some amazing submissions that were IMO at least as good as the best finalists. Maybe the readers should be able to choose half the finalists in the next edition?
2. I'd honestly read more reviews by Scott. Hopefully this review contest didn't dis-incentivize him to post more book reviews.
3. Reviewers should somehow be incentivized to cite papers and other sources to back their opinions on the books. Some reviews were flippantly dismissive of the books' opinions, and I didn't know how seriously to take them as they were unsubstantiated.
Thanks for organizing, would love to see this again. Strongly recommend a word/character cap next year to encourage more concise writing - in my opinion it's more skillful to write a 1000 word review than a 5000 word one.
I agree with this, and yet I think I got the most out of the very long reviews, that probed the matter in a lot of detail and nuance. I don't know what a good compromise would be.
You almost certainly got more out of the longer reviews, they had more content than the shorter one. Presumably you'd get even more out of just reading the entire book, which is arguably just a book-length review of its own arguments.
I think there's an example maximum of nuance in a shorter review that can still give enough room to discuss the book, the reviewer's interpretation, and the reviewer's reaction. IMO an idealized review is short enough that a reader can clearly maintain the reviewer's entire argument in mind throughout the full review-reading experience, something not possible when just reading a book and unfeasible in a review of significant length.
But then again you could get that in a book summary found on Wikipedia and elsewhere. The argument I'm making is that the return on investment of time and effort is often maximized on reading a nuanced 5000 word review, as compared to a 1000 word summary or the 100,000 word book itself.
That's fair I guess I'd rather see more shorter reviews than fewer longer ones
I'd say 5000 words is a reasonable balance. Some of these reviews are upwards of 15,000 words though, which I think is pretty excessive. That's a full hour's reading for a non-speed reader.
"That's a full hour's reading for a non-speed reader."
Dammit, I'm going to have to time myself reading something the next time, because I have no idea how long it takes me to read a piece, and other people tell me I'm a fast reader. So I don't know if I qualify as a speed-reader or not, but somehow "a full hour's reading" doesn't make my heart quail at the prospect.
Okay, did a test run. Took the review "Progress and Poverty" which has a listed word count of 18,343 words.
Read it attentively, trying not to skip or gallop through.
Time taken: 21 minutes.
"But then again you could get that in a book summary found on Wikipedia"
Not really. Wikipedia's book reviews are limited by its commitment to neutrality and its ban of original research.
I didn't feel compelled to finish reviews I was not enjoying. If I lost interest in a review, I would just drop it (and obviously those reviews won't get my vote). I ended up voting mostly for the shorter reviews, since by and large I felt they were higher quality.
Strongly disagree with this suggestion -- one of my favorite reviews, of Progress and Poverty, was one of the longest ones (if not the longest). I also felt no particular compulsion to finish reviews that I was uninterested in (though this was mostly the fault of the subject materials, rather than the reviewers' writing styles which were uniformly quite good -- part of the problem with voting in this contest is that I feel like most of my votes reflect whether I approve of the material, rather than the review).
This is also a celebration of Scott's own tradition of book reviews, some of which get pretty long. No length caps, please.
I agree that there should be no length cap. And by George, Progress and Poverty was my favorite too!
I second this. No length caps, please. There were some amazingly good long reviews.
Same here. Short summaries can be found on other websites, such as Wikipedia. This is ACX, we are not afraid of reading long texts. Heck, even if the review is longer than the original book, as long as it is great, who cares!
Progress and Poverty was my last favorite. I started reading it several times, and got bored each time because it didn't get to the point.
And, to compare the length with scott's own reviews (wordcount by wordcounter.net, not counting the Title):
arabien nights : 4336
a brief history of neoliberalism : 7519
Global Economic History : 2842
why we are polarized : 4573
Origin Of Consciousness In The Breakdown Of The Bicameral Mind : 6800
---- ---- ----
The natural faculties : 7319
progress and poverty : 18343
Scott may write long reviews, but he always gets to the point pretty fast, and if a review of his ends up rather long, that is because it has a number of points, each of which is itself interesting enough on its own.
Even Scott's famously wordy "Meditations on Moloch" is only 14,250 words.
I also didn't like the Progress and Poverty review. Having a quick look at it again, I think that what I didn't like about it is that it's more of a book summary than a review; in all its eighteen thousand words the review finds very little time to actually critically engage with George's ideas instead of simply presenting them.
