Steve Coffman throws around facts and numbers about Amazon.com but seems a bit
shy of facts and numbers about how libraries really work. He does have his damning "50-65% failure rate"—a
factoid that I believe needs a lot of examination. I’ve spent a fair amount of time in a fair number of public
libraries. I can believe that half of those who come in wanting one specific
title don’t get it. I do not believe that more than half of those who come in to the library leave without finding
things they want. It’s rare (in my experience) to see people leave the library empty-handed.
So let’s throw out a few other facts. Some of these come from various previous
library surveys; some are easy to verify. Key ones come from the 1996 public library statistics posted at
http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=1999306.
Queens Borough Public Library facts come from that library’s Web site.
- Amazon.com claims around 3.6 million repeat customers worldwide; let’s be generous
and say 5 million by now. The best estimates I’ve seen are that two-thirds of all American adults (and almost certainly
a higher percentage of children) are repeat users of America’s public libraries: say 160 million at this point.
As Coffman says, "Clearly, they must be doing something right"—more than thirty
times as many people repeatedly use America’s public libraries as repeatedly
use Amazon.com.
- In 1996, America’s public libraries circulated 1.6 billion items (including
10.5 million ILL transactions), handled 284 million reference questions, and had one billion visits. Total operating
expenditures were $5.5 billion. Note that America’s public libraries circulated more than fifty times as many items in 1996 as Amazon.com might
have sold in 1998 if all of 1998 had been as busy as the last three months (which it wasn’t: total 1998 Amazon
sales were around $600 million, not $1 billion). Note also that, adding reference questions to circulation for
an absurdly low "transaction" count (ignoring millions of programs, story hours, in house use, etc.,
etc.), library cost per transaction was about $2.90 including all expenditures.
- Coffman makes a pointless comparison between Amazon’s worldwide operations and one public library system—and he gets that one wrong. Queens Borough Public circulated 16
million items in FY1998, in addition to handling 3.46 million reference and directional questions. (That number
went up to 17.5 million in FY99, but Queens Borough hasn’t posted a full annual report yet.) Amazon’s 23 million
"transactions" is not more than 1.5 times as much as 16 million in any arithmetic system I know of—and,
of course, those 16 million circulations meant at least 32 million transactions, and probably closer to 40 million (assuming that half of them
involved a catalog lookup). A "fairer" comparison might be how many books Amazon actually sold in 1998
and how much it cost them to handle them. We do know how much the dreadfully expensive and inefficient library
system spent on technology to support those 40 million transactions (almost twice as many as Amazon.com), as well
as all the other technology in the library: $2.342 million dollars. That’s less than one-fifth as much as Amazon.com
spent in 1997—a year in which Amazon.com sold a lot fewer books. Does that mean that good public libraries are more than nine times as efficient as
Amazon.com—or does it mean that Coffman’s numbers are meaningless? ("Nine times": even accepting Coffman’s
"1997 expenditures and 1998 transactions" equation, we get $0.54 per transaction for Amazon.com and $0.059
per transaction for Queens Borough. Let’s be really tough on public libraries and cut Queens Borough’s transactions
down to the minimal 32 million—you do have to check everything both out and in. That brings the cost up to $0.073
per transaction, so Queens Borough is only 7.4 times as efficient technologically as Amazon.com. My bad. Clearly,
we should trash Queens Borough’s model for the more efficient private-enterprise.) (This is all silly, of course.
Queens Borough’s total technology costs cover a lot of territory; I wouldn’t be surprised if less than half was
for their catalog and circulation system. I always try to use the figures that would most favor the other side
of an argument: it’s more fun that way.)
- How much would the ELL Computer System ("the system from ELL?") actually
cost? There’s no way of knowing. My best guess is somewhere between $50 and $200 million per year. How much do
American libraries spend on commercial library automation systems, including academic libraries? I don’t have firm
numbers, but I believe it’s around $160 million per year. Savings from the One Big System? Hmm…
- I really don’t understand this one, but it may have something to do with Coffman’s
inexhaustible need to put down librarians and libraries. He says flatly that "libraries have never really
sat down and figured out what [ILL] lending a book really costs." At that point I screamed, "of course
we have: it costs $25 to $30." And three paragraphs later he cites the study that says that. Maybe he doesn’t read what he’s writing? RLG and ARL
studied ILL costs a couple of years back. The average came out to around $30, if you add in all costs for both
borrowing and lending institutions. We (libraries) "sat down and figured out" what it costs.
