If you go to an actual bookstore (or library) and look at newsstand prices for, eg, Canteen or Dwell or Parabola or Foreign Affairs or New Left Review or Ampersand or 34th Parallel (or god forbid Science + Nature) and so on you'd have a harder time arriving at such a neat relationship.
Now, if you grouped by "content niche" and analyzed each niche separately you might find in some niches (like "news", specifically) a clean relationship like what you proposed.
Not really. My claim in the essay was that prices didn't depend on the quality of the content. All I needed to support that were a few counterexamples to anyone claiming they did. Though I did see evidence of a trend toward magazines being priced according to paper content, that would have been more than a I needed.
You're smart enough to understand the concept of a statistical relationship, and also smart enough to know that single anecdotes ("my uncle smoked all his life and didn't get lung cancer") don't really prove anything about a statistical relationship.
But look again at what you're claiming here: "price doesn't depend on quality of content".
Here's the evidence you start from:
- Time: lower quality content, ~$5 / issue
- Economist: higher quality content, ~$7 / issue
Which to an outside observer certainly doesn't look like a case of price not depending on quality of content.
To turn the above facts into evidence for your claim you introduce a "price per page" metric, which lets you salvage your point at the expense of introducing a dubious metric.
Leaving aside the dubiousness of the metric, your chosen datapoints don't hold up once you leave the newsstand:
Difference in quality of content may not be a complete explanation for the above differential, but I'd wager a guess it has something to do with it.
And even at the newsstand you can find a wide range of price-per-sheet values (eg: 4c to 20c by my estimates), enough so that you could just as easily, say, have picked "Foreign Affairs" and "Time" and have it come out that higher quality content yields higher per-page prices.
Which is why this looks like a sloppy generalization: you're trying to argue against a statistical relationship (generally "higher-quality content => generally higher prices") by means of anecdote, and not really a carefully-vetted anecdote at that (as evidenced in the subscription pricing).
You seem to be rather arbitrary about what is a dubious metric. Why would you count issues instead of pages? I think that is a stupid metric. Obviously price/page isn't perfect, but is also obviously better/more relevant. You should also check http://www-cs-faculty.stanford.edu/~knuth/joalet.pdf (PDF) this open letter of Knuth regarding publishing.
It was football day so I really didn't want to get into it. Now it's not so I'll respond to your point, which is well-taken.
About the metric:
I wasn't clear on the sense in which I think "price per page" is dubious. From the consumer standpoint it's not irredeemably flawed; $-per-word is better (gets at the same point more accurately) but that'd take actual work to calculate so it's not going to show up in one of these essays.
That said it has some weaknesses, mainly variations on the "Fallacy of Division".
The major dubiousness as a consumer metric is it runs the risk of double-counting "quality" when applied to the nonfiction written word: you can have a better article either by better writing or by being longer + more comprehensive. Within the constraints of magazine writing you'll see a narrow band of variation in "writing style", leaving "length/comprehensiveness" as the major source of one article's "better quality content" when compared to some other article. Thus the risk of double-counting: if a 10000-word article A on Edo-period Japan is better than a 2500 word article B on Edo-period Japan mainly b/c of how much more nuanced and detailed it is, you double count when you:
- praise A for the higher quality of its contents (thoroughness/nuance/etc.) when compared to be B
- praise A for also giving you more to read than there is in B
...which won't ruin the metric but is an easy way to overstate a case. SHORT VERSION: "local" content quality (at the word or page level) is heavily dependent on the overall quality of the larger item it's a part of (the article); when you say "this awesome 10 kiloword article has 10000 awesome words in it" you double-count the awesomeness.
This isn't enough to sink PG's thesis but it is a danger to be aware of when naively applying the metric.
What sinks his thesis is that I can throw evidence that "quality of content" does impact the price people are willing to pay; his example of newsstand pricing is more about the extent to which magazine retailers can "tax" retail magazine purchasers than about the underlying economics.
The various and sundry evidence for "quality of content" impacting price:
- subscriptions (which cut out the middleman) clearly have what he deems "better content" priced at a marked premium to inferior content; perhaps "newsstands" are just a pathological retail channel?