(Of course I'm biased because I don't like Georgism at all due to being one of those evil landowners whom it is designed to fuck over; I have a sneaking suspicion that most of the people who _do_ like it just happen to be tenants. The odd thing about taxation policy is that everybody just happens to think that the fairest possible system is one that just happens to benefit them. Personally I support a "flat fee" model of taxation, where nations are more like country clubs and every member pays exactly the same and those who can't pay aren't allowed in.)
I'm a homeowner and I'm a fan, so there's one point of evidence to the contrary.
Same here.
The editor part of me fully agrees with and loves your suggestion. The SSC/ACX fan part of me is here in part due to the wonderfully protracted posts -- if I wanted concise analyses I'd go elsewhere -- and hence opposes your suggestion; the reviews are "in character" for Scott's blog. Unifying these parts, I find myself a fence-sitter...
I strongly disagree. The point of this contest, to some extent, is a pastiche of the kind of Book Reviews that Scott Alexander himself writes, and those aren't known for being short.
Given that a kind person has done word counts on the reviews above, it looks like a 5,000 word cap would be more in style for the kind of reviews on here. While I'd be tentatively in favour of "don't make it so long you bore people to death", I don't know how long that would be. If someone can write a Walter Pateresque hard, gemlike flame of a review in 1,000 words I certainly wouldn't discourage them, but neither do I want a bed of Procrustes where a good review gets chopped down to fit an arbitrary limit.
Maybe a suggestion (not hard rule) that people try for around 5,000 words but if they fall short or overshoot, not to worry, unless they end up writing 100,000 words in which case they probably should scrap the review and make it a post on their own blog 😁
That makes sense.
I've read Water Pater. Good? Yes. Short? No.
I disagree, but I'm not wholly unsympathetic to this (nor think it wouldn't be a useful constraint either).
Reviewers aren't getting paid, and cutting to a prescribed length is hard and unpleasant work, so I'd recommend letting them ramble on to their heart's content.
On the other hand, 5000 words is a lot (most of my book reviews are 1200-1500 words), so a maximum of 5,000 words would be fine.
Speaking as the author of one of the longer reviews (~14k I think?), there's a decent chance I could have cut it down somewhat. Maybe to 10k, with some effort and sadness on my part. And quite possibly that would have improved it. But I don't think I could have gone down to 5k without just writing something substantially different that wasn't what I want to write. And at 1k I'm thinking what's even the point.
So I think a word count limit would do two things. One, sure, it would probably improve the quality of the submitted reviews, which is your stated goal. But two, it would impose a filter on the sorts of reviews that got submitted.
It may be that the thing I wanted to write isn't a thing you particularly wanted to read, and that filter would be fine in your eyes, but eh. Given that Scott is capable of judging the skill and quality of writing without imposing a word cap, I think a word cap would not have improved matters.
If you think several of the reviews were too long, I'd suggest instead saying something like: "hey Scott, some of those reviews were an awful lot to read. Presumably you already used that as part of your quality control when deciding which to shortlist. But for me personally I think it would have been better if you had used it as a larger part, only accepting the longer reviews if they stood out more than they already did." And then maybe Scott can try to take that suggestion in future, if he thinks there's a vague consensus in favor among readers.
(I think I'd be broadly in favor of it, despite that I wouldn't be surprised if my own would have not made the shortlist in that case.)
(This comment is probably also too long, but I lack the time etc. etc.)
At a minimum, no restriction, but there's a word count at the top of the review.
Heads-up: Your form presently allows people to vote multiple times. (I grant this is true regardless what setting you use (e.g. if you limit it to one by Google account, people could just make several Google accounts), but there's not even an attempt to prevent it at the moment.)
Thanks for running the contest!
I think the lack of engagement on this reflects the fact that people in this stack generally believe that we're all trying to get closer to truth at the expense of personal defense mechanisms, and that cheating on something as banal as a book review contest with a grand or so on the line is detestable and ignorable.
On the practical side, I'd be surprised if any of the finalists cared all that much about the prize. The clear thought that allows one to do well in this sort of contest translates pretty directly into life opportunities that render a grand or so much less meaningful than avoiding the impression of foul play among your peers.