- Yes, $30 is too much (it’s probably lower in public libraries, but probably
not a lot lower). People who want to improve library service in realistic, incremental ways are working on ways
to bring that down and bring ILL closer to the user. But note that Coffman compares a known, quantified cost with
a misleading, unquantified assertion about Amazon.com and its competitors. He says that these companies "can
afford to individually pack and ship books…out to customers for a small
fraction of their tiny retail margins." Huh? Amazon.com charges
$3.95 to pack and ship a book, in addition
to some fraction of their margin (which is 40% or more for most trade books, not what I think of as a "tiny retail margin")—and they’re
losing money on every order they ship. (Some economists believe that this is literally true: that Amazon’s model
will never be profitable until
they raise prices or increase shipping charges.) Now, for interlibrary lending you can at least double the costs under the best of circumstances:
after all, the books need to get back to the originating library and go back on the shelf. What are Amazon’s actual
costs? I get some inkling based on the $7 extra they charge for my new book {since eliminated}, probably because they don’t get a huge discount
on it. Is half of that surcharge going to pay for handling? I don’t know. Mike Dahn posits that $9 per item is
about the best you can do for a complete borrow-and-return cycle. I think he’s a little too optimistic, but I use
the $9 figure for the rest of this commentary. I’d guess $10 to $15 is more plausible, unless you really can convince
people to work in libraries for less than they’d make at Burger King. (I’ve done ILL paging, at UC Berkeley; it
was great exercise and fun for a month or two, but it gets old real fast…and it’s inherently not a fast-and-cheap process.)
- You have the key numbers here already. Even if the ELL computer system cost
nothing at all (which I’ll suggest is off by $50 to $200 million a year), and if you could get total per-item costs
down to $9, and you spent nothing at all on administering the system [now there’s an interesting set of assumptions],
then, for each 10% of current library circulation that was fulfilled using this new system, you would spend 26%
of the budgets of all public libraries. So, even if you can keep local collections good enough so that the 80:20
rule applies, 52% of total library
budgets would be devoted to this new system. Add in realistic overhead and I’ll assert it’s closer to 60%. You
might be able to cut back local acquisitions by 20%. Public libraries average 15% of their budget for acquisitions:
there’s 3% freed up. Where’s the other 49-57%? Are there really libraries where 57% of costs go to circulation
clerks?
- I do go on (and on and on…) about the details, but the bottom line is above.
To bring forth this great step (forward?), you would have to gut local libraries. It would be a Viet Nam operation:
you’d have to destroy the libraries to save them. In which case, you’d better be damn sure that you’re doing something
worthwhile.
A Few Opinions
Even if ELL made economic sense, I think it’s a bad idea for libraries. Here
are a few of the other reasons why, in no particular order.
The Great Amazon Catalog: Huh?
{While I still believe everything in this section, it's clear that library folk
are so good at trashing ourselves that nobody else seems to want to defend online catalogs or seems to care about
the gross inadequacies of Amazon's "catalog." So, fine, we're terrible, they're great...}
Local Libraries
Mike Dahn is realistic enough to recognize that ELL would not work without huge
amounts of centralization. That seems to be OK with Mike. A central catalog and circulation agency, staffed by
only the most benevolent librarians. A central clearinghouse for fees. A "central buying authority" to
pool whatever acquisitions money remains.
There goes the local library with local emphasis. In its place comes one or a
group of pure monopolies. Run by whom? The Federal Government (as an extension of the Library of Congress, perhaps)?
OCLC? Who do you trust? Who runs The Library of the United States?
I find this whole line of thinking repellant. I don’t like monopolies. I think
the historical record shows that monopolies are inherently not efficient, not benevolent, and not service-oriented.
State monopolies really only work "well" in entirely socialist economies, and the track record for such
economies isn’t all that great either. The benevolent private monopoly? In what alternate universe?
Queens Borough Public Library is reasonably well funded. (Several of the libraries
in my neck of the woods have significantly higher per-capita funding, but they’re all a lot smaller.) If it gave
up half of its acquisitions budget and all of its information technology budget to these centralized authorities,
it would "save" seven million dollars—enough for around 800,000 transactions in this wonderful new world.