- the subscription price also gives nice upper bounds on publication costs; these are pretty tiny (rounding in a way prejudicially to my argument: 50 pages per issue and 50 issues per year of time magazine => .8c/page; given the similar production values of the economist we can safely assume it's similar). I'll say more on this later.
- the consumer price per unit of content varies widely between genres. Look at "Parabola", a new-agey magazine: it's 9.50 / issue at the newsstand, is always 128 pages, and is printed on 5"x8" paper. When you correct for page size you get ~17c per "page" (really: per 8.5x11" paper's worth of printed words). Do you have an explanation for why content-units of "Parabola" sell for higher prices than content-units of the economist other than some difference in the quality of their contents? If time and the economist selling for ~8.x cents per page @ the newsstand is evidence of the market value of their content why is this not evidence for the market value of "Parabola"
I can do this all day: prices per content unit vary much more widely than the narrow range PG's anecdote suggests; this is true @ the subscription price and @ the newsstand price. You'll neither find find consistent cost-per-page to cost-of-product multipliers nor will you find that going super heavypaper + glossy ink (with concomitant jump in cost-per-page) has any obvious impact on cost of product either (cf vogue or seventeen or W).
After awhile occam's razor will suggest: "newsstand" prices aren't good measurements of magazine prices (compared to subscriptions); regardless of where you measure different types of content sell at different prices per unit of content; likeliest explanation is that quality of content has a lot to do with what price something sells at (which is why there aren't any $20 weekly newsmagazines).
Additionally: publishers are already quite close to the post-medium world, in the sense that their costs-of-publication are already quite small vis-a-vis price.
An alternative explanation for the phenomenon PG is observing is something like:
- (a) each type of content will fetch a different price; this price has a lot to do with what the content is
- (b) for a given "type" of content there's a constraint on what is involved in producing an instance in it; eg: there are no 300 or 500 page newsweeklies, but there are 50-100 page newsweeklies
- (c) within each "type" prices will cluster around some central point but higher quality content will fetch higher prices
- (d) within the constraints of the genre there's a strong correlation with longer == better
- (e) together, (b) (c) and (d) make it look roughly like the time + economist situation: 'economist > time but priced < time per content unit'
...but b/c publishing costs are such a small part of the price to consumer (which has more to do firstly with content genre and secondly with relative quality of content within genre) the "length" as such isn't a big factor in setting price; the relationship described in (e) is literally a red herring.
You can differentiate this scenario from pg's by checking across content genres; if prices-to-consumer are hovering within a narrow range per unit price his looks better, but if price per content unit vary widely across genres mine looks better.
In fact consumers never really were paying for content, and publishers weren't really selling it either. If the content was what they were selling, why has the price of books or music or movies always depended mostly on the format? Why didn't better content cost more?
and paragraph 4:
Almost every form of publishing has been organized as if the medium was what they were selling, and the content was irrelevant. Book publishers, for example, set prices based on the cost of producing and distributing books. They treat the words printed in the book the same way a textile manufacturer treats the patterns printed on its fabrics.
Consider these two claims:
- CLAIM A: "prices don't depend on quality of content"
- CLAIM B: "prices depend on manufacturing cost" (=> CLAIM A)
...which claim would you imagine the author to be advancing? "CLAIM A" -- that content quality has nothing to do with pricing? -- or "CLAIM B" -- that 'publishers...set prices based on the cost of [manufacturing]'?
If, further, interposed between those paragraphs you saw paragraph 3:
A copy of Time costs $5 for 58 pages, or 8.6 cents a page. The Economist costs $7 for 86 pages, or 8.1 cents a page. Better journalism is actually slightly cheaper.
...which claim does that paragraph do more to support?
If I've misread the essay I think I can be forgiven for assuming pg attempted to argue that price is set based on cost of manufacture (and not just that 'quality of content does not dictate price').
If you go to an actual bookstore (or library) and look at newsstand prices for, eg, Canteen or Dwell or Parabola or Foreign Affairs or New Left Review or Ampersand or 34th Parallel (or god forbid Science + Nature) and so on you'd have a harder time arriving at such a neat relationship.
Now, if you grouped by "content niche" and analyzed each niche separately you might find in some niches (like "news", specifically) a clean relationship like what you proposed.