It's hard to untangle liking the content of the book vs the content of the review. And without reading all the books it's hard to tell if the review is accurate. Any advice for voting more rationally?
Yeah, picking a good book to review does seem to be part of it. Except that it's hard to know if the book is good before you're read it. I suppose a good reviewer might read many books and only review the ones worth writing about.
My own advice is not to think overthink it. If you like it (for whatever reason), vote for it.
Alternatively, you could ask yourself: would you want the person who wrote that review to write another book review? If yes, then vote for it, otherwise no.
I did try to be careful about that, so if I thought the review was good, regardless of "did I like the book or not?", it got the yes vote.
My thought is that the point of the contest is to encourage reviews that the community finds interesting/useful/thought-provoking, not as a competition of sheer writing prowess, so I'm not attempting to correct at all for "well-written review of a book that isn't interesting to me".
(But some correcting against "this book says things I agree with" vs. "this book says things I disagree with" is probably merited)
I'd say, choosing a good book to review is important part of writing a good long-form book review. Ideally a review imparts worthwhile information, thus worthy books make worthy reviews.
While a skillful and knowledgeable writer can write an engaging and informative review of a book that would not stand as tall on its own merits, reviewing a book with a good content and relating its high points is more reliable way to disseminate good information. Thus not something I would penalize while assessing a book review (as long as the reviewer's own contributions are not bad).
However, this is a point of view of judging reviews as source of information. One could choose choose to reward reviewers writing skills and such, like a teacher writing comments on students' papers. Such educational motive is not bad, but IMO not best suited in competitions.
Yeah, is the goal of the reader reading the review to answer the question, "Should I read this book?" or "What's the summary of this book?". The contest is a bit artificial as my goal in reading isn't really either of those, but more like, "Is there something interesting or useful I can learn from this review?"
I agree ,JD. I liked a lot of the reviews so I’m glad this thing is happening, but this might be a dumb thing to “vote” on. I’m abstaining.
Yes, there were many of the reviews i liked not because of the criticism/review, but because of the book content (and vice versa); but I assume for this contest it is impossible to untangle, so am just voting on overall enjoyment.
The way I see it I'm voting for the one I most want other people to read and part of what goes into this is picking a good book.
Thumbs up for the book review! As others are saying, it was great and thanks for all the effort - from scott and all the contributors! If this ever gets repeated, here are couple of books that a budding reviewer might enjoy reviewing (I'd do it myself, but my writing style is bland. I also dont have the guts to compete so openly...)
* Ted Chiang's short story collection. it is very much sci-fi for 'clever people', for lack of better term. there are some stories where i notice things that a layman couldnt possibly notice (e.g. one story is just one big analogy to entropy); makes me wonder how much I am not noticing in other stories!
* Nothing is True and Everything is Possible, by Peter Pomerantsev. I have not read it yet, but his podcast interview on 'Subject to Change' podcast is packed with some fun and intriging stories of life in Russia.
So, how do I vote tactically with approval voting? Assuming it's _very important_ to me that my favourite review wins, and, if it doesn't, it's _very important_ that my second-favourite wins and so on. How many should I vote for?
(I'm mostly joking; I know it doesn't matter that much. But I am interested in the theory here...)
Plainly we need Proportional Representation via Single Transferrable Vote, then a rake of tallymen, vote-counters, returning officers, and hire a count centre. Crowds of cheering supporters hoisting the winner aloft, tight-lipped losers proffering congratulations.
And naturally, pundits to forecast the way the vote is going! And graphs!
Lots and lots of graphs! https://www.youtube.com/watch?v=FuaKooWGq4s
I had initially thought this election was using Irish-style STV, and put considerable thought into ranking the 17 reviews. I'm a little peeved since approval voting is a lot easier (but also makes it difficult for me to differentiate reviews I loved from those I merely liked).
Approval voting is actually mathematically better. STV has a lot of weird possibilities for example it's possible that voting your honest preference can change the result to an outcome you like less than what the outcome would have been had you not voted at all
Approval voting is not "mathematically better" than either IRV or STV; they satisfy different criteria (notoriously, approval voting fails later-no-harm in an obvious way, while IRV and STV satisfy it).
STV does have its issues, though; while IRV's failure of no-favourite-betrayal is highly non-trivial to use, STV's is quite blatant.