That’s five percent of Queens Borough’s 1998 circulation, with a substantial loss of local autonomy and service.
What a wonderful new world! Each of Queens Borough’s registered users would be able to borrow one item a year from ELL and it would only cut Queens
Borough’s local collection development in half. Great stuff.
Local libraries should reflect local needs. I suspect that thousands of local
public libraries have local history collections; hundreds have oral history collections and important local-history
projects. I’ve yet to hear of a decently-funded local library that didn’t put on local programs, serving the special
needs of the local community, and that didn’t tailor its collection to meet local needs. That costs money—money
that would disappear into the ravenous maw of ELL.
Privacy, Equity, and Other Silly Concerns
The special services that add so much to Amazon.com depend on detailed personal
profiles, including profiles of everything you’ve purchased. That’s one of several reasons I don’t buy from Amazon.
(Which is pretty silly, since my reading habits are remarkably innocuous. Blame it on a father who’s a true Constitutional
Conservative: I think my reading habits should be my own business.)
A single national circulation system would be a goldmine for the FBI, NSA, and
miscellaneous hackers around the world. A single worldwide circulation system would be even more wonderful. "Let’s have a list of everyone
in Myanmar who’s checked out a book that’s anti-regime." "How about a list of everyone who looked at
works on socialism?" "Who’s looked at works that discuss juvenile sexuality?"
Think this wouldn’t happen? The same tools that send e-mail when a "book
that would interest you" arrives are the tools that make it easy to produce such lists, and virtually assure
that such lists would be produced.
Opt out? Not and keep checking out books, you couldn’t. Eliminate the history
of what was checked out previously? I hate to say it, but if I was advising the NSA or the internal security forces
of some other country, I’d suggest hiring a hacker to make sure that the one
single ELL site was continuously and unobtrusively copied to a secure
server for later examination. {This section has been called paranoid. I worked at the UC Berkeley Doe Library in
the 1960s and 1970s, and remember FBI visits demanding records of past circulation. That may explain the paranoia.
But of course, we're all better now, and no future government of any political bent would ever be as invasive...right?}
Equity? This new system would be grossly inequitable. Citizens in poorer communities
would have even less access than they have now. Poorer citizens in wealthier communities would lose access to the rich local collections—and
be expected to pay for most access for the Great Library in the Sky. "Partially-subsidized" is a code
word for empowering the haves at the expense of the have-nots. It strikes at the very core of public library values.
But then, Coffman seems to think that subscription libraries are a neat idea, and feels it’s foolish to deny some
citizens (the wealthy) services just because a library can’t offer them to everyone (the poor).
My other silly concerns generally resonate with those already stated. I believe
ELL would:
- Destroy effective local libraries
- Establish monolithic national (or worldwide) agencies that would control library
service, no matter what the original intention
- Cost more to do less than current public library services
- Empower the powerful and disadvantage the disadvantaged.
Picking at Pieces, 1:
Points in the Original Article
I’ve tried to avoid repeating stuff discussed above, and much of this won’t make
sense unless you’re looking at the original article.
What studies show that Amazon.com users are heavy library users? Oddly, Coffman
doesn’t mention any. If they’re Amazon’s own internal studies, well…
The Collection
Do public libraries actually fail their borrowers 50-65 percent of the time? I flat-out don’t believe it. In my experience,
most public library users (myself included) don’t walk in the door with one specific title in mind and walk out
if that title isn’t available.
Most library catalogs don’t go beyond the local collection? In what public libraries?
In the ones I use, online catalogs almost always include options to search farther—and the reference desk probably
has access to FirstSearch or Eureka.
But now Coffman starts offering solutions, and now it starts getting nasty:
- "Do away with the local catalog." Charming. Most users in most libraries
want to use the local collection first.
Good union catalogs recognize that and work accordingly: you see your own collection first. Coffman wants to drop
the local catalog—but then, it becomes apparent that Coffman doesn’t believe in local libraries. (Many users, perhaps
most, go directly to the shelves—but, as we see later, those shelves have been stripped of all but the most ordinary
in order to support ELL.)
