Question for booksellers: what do you do with the ones that have no barcode?
Summary: yair25 is developing a book listing tool focused on older books lacking barcodes, which reads title and copyright pages to gather information. They are seeking feedback on how booksellers currently handle such items. 1786davycrockett shares concerns about the tool's limitations in accurately determining editions due to the complexities of book publishing, such as ISBN reusability and varying edition indicators. They advise caution regarding the use of copyright dates for dating publications and suggest understanding the intricacies involved in bibliographic research. yair25 acknowledges the limitations and clarifies the tool's scope, emphasizing that it doesn't replace expert judgment but offers a more reliable way to process straightforward data from the books themselves.
I've been building a listing tool that's only for books, and I keep running into the same thing — the general AI listers do fine on anything with an ISBN and fall apart on older stock. They tend to read whatever text is boldest on the page, which on a 1930s book is almost never the edition line.
So the one I built reads the title page and copyright page instead, and for every field — author, translator, publisher, year, place, edition, printing — it shows you the exact line it came from. If the book doesn't say, the field stays blank and gets flagged rather than filled in with a guess.
What I actually want to know is how people here handle this now. Do you type those in by hand? Skip them? Lot them together and not bother?
It's free and I'm still in testing — after feedback, not customers. I can't post links here, so message me if you'd like to try it on a batch and I'll send it over.
1786davycrockett
·Aug 18, 2026 at 5:21 AM@yair25 I neglected to mention that ISBN codes are not engraved in stone -- they are regularly "pulled" by the publishers as titles stop selling or become outdated, or for any number of reasons.
So the publisher "recalls" those ISBN codes -- which, at a bookstore level, requires employees to hunt those particular titles down, pull them off the shelves, and return them to the publisher -- resulting in a "dead" ISBN.
Months later, that identical ISBN will be re-issued by the same publisher -- but very often with an entirely different book! So ISBN codes are kind of a "revolving door" in the publishing business.
Years ago, I attempted to explain to whomever was setting up the eBay catalog, the nature of ISBN codes, and that the ISBN is not intended as a "constant" for individual titles; but I got nowhere -- even when I demonstrated THREE different books from the same publisher, in three different genres, that had the identical ISBN code!
So your project may, unfortunately, be an exercise in futility -- cataloging ISBN codes is like chasing after shadows!
yair25
OPAug 18, 2026 at 6:00 AM · EditedThanks for that, and you're right.
The tool reads the copyright page and reports what's printed there. On a Grosset & Dunlap reprint that's 1927 on a book actually printed thirty years later — and calling that a "publication date" is simply wrong. I'm going to relabel the field as the copyright date, and flag any book where the © is the only date on the page, so it reads as "don't trust this year" rather than an answer.
The Hardy Boys problem I can't solve. Comparing flyleaf listings front and rear, cover art variations, price changes — that's expertise, not text on a page, and I'd be lying if I said a photograph gets you there.
What it does handle is narrower than I've been describing:
— Books where the copyright page does state the printing: number lines, "second printing", "first edition thus"
— Limited editions, including the copy number
— Ex-library copies, book club editions
— Translators, which most listing tools drop entirely
— Facsimiles where the title page says 1922 and the copyright page says 2003
— And for every field, the exact line it read it from. If the book doesn't say, the field stays empty and gets flagged — it doesn't guess.
So: no help on a series reprint that gives you nothing to read. Useful on the general pile where the book does state what it is, and where the alternative is typing it in.
On the ISBN point from the other reply — that's an argument for this rather than against it. I don't catalogue by ISBN at all.
If you'd tell me what else you'd flag, I'd take it. You clearly know where this breaks.
1786davycrockett
·Aug 18, 2026 at 2:24 PM@yair25
It sounds as if the listing tool which you are designing will only be applicable for certain "bare-bones" information, but perhaps side-stepping some item descriptions which are the indicators establishing the collectibility of particular editions, which may not be revealed simply by listing the copyright date or ISBN.
Many eBay sellers simply enter only the portion of the item descriptions which is required by eBay -- in the space designed for the ISBN, we simply enter "Does Not Apply," since it seems that actually filling in an ISBN code will tie that item into the eBay catalog, which unfortunately is jam-packed with spurious information: since eBay insists that the seller is responsible for all information in the description, many eBay sellers do not want their items tied into eBay's incorrect data, which may then result in "not as described" cases.
As antiquarian booksellers have acknowledged for well over 150 years, there are simply too many esoteric variations involved in establishing "which edition is which" -- some publishers use a letter code, and some use a number; some only put a year on the title page for a first edition, but not for later editions; some use a variation of a date on the last page; some use a letter code at the last page, in the spine area; and so on, and so on.
So I'm not sure how the listing tool which you are describing would actually be of assistance to any but casual booksellers, who may be totally unaware of how to determine what features actually define the value of particular editions.
Establishing a publication date is quite often not as easy as it may seem, and I'm not sure that that the listing tool which you are designing may account for all the variations which many publishers use to distinguish one edition from another.
yair25
OPAug 18, 2026 at 3:25 PMYou're right, and I'm not going to argue the parts you know better than I do.
