User talk:Fixer/Archive/2011

Duplication

I accepted the submission adding this pub, but afterwards saw that another identical one was already in the db. (It's probably the record you accepted earlier in the day: 2011-01-15 / 02:37:12 / 1515837 - NewPub / Ahasuerus / Ahasuerus / Undead Hearts). Before deleting the record, I wanted to let you see it. I thought Fixer searched the db for ISBN and titles before making a submission. The new one did not come up as an "Add Pub" submission. And the ISBNs are identical. Thanks. Mhhutchins 17:24, 15 January 2011 (UTC)

This was an odd one. I was running Fixer late at night and the connection between ISFDB and the development server froze for a few minutes when this submission was in the pipeline. Eventually Fixer received a "timeout" error on his end, so I assumed that the submission didn't make it and resubmitted/approved the book manually. However, it looks like the original submission had been added to the queue after all, I just didn't notice it (apparently, I need new glasses.) I have deleted the duplicate -- thanks for catching it! Ahasuerus 21:05, 15 January 2011 (UTC)

Leftovers from the latest batch of submissions

I'm leaving six or seven submissions in the queue to let someone else decide whether to accept them. Their Amazon pages give almost no information about the content of the books, neither summaries nor reviews. Some titles appear to be spec-fic but titles can fool you. Mhhutchins 06:04, 20 January 2011 (UTC)

Unfortunately, these late bloomers are always problematic. Sometimes Amazon adds a record and then deletes it within a few weeks, other times they don't bother and the record is stuck in limbo for months if not years. Perhaps Bill will have additional ideas as he probably knows the UK scene better than us trans-atlantians! Ahasuerus 07:22, 20 January 2011 (UTC)

Pub missed ("Supernatural")

I entered this pub based on info in a review in the latest Locus. Amazon has it categorized under "Supernatural" but it appears Fixer missed it. Is he searching this category? Thanks. Mhhutchins 02:28, 1 February 2011 (UTC)

Nope, no "Supernatural" at this time. I think I tried it a couple of years ago and the results were mostly New Age-y. The rest had significant overlap with other SF subjects, so I didn't include it in Fixer's list. I'll try it again later this week and see what Fixer finds -- thanks for the heads up! Ahasuerus 06:11, 2 February 2011 (UTC)

Fixer has been re-run for 2010-2011 using "supernatural" as the subject header. Most of the new hits (as opposed to "late arrivals" that were not listed by Amazon when Fixer was last run) came from New Age/paranormal non-fiction, but there were a few SF books that would have been missed. I'll add it to the list of subjects that Fixer uses -- thanks for finding it! Ahasuerus 16:09, 18 February 2011 (UTC)

You're welcome. I supposed combining "supernatural" with "fiction" would weed out a lot of that New Age stuff. I've noticed a number of Christian fantasy titles from Fixer lately. Would those have been found if Amazon had placed them only in the Fiction category, without any subcategory like fantasy or science fiction? Or would "Christian > Fantasy" have been sufficient? Searching through Amazon just now I found this pub which is not in the database. Mhhutchins 18:04, 18 February 2011 (UTC)

Just found another one not in the database, and it's only categorized as "Religion & Spirituality > Fiction". It happens to be in a series for which we have a later book. That one was categorized as "Science Fiction & Fantasy > Fantasy" and was naturally found by Fixer. I can't imagine anyone, even Fixer, looking for spec-fic titles in "Religion & Spirituality > Fiction". This is obviously a failure on Amazon's part to properly categorize the titles. Mhhutchins 18:13, 18 February 2011 (UTC)

Sorry about the delay, I have been playing with Amazon's interface on and off. I checked my notes from 2008 and apparently I was unable to get searches for multiple subjects to work. I could get a search for "subject=fantasy and pubdate=2008-11" to work, but I when I tried, e.g., "subject=supernatural and subject=fiction and pubdate=2008-11", it didn't return what I expected. However, Amazon's internal logic changes frequently, so what was broken in 2008 may have been fixed in 2009 or 2010. I have tried similar searches over the last couple of weeks, but the results are somewhat inconsistent. I will have to try a few more things to see if I can discern a pattern. Ahasuerus 20:48, 28 February 2011 (UTC)