- "Build a global catalog…[based on] WorldCat." In other words, swamp
users with grotesque amounts of irrelevant data when they’re trying to find stuff in the local collection.
- "It would make every library that used it…the equal of the richest and
best-known research institutions in the world." Nonsense. Locality matters…and, incidentally, a portion of that 40 million titles can’t be borrowed because it’s unique to those
research institutions.
Circulation System
Here, Coffman trashes existing library systems: "the most rudimentary online
catalogs and circulation systems." They offer "only the most limited functionality" and "the
most basic search options: author, title, and subject." More or less like Amazon.
"Patrons generally cannot simply check out books by themselves." Not
true in the libraries I use. "They must keep staff standing by to handle exceptions." This is a bad thing? The little old gentleman who doesn’t
understand the Web should just go to hell? He doesn’t use Amazon: count on it. Once again: 25 to 50 times as many
people in the United States
use public libraries as use Amazon in the world. Tens of "millions
of people" seem to find public libraries workable.
"Many people still want to visit us in person; just as many wish they could check a book out on the
Web." That’s an absolutely unfounded assertion, one that I suspect isn’t true. Coffman’s making up his facts
as he goes along.
Costs
Yep, it’s going to cost "lots of money." And now Coffman starts getting
really tricky. He cites $12.5
million as Amazon’s online-systems cost—but that’s for 1997, not 1998, and it doesn’t cover any where near 23 million
transactions. We know that Amazon’s losses are growing at a phenomenal rate.
Here again, Coffman trashes local catalog and circulation systems. "We would
only need one circulation system for the entire catalog." Privacy? "A software design issue." Right.
So is a totally stable Windows environment: just a software design issue.
"It would likely still cost far less than libraries currently spend running
their own feeble systems." Why? Because Coffman says so, based on absolutely no facts whatsoever. "Substantial
savings in staff costs"—because we don’t have people to handle exceptions. "All those clerks standing
behind circulation desks" don’t make up a huge percentage of my own library’s costs, and I’ll bet they don’t
elsewhere.
Aha—but Coffman thinks the system will only cost "$12 to perhaps $20 million"
to develop and operate. Why? Because Amazon, handling some fraction of one library system’s circulation, did it for that. Nationwide, American public libraries
circulated something like 1.5 billion
items in a recent year. Figuring a nominal two transactions per circulation, that’s three billion transactions:
150 times the traffic of Amazon. For the same price? Nonsense.
Reality Check
Coffman basically asserts that libraries shouldn’t bother with well-rounded collections:
they’re "wasting patrons’ money." So we need to "shake off the old ways of doing things"—in
other words, scrap local management and local libraries.
Public libraries should have "key current titles" and cut back on balanced
collections. For the exceptions, the 20% of circulation that are what
libraries are all about, well, let ‘em wait. Or you buy a copy when it’s
been requested twice…not a bad figure, but it won’t work.
Coffman counts on local collections to make the scheme work; oh well, it will get diverse enough. And we’ll collaborate
on collection development. And, of course, it’s really a parasitical system, relying on academic libraries to actually preserve the records of the civilization.
"Large libraries could make a bundle reviving and reprinting forgotten books
from their own collections." Nonsense, and he knows it. But he’s a market man from the word go: libraries
can make money from this and that. "Entrepreneurial libraries"—now there’s a phrase to warm your heart—become
mini-publishers. Hey, maybe they can turn profits! Oh yes: and we can have B&N handle returns, until Amazon
puts it out of business.
New Revenues
Coffman wants libraries to be money-making enterprises. Oh boy: we can refer
people to Amazon. We can sell used books; after all, we’re going to dump everything that’s not current. So we turn
this megacatalog into a used bookstore.
"If libraries truly began culling their collections as they should": again, Coffman simply doesn’t believe
in diverse collections. Ah, but next…
Libraries "allow patron access to the complete collection, subsidize what
it can, and ask the patrons to pick up the rest of the tab."
Here it is: subsidize the haves, and forget about the have-nots. Safety net? That’s old thinking. Revenue stream: that’s where it’s at. People don’t
have personal computers? Too bad. People can’t afford to pay a few bucks for anything that isn’t popular? Who cares?
Oh, maybe we can establish special needs-based welfare library rights: that will add dignity to the safety net
that public libraries represented before ELL.