IIRC there's a theorem that no electoral system can completely eliminate tactical voting *except for* random ballot (i.e. select a ballot at random, whoever that person voted for wins).
(For prize distribution where cents in prize >> #voters, you can get the advantages of random ballot without the randomness; simply break up the prize according to the distribution of votes.)
Uh, do you mean instant runoff voting, like in Australia? That's not a mathematically stable system. It fails the monotonicity criterion, and in simulations it tends to produce weirder, less predictable results than other systems. e.g. http://zesty.ca/voting/sim
Also, ranked-choice ballots are harder to fill out than Approval or Score ballots. I would have preferred Score voting (e.g. an 8-point scale from "Just O.K." to "magnificent") over Approval (the neutered binary version of Score voting).
Also nice is that Score voting lets you express *how much* you prefer A over B. And, Score and Approval voting produce clear numeric results (an average score for each candidate) whereas a collection of ranked-choice ballots can be interpreted in a variety of ways according to one's desired "spin".
STV makes more sense in contexts with multiple winners, but in such cases I still prefer Direct Representation or a simplified version thereof, Simple Direct Representation. https://medium.com/big-picture/simple-direct-representation-cd43becd9837
We also need a prediction market that trades in probability futures, which can be organized into tranches and then re-sold as a meta-predicitive instrument. I mean, obviously, right ?
Approval voting is not an ordinal voting system. You are asking how to vote strategically in a cardinal voting method by giving your ordinal preferences; this is what makes your question so difficult to answer. With no information on the strength of your preferences, there is no way to tell where you should place your cutoff.
The optimal threshold strategy for score voting (including approval voting, which is just score voting on a binary 0-1 scale), is to just approve everyone you like more than your expected utility.
http://scorevoting.net/RVstrat6
It almost certainly wouldn't make sense to use "Irish-style STV" for a list, since that's a proportional voting method designed for emulating the diversity of the electorate within a legislature, whereas you'd expect the point here to be more "order this list by total popularity". Approval voting is better for that, and simpler. The optimal compromise would be score voting—just rate each option on a 0-5 scale.
If I understand this right the cutoff threshold is the expected utility of the winner.
So if you expect the winner to be a book review you approve of, you'd approve of every book review you approve of, and if you expect the winner to be a book review you don't approve of, you'd approve of every book review you approve of.
So it's just honest voting, isn't it?
Bwah-hah-hah! Cast my votes, so now waiting with evil glee the final results and the resulting "what? no way!" disagreements.
This was great fun, I'd love to do it again. Thanks so much for giving us all this chance to read and vote on a "selected by personal interest of the reviewer, not a pre-selected slate of reviews" assortment, and thanks to the entrants for their time, effort, and willingness to put these before us for criticism and assessment.
I approve of approval voting, but it would be nice if there was a way to participate in the voting even if we haven't read all of the reviews—like a Yes / No / Didn't Read, where the latter would not affect the score of that review relative to the others.
I don't think there's a way to do that without fundamentally 'breaking' the simplicity of approval voting.
It basically becomes a reddit-style upvote downvote system, where you've got to decide whether "approved by X% of Y% voters" beats out "approved by (X + N)% of (Y + M) voters".
I think if you just haven't read all the reviews because you haven't had time, you've got through June to find time or else skip voting. If you skipped a review because the topic just wasn't interesting to you, I do actually think that's probably a valid reason to disapprove.
There's no reason Approval can't have a "no opinion" option other than "it's not the standard formulation of Approval". Yes, Approval is usually proposed without this feature because simplicity is its big selling point - if you aren't dead-set on simplicity, you go with Score voting instead.
Now, if there's a big imbalance in the *kind* of people who abstain from voting on certain reviews, there could be a problematic bias, but I'm not sure if this is ever worse with a "no opinion" option.
Suppose review of book B is featured on subreddit /r/B where they all love B, causing hordes of voters to read and vote for this particular review, ignoring all the others. With a "no opinion" option, you could get lucky and most voters say "no opinion", so B gets a big bump from these many extra votes — and that's all. Without "no opinion", the hordes arrive and vote "No" on all other reviews, or, if they're honest, abstain from voting entirely. I expect this second case to produce a much stronger pro-B bias than the first case.
range voting (https://rangevoting.org/) has this 'no opinion' feature. Basically, rank each review 1-10, or 'No Opinion' if you didn't read it. The highest average wins.