Thanks for trying. There's always going to be an obscure title by an obscure author from a obscure publisher. We're not going to find all of them. I only strive for completeness when it comes to the Big Name Authors, which probably accounts for 90% of database user searches anyway. BTW, when will you be releasing the library pubs that Fixer's found? Mhhutchins 20:54, 28 February 2011 (UTC)

I did some work on library data while I was away, but it's still not quite ready for prime time. The parsing algorithm does a decent job of extracting the most commonly used fields, e.g. if field 300, sub-field "c" says "18cm.", Fixer knows that it's a mass market paperback. If field 520, sub-field "a" reads "This epic novel, based on the gripping ten-part Sci Fi Channel miniseries from DreamWorks Television, spans sixty years of American history... --p. [4] of cover.", Fixer knows to stuff it in the Synopsis field of the submission record and so on. However, there are more obscure fields that Fixer doesn't know how to handle yet. Most importantly, Fixer is still Amazon-centric and needs reprogramming to help him handle books that are not listed by Amazon. That will take some time, but I will be once again on the road in late March and hope to get it to a point where it's at least semi-functional. Ahasuerus 02:36, 1 March 2011 (UTC)

Makeshift God

Please look at the ISBN of this submission: 2347736930. It's my understanding that ISBNs beginning with a "2" are French language books, so I knew something was amiss from the start. A little research shows we already have a record for 0234773693, only one number off from the Fixer submission, same author, same title, basically the same publisher. I was going to reject the submission but wanted you to see it first. Is there a way for Fixer to determine that certain ISBNs just shouldn't be submitted, especially those that begin with numbers other than "0" and "1". Or has the search for non-English language books began? Thanks. Mhhutchins 19:03, 7 February 2011 (UTC)

When Fixer runs and queries Amazon using the standard query process, he automatically rejects 999- and other obviously bogus ISBNs. He also suspends all non-0/1 ISBN and puts them in a separate "on hold until further notice" table. This is the same table that picture books for young children and other borderline pubs end up in.

However, the last run was experimental because I tried a new algorithm to find "notable" SF pubs not in our database. Since I wasn't sure what Fixer was going to find, I asked him to bypass the standard auto-rejection/auto-suspension process. As it turns out, it resulted in too many flaky ISBNs and I didn't catch some of them when I was reviewing Fixer's internal queue prior to submission (it was 3am). Lesson learned -- I won't ask Fixer to bypass the standard vetting process when experimenting with this algorithm again. Ahasuerus 01:33, 8 February 2011 (UTC)

Tor Books

Is there a way to change the publisher on any Fixer submissions based on Amazon records from "Tor Books" to simply "Tor"? We have merged the two publishers, giving the preference to "Tor" (I can't recall any book published by Tom Doherty stating the publisher as "Tor Books" on the title page). See this submission which show "Tor Books" as an unknown publisher, at least until someone makes another submission using it. Also, look at the ISBN of this submission: 5556085269. Obviously not in the Tor range (or anybody else's for that matter). It may have been a temporary number until final publication, which I've seen quite often with Tor, and have made an effort to remove them from the database. Can Fixer be programmed to ignore such numbers? Thanks. Mhhutchins 19:10, 7 February 2011 (UTC)

Agreed, if WE have decided on a Canonical publisher name then Amazon's choice shouldn't affect ours and it would help if Fixer auto-corrected them for us. There's always a risk that we won't get the "Unknown Publisher" warning for other "Tor Books" additions between approval and subsequent correction. I think I've rejected some other 5-prefix-ISBN Fixer submissions as they came from obviously non-Russian publishers. BLongley 00:37, 8 February 2011 (UTC)

Yes, it's possible to tell Fixer to convert "Tor Books" to "Tor" prior to submission. The only challenge is to make sure that the conversion occurs in the right place(s) since I allow overrides of "forced" transformations at various points. Ahasuerus 03:19, 8 February 2011 (UTC)