"For an additional charge:" how is it that public librarians have not
risen up with appropriate indignation to this constant theme?
The Home Stretch
What’s really happening here? Not a marvelous new collection of 43 million whatever;
at least not without several billion
dollars per year in new funding. Browsing categories? With 43 million titles? Spelling correction? It’s almost
impossible to use a spelling variation that doesn’t represent actual titles in very large catalogs. Recommendation
service? The NSA and others will love it.
"We never waste a patron’s time." Nonsense. We’ve gutted the local
collections to pay for a portion of this monstrosity, so when they just want a good mystery, their selection is
far more limited. "A little longer" will be weeks, months, or never for arcane items; but who cares?
"Some fees involved." Bye-bye, equitable public library. "Radically
reduce the traditional costs of library operations." No evidence
whatsoever. Let’s face it: it will be a lot more expensive to have clerks picking up those books all around the country instead of
checking out stacks of books that patrons have picked up. Using a money-losing system as a model makes no sense.
"All sounds too good to be true, doesn’t it?" Yes, because it isn’t.
It’s a for-fee dystopia that would ruin America’s libraries.
Picking at Pieces, 2:
The Response to the Article
"The vast majority" of 250 responses were positive. Okay, let’s assume
that’s true (we have only Coffman’s word for it, but)… Actually, it’s likely, since many of us who thought the
idea was stupid simply said "there goes Coffman again" and dropped it.
Here, Coffman’s trying to respond to objections.
The Miscellaneous Stuff
"Who cares if Amazon loses money?" We care because Coffman’s purported
cost structures are based on Amazon.
He repeats an out-of-date and possibly inaccurate figure for what it costs Amazon to do something. We spend "many
times as much" on systems "and get much less out of them." Nonsense. We spend perhaps 15 times as
much to do 150 times as much work: library systems are far more efficient than Amazon.com.
In the next paragraph, Coffman drops the next bombshell: books must die for his system to work. "We will
have to deliver a lot of that food electronically." Here, of course, is an answer to many objections: make
it digital and all problems are solved.
The problem of people who don’t own computers or prefer not to use them. "Temporary."
They’ll die out. Who cares anyway? Oh yes, we’re going to have "lots of good customer service people on the
floor"—with what money? The money we saved by firing all the…customer service people?
Interlibrary Loan
Amazon.com et al don’t ship individual items for "next to nothing."
They charge $3 handling, and it’s pretty clear that they lose money on each order.
I’ll stick with $9 as a highly optimistic figure for total ILL costs, per book. That’s with no staff involvement in selection and
automatic routing.
Once again, Coffman chooses to regard the 20% as irrelevant: turning public library
policies on their head.
But then, his real answer is "all of this material—or all of it that matters, anyway—will be digitized."
Whoops. "All of it that matters."
Pay attention to the attitude behind that statement.
Library Cooperation
Why didn’t people focus on the costs? Because they have no idea what they could
be. Five billion dollars incremental
is just too big a number to even consider.
And if we don’t do it, the world will pass us by. Or at least Steve’s World:
the world where market is everything, and public agencies are irrelevant.
Picking at Pieces, 3:
Mike Dahn’s Commentary
First off, I wonder that there are no commentaries from real, honest-to-goodness,
public librarians. Coffman isn’t one. Dahn certainly isn’t one. But never mind.
I found Dahn’s commentary particularly interesting because he almost sees what’s
happening—but refuses to accept it for what it is. In the process, he’s argued for a remarkably statist, monopolistic
future. But then, I’m an enemy: I would say that.
More than in the other areas, these comments make almost no sense unless you
have the Dahn piece in hand. Sorry about that.
Building the Catalog
Dahn does an interesting analysis. He fails to mention one fatal flaw of distributed
Z39.50 supersearches: they rule out most "fancy" access, including browsing. It just won’t work.
After lengthy discussion, Dahn again goes with the "establish a monopoly"
model. Amazing that this doesn’t seem to bother anyone.
Incidentally, the problem of "filtering" a huge catalog so that only
the local holdings show is not
trivial, particularly as it affects anything but known-item access. If you think about it, browsing doesn’t work
here either—particularly since your local public library will almost never have even one percent of the titles.
Oh, that’s right, we really don’t care about local holdings.