I didn't like any of the book reviews, but I voted to approve 2, 4, 6, and 15.
Of course, I approve of doing this again, just to see what people come up with.
The ones you disliked the least?
Yes.
Genuine non-snarky question: what are examples of book reviews you like?
Scott's and Branko's are pretty good.
Thank you Scott for organizing this.
I want to use this as an opportunity to kindly request you host more user generated content contests in the future.
I believe there are many great writers in this community with fascinating things to say. Unfortunately, many of these people don't write (or share their content) because they don't have an audience. I can attest to how discouraging this can feel. By offering the possibility of an audience (even in the form of the non-successful entries linked to in the Google Doc), it will help promote lots of great writing.
To be specific - I think you could very easily run a "best SSC-like essay competition" amongst the userbase and generate absolute gold and help people discover lots of wonderful new writiers. To make your life easier, you could get a team of volunteers to read the initial submissions and run the process similar to the book review one.
Thumbs up for this comment!
I do think broadening the scope to allow essays that are interesting but not book reviews might lead to some good content.
Would probably have to be restricted against culture war topics, though - at least if it were me, I wouldn't want to deal with the potential fallout from someone *else*'s contentious culture war views being spotlighted on my platform.
+1, and also, love Scott's recent 'peer' review post. If Scott does more peer review, he will end up with fewer essays but higher quality; he could fill in the gaps with community content.
Is there a difference between voting No and not voting on an entry? (Is the winner just going to be the one with the most Yes votes?)
Finally!
Somehow I got the impression that all 80 were finalists XD Will the remainder be put up on ACX at some point?
The remainder were already put up awhile back, and the ones that received the most votes were added to the main contest.
Not on ACX though.
Did the reviewers pick the book they wanted to review?
As far as I know, yes.
Thanks for running this, tons of enjoyable content. So many, it's hard to remember all of them!
These reviews were all so good and so interesting! I was really impressed by the quality of all the finalists. Many thanks to everyone involved.
I did enjoy the reviews quite a lot, and I hate to say it, but... I think 17 is just too many. They kind of tend to blur together after a while :-(
For me any more per week would have almost certainly made it start to feel like a chore to get through them.
The 1001 nights review is still the best one by far, lol
Wait it's not in the finalists?
It was written by Scott, not a reader.
Personally, I was a bit disappointed by the theme of the reviews, as I was voting I honestly had to remember "wait, which one of the 5 books that make analogies between the decline of the Roman empire and the decadence of the US is this one?"
It does feel like the selection could have used more books that are just fun factual rides (e.g. the one about Galen, or the one about the digitalization of books) or at least fun investigation/science/life-experience, etc. Politics is such a silly subject and it kills the mind. Pundit predicting the fall of America because:
1. Berton Woods
2. Waves at China and Roman Empire
3. I'm not a racist/evangelist/anti-equalitarian <BUT>
are a dime a dozen.
Wait, there was one saying #3?
I'm going to list off word counts, in case anyone wants to take them into account when voting.
Order Without Law review: 14,595 + 5,191 on another site
On the Natural Faculties review: 7,319
Progress and Poverty review: 18,343
Are We Smart Enough to Know How Smart Animals Are review: 5,330
Why Buddhism Is True review: 4,341
Double Fold review: 6,817
The Wizard and the Prophet review: 5,669
Through the Eye of a Needle review: 5,412
The Years of Lyndon Johnson review: 2,138
Addiction by Design review: 5,064
The Accidental Superpower review: 9,017
Humankind review: 7,256
The Collapse of Complex Societies review: 5,554
Where's My Flying Car review: 2,811
Down and Out in Paris and London review: 13,445
How Children Fail review: 4,509
Plagues and Peoples review: 4,200
Quotes of the source are included because you have to read them and they were included by the reviewer; Scott's notes at the top (and in one case bottom) are not.
I'm amazed to see how long my favourite one was. In my rankings I penalised a couple for being "too long" in the sense that I got a bit bored reading them, but I must've enjoyed my favourite one so much that the words just flew by. :)
(Just to clarify, I'm aware it's approval voting so there's no such thing as ranking, but as I was reading them I rated them out of ten and took a few notes to help me remember which was which and what I thought.)