While I'm at it - many of the Fixer-submitted Amazon-sourced entries for Borgo Press were 1999 Wildside Press editions that had Copyright or First Edition dates on instead of real ones. A 1971 Borgo Press title rung alarm bells with me - as did 1992 e-books editions. I did try and sort out Publishers at one point, if we can use the data I'd be happier. BLongley 00:37, 8 February 2011 (UTC)

Yes, I saw that and left a note on the Moderator Noticeboard about Borgo Press. It's one of the problems with using Amazon's data for older books -- the older the pub, the flakier the data. We are probably better off getting older data from library catalogs (I have 10 Gb worth of stuff sitting on the hard drive), but that presents other challenges. Ahasuerus 03:19, 8 February 2011 (UTC)

If I provided a list of dodgy publishers e.g. "Littlehampton Book Services" are listed for a lot of historical Amazon UK data, or dodgy dates - e.g. Borgo Press definitely didn't publish in 1971 - would you use those to provide automated warnings? BLongley 01:06, 9 February 2011 (UTC)

If the number of dodgy publishers and suspect permutations is not too high, then I can add these checks to the scripts that handle moderator pages. It, OTOH, it requires additional tables, then it will take longer to implement. Either way, if you could create a feature request and list the suspects there, it would be a great start. I am having an increasingly hard time keeping track of everything that needs to be done because so many changes depend on other, often subtle, changes... Ahasuerus 04:09, 9 February 2011 (UTC)

Another missing Amazon category?

I just added Toys from the James Patterson writing corps. I suppose the only relative category that Amazon places it in is "genetic engineering" (other than "thrillers" which I'm sure Fixer wouldn't look for titles.) Should Fixer be looking for this combination of categories? Mhhutchins 22:02, 9 February 2011 (UTC)

I guess I can ask Fixer to search for things like:

"subject:genetic engineering" AND "subject:fiction"
"subject:supernatural" AND "subject:fiction"

This approach works well when Fixer interrogates library catalogs, but I don't think I have tried it with Amazon, which uses a totally different set of commands. We'll see how it goes ;-) Ahasuerus 23:11, 9 February 2011 (UTC)

ebook format

Fixer submitted a Vintage Digital publication with format "e-book" instead of the "ebook" called for by the help. --MartyD 11:24, 14 March 2011 (UTC)

Fixed, thanks! Ahasuerus 20:31, 14 March 2011 (UTC)

Kindle and no Nook?

Mr. Fixer, are you planning on submitting Kindle ebooks from the amazon website? I might note that this would have a tendency to give Amazon a competitive advantage over Barnes & Nobles incompatible Nook ebooks. I would say a pox on robot entered ebook titles from both sources until there is a common format. Barnes & Noble Nook ebooks can actually be read on devices other than the Nook while the Kindle DRM ebooks can only be read on a Kindle or a PC.--swfritter 22:06, 14 March 2011 (UTC)

I suspect Fixer can't distinguish DRM from non-DRM. Diane Duane has suggested we might look at her non-DRM Kindle books, and I admit I'm not going to enter them myself although I own a Kindle. We're busy enough anyway. BLongley 01:53, 15 March 2011 (UTC)

The biggest technical problem with Kindle is that Fixer relies on ISBNs and many Kindle books do not have ISBNs. It make sense financially since ISBNs cost money, but it's a headache for the poor 'bot.

Policy-wise, we still haven't decided whether we accept electronic books without ISBNs. I am becoming convinced that we should, especially now that Amanda Hocking has shown how many people read them, but that's fodder for the Policy page. Ahasuerus 02:35, 15 March 2011 (UTC)

Consider: A) Our policy discussion process is mostly broken; B) There are far too many significant ebook works without ISBN's and C) I, and others, have put a fair number of non-ISBN ebooks into the the system. The policy has been changed by common practice although I think a human should be making the decisions in these cases.--swfritter 13:37, 15 March 2011 (UTC)

AddPub errors

It seems you're often confused when a series and title are present in the publication name. For instance, today you wanted to Add this pub to this title, when of course it should go under this one. Is it possible that you could look at the series that an existing title is in to help improve the determination of which part is the series and which is the title? We have a few other examples: e.g. it looks like a Moderator has let through some extra titles into this that should go there. BLongley 19:23, 4 April 2011 (UTC)