I’m fascinated by the idea that the library community doesn’t include professional
Web developers…unless that community is defined as excluding RLG, OCLC, and a few dozen others like Innovative,
TLC, Ameritech, and so on. What, exactly, would outside Webmavens bring to the picture?
Oh yes, we would need numerous heavy-duty servers and high-speed network links
and support staff and more. Not free? True.
How much would it cost? "A lot of money." But never mind…
Circulation
Polling for circulation information? Possible, but incredibly cumbersome.
Seamless ILL
Here it gets interesting: "we" must "completely rethink the way
we allocate library resources." What does this mean? Let’s see…
This discussion presumes a national datase of patron and circulation information.
Forget privacy.
This model requires circulation clerks for remote materials. But we’ve eliminated them to save money.
Aha: people might be going less and less to their local library. Big loser: communities.
Another big loser: true local libraries.
How about the problems?
- (1) If any library refuses to check an item-in-hand out to a local patron because
someone else has requested it cross-country, that library is abusing its taxpayers. "Might" cause political
difficulties? I certainly hope so.
- (2) Extended loan periods further reduce availability.
- (3) We pay for this nonsense by not
buying books—even though we know (or should) that even the cheapest ILL must cost more than the cost of local circulation. Small librarians will not be able to offer their customers more for less: there’s
no cost model that can make that case. And what the heck, we’ll give children less access; just because they take
out more books… Patrons become "customers," able to check out a given number of books per year. There’s
a wonderful new twist in public library service! But then we sell those customers more credits. Now we’re selling library service. Can I sell my welfare allotment (that is, the portion of local welfare
that might represent "my share")? Call me one of those critics claiming inequity, particularly since
$9 is at least three times the typical cost of a circulation at present. Ah well, it’s "harsh reality."
We can’t afford public service: we’re too busy serving the public.
Another monopolistic, centralized, system: this time for billing every library
in the country. With all libraries paying for it.
Dahn offers a good discussion of the problems involved in getting larger libraries
to sign up. His "non-bureaucratic" ideas would be very bureaucratic indeed. Determining "low-quality
materials"? Wow! Let’s say that the most specialized library winds up paying $18 for ILL, while UC Berkeley
(say) pays $6.75. This formula absolutely advantages very large libraries: they have more to offer.
Of course, the incentive of "billionths of a cent" is meaningless unless there are billions of circulations—and
if there are, we’re talking $9 times however many billions. Where’s the money? Otherwise, the tax differential
must be significant.
The whole system pretty much falls apart at Stage III, short of nationalizing
all libraries and then tripling their budgets. (But then, that’s really what’s going on here...except for the budget-tripling
part.)
Partially Shared Collection Development
Pooled collection development. With pooled authority. A "central buying
authority." Another monopoly, in other words. A bit scary? Only a bit?
"The benefits to be gained by partial centralization far outweigh those
fears." We all know that monopolies are more efficient. Why not just turn all our budgets over to LC? And
we "imagine" operating costs of $20 million per year? On what basis? Because Amazon may be spending two-thirds
that much to handle less than one percent of that traffic?
Guaranteed: the people in such a service will never feel like employees of the nation’s libraries. They will be the rulers: not only over
libraries, but over library-dependent publishers.
Some Final Thoughts
As Dahn admits, he really doesn’t offer details—or anything like numbers. But
he believes. And here it gets
good:
- "There are probably a million reasons why this project won’t work. Let’s ignore them." Well, so much for this
critique: naysayers are just worriers.
- Now I’m an "enemy." A naysayer is an enemy. Interesting language for librarians. And we’re told that "we must persist"
no matter how wrong-headed it is.
This is a crusade, not a plan. "Creative, flexible centralization."
We’ll offer advertising. On a publicly-funded system. Great. Not offensive? "Exclusive
partnerships" with booksellers. Not offensive?
Does Dahn think of himself as a socialist? Probably not. Is his proposal a great
leap into statist libraries? Absolutely.
Mere details: $5 billion here, $9 billion there. Mere trifles.
How can you argue against this nonsense? In fact, you can’t.
And yet, I just have. What a waste of time.
Written and posted August 30, 1999. Revised (to separate the background section,
otherwise unchanged) September 8, 1999.
Trimmed slightly August 26,
2000.
Return