Some things no tool can do. Establishing which edition is which — letter codes, dates on the last page, spine codes, cover art variations — takes a trained eye and years of knowing what to look for. I'm not building that and I'm not claiming to.
What I built handles the part where no eye and no knowledge are required. What's actually printed on the page:
— Author, translator, publisher, year, place of publication
— Any stated edition or printing: number lines, "second printing", "first edition thus"
— Limited edition statements, including the copy number
— Ex-library copies and book club editions
— For every field, the exact line on the page it was read from
— A blank, flagged field when the book doesn't say — never a guess
— A flag when a year comes only from the copyright page and can't be trusted as a publication date. That one came from your last reply.
It reads. It doesn't judge. On a Grosset & Dunlap reprint it returns almost nothing and says so, which is the correct answer.
Your point about "Does Not Apply" in the ISBN field is the most useful thing anyone's told me this week. If sellers are deliberately staying out of eBay's catalog because the data in it is wrong, then reading the book itself is the alternative rather than a shortcut back into that catalog. I hadn't framed it that way.
booksnap-bench.vercel.app — free, no signup, if you're ever curious.
1786davycrockett
·Aug 18, 2026 at 6:09 PM@yair25
Entries for translator, year, and place of publication will be tricky.
Not every foreign work -- particularly pre-World War II editions -- include identification of the translator -- and many current works (particularly books in public domain) don't bother either.
If there is no accurate edition notification, how would you establish a publication year for those editions?
By "place of publication," are you referring to the publisher, or to the printer, since both may have differing locations (and sometimes different countries!). Many books will often list multiple locations as well -- the original publisher (and sometimes date published), plus the reprint publisher (and sometimes date published), as well as the name and address of the printing company.
All of this does get complicated, doesn't it?
1786davycrockett
·Aug 18, 2026 at 5:04 AM@yair25
Copyright dates on books before 1969 are iffy -- not every publisher bothered with edition dates -- so please don't attempt to "date" an edition by the original copyright date.
There are far too many variations that you may be overlooking. For instance, I just listed about 14 different HARDY BOYS books -- and obviously the original copyright dates are useless with those!
Dating the HARDY BOYS books required comparing the front flyleaf listings with the interior listings, and then finding the correct dates for the rear flyleaf listings, as well as the correct dates for the rear cover listings.
And other HARDY BOYS sellers also require a breakdown for the cover art as well as cover design, as well as if there were any changes to the "suggested retail price" listings (including for Canada) on the front flyleaf -- not to mention any text differences; the number, artist and style of interior artwork; the type of paper upon which the frontispiece is printed (or whether a frontispiece is even available); if any additional advertising is included at the rear pages; etc.
So it kinda makes me wonder if the project upon which you are working is really worth the effort, since there are so many variations that are generally unaccounted for by the casual book person -- and copyright seems to be the source of so much misinformation in the used book business.
yair25
OPAug 18, 2026 at 6:17 PMAll of it does. That's most of why it's interesting.
Translator. If the book doesn't name one, the field stays empty. It isn't inferred from the language or from any outside source — nothing goes in that field unless a printed line says so.
Year. If there's no edition or printing statement, no number line, and no stated publication date, then the only thing on the page may be the copyright date. In that case the tool records that the year came from copyright alone, flags it, and leaves the publication date blank — the description says “copyright date 1931; printing not stated.” That change came directly from your earlier reply. If there's no date at all, the field stays blank.
Place. It records the publisher/imprint locations actually printed in the book, not the printer's location. If several publisher locations are stated, it keeps them rather than choosing one. Printer information is treated separately. It doesn't decide which location is “correct,” because that's a judgement and it doesn't make judgements.
The rule underneath all three is the same: every field has to trace back to a literal line of text on the page. If it can't, the field stays empty and gets marked. An empty flagged field is a correct answer; a plausible-looking guess is not.
That's also why it deliberately stops short on books like the Hardy Boys examples you described. Those require bibliographic judgement from clues outside the straightforward printed metadata, and I'm not pretending to automate that.
What I'm more curious about now is the ordinary part of your inventory: books where you already read the title/copyright pages and type what's stated there into the listing. If Bench saves no time on ten of those, that's probably the most useful answer I could get.
1786davycrockett
·Aug 18, 2026 at 8:26 PM@yair25
Well, good luck to you, one way or the other!
Something that I have learned in the 60 plus years that I've been selling used books, magazines, comics, records, et al, is that there always seem to be new
surprises that nobody else had ever considered before; and I've been on the long end of bibliographic research to know when it's becoming too overwhelming.
Have at it!
yair25
OPAug 19, 2026 at 4:38 AMThank you. Sixty years of surprises is exactly the kind of thing I can’t build my way around, and it’s useful to hear it said plainly.
The copyright date change came from you and it’s in the product now. If another surprise turns up that you think I’ve got wrong, I’d genuinely like to hear it.
All the best,
Yair