Fixer and I tried to brainstorm this issues a couple of weeks ago. Unfortunately, no bright ideas emerged at the end of the session. The underlying problem is that Amazon has a bewildering number of ways to add series info to the title field. Sometimes they embed the series name and the superseries name using nested parentheses, but there is no rhyme or reason to it. The most we can realistically do is to try to do a better job of identifying SERIES: TITLE situations. It's not much, but it's better than nothing. Ahasuerus 04:53, 18 April 2011 (UTC)

That's why I was thinking we could do something with the ISFDB series, if we have one. BTW, there was another oddity today: "Phantastes: A Faerie Romance for Men and Women" was put under "Phantastes" although we have the full title already present. BLongley 14:50, 18 April 2011 (UTC)

It looks like Fixer checks whether the "ABC" in ABC: XYZ matches an existing title before checking whether the whole string matches an existing title. Silly robot! Ahasuerus 01:36, 19 April 2011 (UTC)

Better a Silly Robot than an Angry Robot. (Actually, I'm not sure which causes more problems.) BLongley 02:17, 19 April 2011 (UTC)

I've had a few problems with AddPubs and series designations, too. One particular one that gave me a headache was for The Dead Town, which was titled Dean Koontz's Frankenstein: The Dead Town. Fixer linked it to the SFBC omnibus Dean Koontz's Frankenstein. Took me several submissions to fix the problem after accepting the Fixer submission, unmerging, adding a new title record, merging with the correct title, etc. Mhhutchins 02:31, 19 April 2011 (UTC)

The Secret Kingdom by Jenny Nimmo

I'm going to accept this submission but will make some changes, so I wanted you to know before I do. The first thing that popped out was the price. When I looked up the ISBN range, I saw that except for one title they were all audio books released by Random House. Then the price made sense. We have a title record for The Secret Kingdom with only a UK pub. Scholastic has published books in the US by Nimmo before. A Google search brought up this listing on Publishers Weekly website for the printed book. Did Fixer miss this listing on Amazon, or is that in the next bunch of June submissions? I'm going to go ahead and create a record for the book based on the PW and Amazon listings, and change the current submission to an audiobook from Random House. Mhhutchins 16:07, 15 April 2011 (UTC)

Thanks for taking care of this unruly ISBN! Normally, Fixer auto-suspends audio-books, but in this case there was no explicit indication whether it was a printed book or an audio one. I should have noticed the unusually high price and the lack of a page count, but it was probably late at night... Ahasuerus 00:27, 17 April 2011 (UTC)

P.S. And yes, Chronicles of the Red King: The Secret Kingdom was in the next batch of Fixer's submissions. Ahasuerus 01:49, 17 April 2011 (UTC)

Crown of the Conqueror

Not sure how to handle this. The price is strange, and it's not listed on Amazon.co.uk's site, so where did Fixer get it? Also there's another edition coming soon after this from the same publisher one with a normal price, but with an entirely different ISBN. Mhhutchins 21:47, 17 April 2011 (UTC)

A price like "£4.93" typically means that Fixer erroneously grabbed the UK price rather than the US price. However, Amazon.com doesn't list this ISBN any more. The only source that thinks that this ISBN is valid for the US edition is Fantastic Fiction, so it looks like this ISBN has been canceled and replaced with another ISBN. I will change the date to 8888-00-00, thanks! Ahasuerus 03:59, 18 April 2011 (UTC)

I checked and there's a few more 9780062021 prefixes from Angry Robot that appear to be vapourware. That prefix looks as if it actually belongs to Avon. BLongley 14:31, 18 April 2011 (UTC)

I must look into fixing advanced price searches. ISBN prefix checks will take a LOT longer. :-( BLongley 02:32, 19 April 2011 (UTC)

US Publishers on Amazon UK

I think you can safely assume that "Egmont USA" and "Bloomsbury U.S.A. Children's Books" data would be better-sourced from Amazon US than Amazon UK. BLongley 20:26, 6 September 2011 (UTC)