I did the same, but also tried to order them best-to-worst, deciding where each new one fit as I read it. Which was sometimes difficult, because it might have been months since I read the others. But I don't think my ratings out of ten were very reliable either, because my lowest was 5 and my highest was 8 and I'm pretty sure that's not because the quality range was only 3/10.
(The two systems were mostly consistent, but not entirely. Best-to-worst, my ratings out of ten were 8, 8, 8, 7-8?, 7, 7, 7, 7, 7?, 7, 6, 6, 6-7, 6, 5, 6.)
I think I voted to approve of the top 5, somewhat arbitrarily.
If there's one place where "wow, that's a lot of words" should not be a barrier, it's here 😀
I do think that if you're enjoying a piece and find it interesting and amusing and educational, then you're not going to think it too long. There were a few reviews I felt were too wordy, could be more concise, but that was because they struck me as "you're not making your point, or not making it clearly, say what you mean and tighten this up".
P&P had a couple thousand of those words in an optional appendix, did any others do anything similar? I can’t remember now.
It was amazing! A big thanks to all reviewers and to Scott!
One thing that was not working properly: in the runners-up file, links were generally broken. I know this is hard to fix in hindsight. But if you host a contest again, Scott, (I so much hope you do!), you could ask for a format that allows you to keep the links working without much effort.
Is it possible to see what the votes for the non finalists looked like? I voted for a few and would be interested in seeing how my thoughts match the consensus. Additionally the non finalists were hit or miss enough that I don't want to read all of them but would probably be interested in ones that just missed the cut.
I second this request.
I voted for Progress And Poverty (interesting topic), Double Fold (the librarians flooding the comments to explain how it was "all wrong" was rather revealing) and Addiction By Design (important topic).
I was about to comment about why Arabian Nights wasn't here. I definitely hadn't realized that one was written by Scott!
I was conflicted between voting "more reviews" because more is better, and "same as now" because so many reviews are hard to remember and evaluate.
My actual preference is "same number of reviews, but more often".
Reading the entries was incredibly time consuming. I strongly suggest that you filter down the finalists better next time.
They are all so good...
Dear Scott, would it be possible to extend the voting time by, say, two weeks? It seems hard to match reading the final entries with a busy schedule. Also, thanks a lot for organizing this, I think it was a great endeavour!
On the Natural Faculties was probably my favorite because it 1) changed my views about ancient medical science, 2) concerned a text we all weren't already reading, 3) and most importantly was written fairly well.
I really enjoyed that one too. In particular I thought it was good as a book review because it's a review of such an ancient text that needs so much proper contextualization just to get started, as opposed to reviewing, say, "How to win friends and influence people" totally doesn't.
Note to self: Stop reading Open Threads.
I can't imagine why you decided to gather less information in the main contest (Approval Voting) than you did in the auxiliary book review contest (Score Voting).
Also, Score Voting is something that ought to work well with a separate "ballot" on the bottom of each review (it's always easier for people to rate the thing they just read).
Funny story. I thought my favorite review of the contest, by a longshot, was the Arabian Nights one. I was just waiting to vote for it. I also figured it was the one Scott was pretty sure would win in those predictions he posted a while back. Only after I reviewed and quantified all 17 in order to vote did I wonder where it was and go find it, only to find out it was his review and not in the contest at all! Doh! No wonder if was so fitting with the themes of the blog.
If you want to keep exploring alternate voting methods for these, I suggest using this website next time for quadratic voting: https://qv.geek.sg/
Was on the fence a bit about recommending the Down and Out review, as it mostly summarises highlights from the book in a digestible format and praises Orwell, rather than being especially strong as criticism per se. However I pondered on this and realised that "summarise highlights of a book into a format digestible over about 20-30 mins" is actually also a good thing which I in fact come to SSC/ACX for, so I don't feel bad about voting for it.
Also voted for the review of The Wizard and the Prophet, but felt it might have done without the bit about the tattoos, however neat a framing device it is.
Hi Aman!
I understand how difficult it can be to cope with the academic workload, especially when it comes to complex topics. There are several platforms where you can find student freelancers or professional writers ready to help with essay writing. For example, sites like Upwork or Fiverr offer a wide range of freelancers with different experience and prices. Or you can find online sites where they offer to buy essays.