Here's a record that I just accepted from a Fixer submission. Should I remove the UK price and add the US price of $16.99 (from Amazon.com)? Mhhutchins 20:48, 6 September 2011 (UTC)

I would. £10.50 is not a standard UK price, except maybe for some Library Editions. BLongley 21:12, 6 September 2011 (UTC)

Done. I'm also going back to other Bloomsbury USA books and finding the US price on Amazon.com. BTW, we have a lot more books as "Bloomsbury USA Children's Books" (without the periods in U.S.A.), so I'm converting those otherwise. Mhhutchins 21:15, 6 September 2011 (UTC)

Thanks, I was just about to say I would also consider merging "Bloomsbury U.S.A. Children's Books" and "Bloomsbury USA Children's Books" but I'll leave that to a US Mod. I'll look into "Bloomsbury Children's" and "Bloomsbury Children's Books (UK)" from the UK point of view. BLongley 21:17, 6 September 2011 (UTC)

A US book submitted with a UK price (or vice versa) often means that the US publisher had canceled the ISBN and it was subsequently removed from Amazon's US database, but their UK database hasn't been updated yet. There are other permutations that can lead to the same scenario, but more often than not these cases require additional digging. Ahasuerus 22:55, 6 September 2011 (UTC)

So does Amazon.co.uk convert the original US price to UK pounds which is the reason why we get such odd prices on some of these? Look at the submissions I have on hold. Mhhutchins 20:38, 8 September 2011 (UTC)

I think that happens in both directions. You can usually tell by looking up the same ISBN on the other Amazon, and if you see a "normal" price, that's likely the right one. Amazon.com is often nice enough to add the telltale "[import]" as corroboration. --MartyD 00:40, 9 September 2011 (UTC)

It's not just bi-directional. Look at "Prospero Brule" - the only way that make sense is if the language is French and the price is €11.50 - sourced from Amazon.FR. BLongley 15:52, 9 September 2011 (UTC)

Normally, French ISBNs start with 2 and Fixer auto-suspends them, but it looks like this is an exception. Ahasuerus 20:53, 9 September 2011 (UTC)

I asked one of experts about it User_talk:Hauck#The_Horus_Heresy.2C_T15_:_Prospero_Brule but it seems Hervé's preferred option is to delete it. I don't know if that's just to keep things simple, or because we haven't got language support up to the right level yet, or if he just doesn't like Warhammer or Dan Abnett. :-/ But we really could do without submissions that take hours of discussions between Mods - OR we need Fixer to do better at sourcing. :-( BLongley 23:38, 9 September 2011 (UTC)

All four options at once ;-). Hauck 08:32, 30 October 2011 (UTC)

Unfortunately, I don't think Fixer can do much about non-English books whose ISBN starts with 0- or 1-. There are quite a few Spanish language books published in the US and Fixer dutifully reports them every month. I then have to suspend them manually before I create submissions... Ahasuerus 03:59, 10 September 2011 (UTC)

Stoke Books

I think the current queue suggests that Fixer needs to learn how to do CHAPTERBOOK submissions. :-/ BLongley 00:39, 30 October 2011 (UTC)

The thought has crossed my mind, but the current de facto standard is that juvenile books that are just under 100 pages long are entered as Novels to avoid clutter (which we may want to formalize at some point.) That makes it hard to distinguish between 96 page novellas for the adult market and 96 page "novels" for kids. However, I should definitely teach Fixer how to submit the under-70 crowd as chapterbooks: there have been a lot of 1940s-1950s reprints lately. Ahasuerus 05:07, 30 October 2011 (UTC)

Agreed on the formalising of juvenile "Novel" rules. But I wouldn't make this a priority - indeed, I would suggest that the language improvements are far more important, even if I'm not someone that will actually use them much. We have active European moderators now, and active editors feeling a little frustrated, so that's where I'd direct my efforts. (And in fact, have done, but I'm waiting on feedback on the last few improvements. Sorry if I'm reminding you of the bottlenecks, I feel like I am one too.) BLongley 02:12, 14 November 2011 (UTC)

Ebook submission, 4 months early???

Why is Fixer making a submission for an ebook that hasn't been priced and won't even be available for four months? Mhhutchins 01:09, 14 November 2011 (UTC)

I have to agree: 4 months in advance is usually too much, especially with Amazon data. And we don't seem to like ebooks much anyway: I've corrected a lot of "Angry Robot" e-publications but in hindsight would prefer not to have had them here in the first place. We're overworked anyway and I would much prefer you (Ahasuerus, not Fixer) catch up on the language support software improvements rather than give us more dodgy Fixer submissions. I know we probably need them if we're going to remain current, and the recent Amazon changes have made things more complex, but please consider reducing the workload for us mere humans. BLongley 01:38, 14 November 2011 (UTC)

Sorry, folks, it was unintentional. As you said, Bill, the new process is considerably more complex and it involves moving data from queue to queue to file to table to queue etc. This particular ISBN escaped into the wild before it could be properly handled.

As far as implementing language-specific enhancements goes, I just want to make sure that the new Fixer process works successfully from A to Z before I go back to Python development. It has taken longer than expected -- and I have repeatedly underestimated the amount of work that was still outstanding at different junctures -- but we are quite close and I just want to get this monkey off my back. Ahasuerus 03:56, 14 November 2011 (UTC)

Another misidentified publisher

Over the past year there's been a few dozen records submitted by Fixer for a publisher identified by Amazon (and most other catalogs) as "DP". It appears to be solely trade paperback editions, most of which are reprints of mass-market editions published by Leisure Books (an imprint of Dorchester). Well, "DP" really doesn't exist. The Amazon "Look-Inside" feature reveals that every book in the database under the "DP" publisher should have been Dorchester Publishing. The problem occurs because of the logo used by Dorchester which someone has misidentified as the name of the imprint. (Look at the title page and copyright page of this title). Looking at their website, it even seems that the company identifies "Dorchester Publishing" as an imprint of the eponymous company that also publishes the Leisure Books and Love Spell imprints. Now to my question: is it possible to program Fixer to automatically change future submissions from "DP" to "Dorchester Publishing" without too much jumping through hoops? Thanks for considering it. Mhhutchins 02:32, 22 November 2011 (UTC)

Done -- it took less than 2 minutes to make the change. Wish they were all that easy...

Which reminds me... But I think I'll post it on the Moderator Noticeboard. Ahasuerus 02:42, 22 November 2011 (UTC)

Fast work, indeed. Thanks. Another thing: whats the latest about the Fixer submissions from Amazon.co.uk for what's obviously US editions, but have UK pricing? I corrected a few of the Dorchester Publishing records, which were in UK pounds, to US dollars, which I got by going to their Amazon.com link. Mhhutchins 03:33, 22 November 2011 (UTC)

Fixer uses a special algorithm to determine whether an ISBN's data should be imported from Amazon.com or from Amazon UK. If the publisher:

• is obscure, or
• has presence on both sides of the Atlantic (e.g. Tor), or
• the two stores can't agree on the name of the publisher, then

Fixer lists both sets of data and asks me to pick the store at submission creation time. Sometimes I pick the wrong one. Other times Amazon's data is bad and Fixer picks the wrong store on his own. I keep trying to improve the algorithm, but there are always ways for things to go wrong... Ahasuerus 04:22, 22 November 2011 (UTC)

Pubs that didn't merge titles

There were two submissions for pubs, both by the same author (Mary Downing Hahn), that had exact matches with two titles already in the db: Time for Andrew: A Ghost Story and A Doll in the Garden: A Ghost Story. For some unknown reason, they were submitted as new pubs and titles, and not as just new pubs, requiring me to merge them with the existing title record. This hasn't happened in a long while, so I'm wondering what may have happened that caused fixer to be unable to match the titles. Thanks for looking. Mhhutchins 06:33, 27 November 2011 (UTC)

It's the old "colon" problem again. I thought I fixed it a while back, but it turns out that the fix was flawed. I have made further changes and it should be working now, but please keep in mind that Fixer tries to err on the side of caution, so if there are multiple identical (or near-identical) titles on file, they won't be auto-merged. Ahasuerus 06:59, 27 November 2011 (UTC)