Small combining changes

I've made some changes to the combination process.

1. The combination page now shows which split a work belongs to, if the author is split.
2. That process now works. If you combined A and B, and A is the winner, it won't change the split A is assigned to. It was doing this before, quite wrongly.
3. A warning is giving if your combinations cross authors.
4. Another warning is given if your combination crosses author splits. (See http://www.librarything.com/topic/241967#5854053 for context.)
5. It looks a little different, with real links for the works.
6. The combination page is now always organized with the "winner" on top. I'm not calling this out, because the title and even author name of the "winner" can change, and people wouldn't necessarily understand that.

Apologies for the fuzzy:


I decided against Collectorator's suggestion in http://www.librarything.com/topic/241967#5852891 (explanation there) but this should at least help the problem they identified.

Excellent! I haven't seen it in action yet, but I'm looking forward to it.

Any chance you could also check when combing for works that are related by work-to-work relationships? E.g. provide a warning - or total block - if one work is "Contained in" the other.

(I'd prefer a total block, preventing the combination. If combining is really the right thing to do, then the relationship is wrong, and should be fixed first.)

Okay, so the rule is if any work is contained within any other work on the list, block?

If there's any work-to-work relationship that involves both works. Abridgement of, Study guide to, etc.

Hmmm. So if there's any work-to-work relationship where one book on the list points to another book on the list?

I feel like there must be an exception. But I'll start in.

I don't think so - I've just looked at the list, and none of the available relationships should apply in cases where the works could reasonably be combined. In every case, the relationship exists because there is some difference between the works. "is exactly the same as" and "has the same text as" aren't on the list :-)

Someone could perhaps incorrectly put a relationship between a work and an illustrated edition of the same work, but that's easy to get around by just removing the relationship.

Another way to look at it is - would any of the relationships make sense after combination? You'd end up with "This work" is some relation to "This work".

It should just block, and then if they really need to be combined the relationship is easily removed.

We also need a block that prevents the combining of a split author page whole hog into another author page. I do not know how they do that, but they do. I just had to separate Pliny from some other bigger name Plinio somebody.

Snort. Well, I guess Plinio is a nice unique name. Isn't that the idea?

What do you mean, exactly, though. Do you mean using the combine feature on works, or aliasing an author somewhere, or what?

See http://www.librarything.com/topic/244705 . More changes.

>8 Collectorator:

We also need a block that prevents the combining of a split author page whole hog into another author page. I do not know how they do that, but they do.

Easy as pie.

Here's Michael Smith's page*. Notice that "Improve this Author" suggests combining it with Michael S. Smith. All I'd have to do to combine them is click on "combine". (On top of that, Michael S. Smith is a split author, so I could combine a split author page with another split author page - and I've seen that happen.)

*Don't know why I can't touchstone that.

>12 lilithcat:

So I think what we need is for author combining to go to an intermediate page, not happen right when you click. That is, it should work like work combination.

Does this sound good?

My second thought is that it, when we have such a page, a number of situations can be flagged as warnings or show-stoppers, as happens on work pages now.

So what should be flagged?

>15 Collectorator:

I'm responding to her, but not talking to her only.

I suggest flagging author combinations when one or both have a canonical name set? These canonical names are often used (wrongly) as a mask.

Okay, good.

This would be in addition to warning whenever combining with an author that is split.

Okay, but just a warning--there's nothing wrong with it per se.

I need to look at all the variations of what happens when you combine splits, splits with aliasing, etc.

>14 timspalding: what we need is for author combining to go to an intermediate page

Sounds great - showing both system calculated name and canonical name for both authors?

Tim, you may characterize this as "Small combining changes" but for me these are the most exciting changes to LibraryThing in a long time. Thank you!

Another suggestion, if I may: 99 times out of 100 I will "never" suggested author combinations. Could that block just be removed from the page? Users who know what they're doing will know what to do without the suggestions, and users who don't know what they're doing shouldn't be encouraged to do what is almost always the wrong thing.

Another suggestion, if I may: 99 times out of 100 I will "never" suggested author combinations. Could that block just be removed from the page? Users who know what they're doing will know what to do without the suggestions, and users who don't know what they're doing shouldn't be encouraged to do what is almost always the wrong thing.

You will never them, or never never them? You're saying author combination should not be a normal, public option?

The suggested author combinations are almost always really bad ideas. Just get rid of these suggestions. Author combination should be a normal option, but LT should not be recommending really bad combinations.

>18 timspalding: You will never them, or never never them? You're saying author combination should not be a normal, public option?

I almost always "never" the suggestions.

I overspoke when I said "Could that block just be removed from the page?" It's just the list of suggestions that should be removed; the option to search for authors names to combine with should still be there.

Perhaps something like this, with the search box and button always visible, and the search box pre-filled with the author's name.



Or perhaps that's still too much encouragement. Maybe still have the box and button normally hidden, and clicking the text makes them visible, like it does currently.

(The original version of this message was posted on my phone as I was going out the door.)

Thank you so much for these changes!

Also echoing >17 r.orrison:, >21 Collectorator: and >22 Collectorator:

These are all like late Christmas presents! Thank you.

#10

''Snort. Well, I guess Plinio is a nice unique name. Isn't that the idea?''

I don't understand. I thought different language versions of authors' names get combined as a matter of fact. Moreover, publishing conventions in different countries (often in Europe) commonly use only the last name or most famous name (Flaubert, Cicero...)

This excessive author name splitting is likely to wreak havoc (I think has already, going by what I thought I was seeing in my catalogue recently) especially in classical literature.

As for author combining suggestions, I just looked at Lucius Annaeus Seneca (not a single Cyrillic or Greek name version included until I combined two Russian entries) and almost all suggestions were spot on, with these arguable exceptions:

Seneca. Apocolocyntosis (combine)
Lucius Seneca (combine)
senecatheyoungerluci (combine)
Seneque (combine)
Sénèque (combine).

And even those, as I mentioned, probably more often than not refer to the more popular author.

>25 LolaWalser: I just looked at Lucius Annaeus Seneca ... and almost all suggestions were spot on

Interesting. Using the Random Author feature to look at a couple dozen randomly selected authors, the suggestions I've gotten just now are:

• On the Paul E. Eden page, for "Paul Eden" and "Eden Paul" - but Paul E. Eden already has Paul Eden (1) aliased in, so shouldn't be combined. (Side note: those two names are already combined with each other, so there's little point listing them both.)


• On the Frances Hodgson Burnett page, for "L. Frank Baum L.M. Frances Hodgson Montgomery" a hodge-podge mixup of multiple authors in one entry, which shouldn't be combined.


• On the Annette Smith page, suggestions for "Randell", "Beverely Randell", "Beverley Randell", and "Beverley et al Randell"


• On the William R. Ferris page, suggestions for "Wilson reagan Charles", "Mary L. Hart", "Charles Reagan Wilson", and "ed. Charles Reagan Wilson"


• On the Edward Marston page, suggestions for "M. Jakubowski", "ed. Maxim Jakubowski", "Miriam G. Monfredo", "Miriam Grace Monfredo", "Sharan Miriam Grace and Newman Monfredo, eds.", and "Sharan Newman"

There were plenty of pages with no suggestions, but not a good one in the lot. (I haven't nevered them in case anyone wants to look.) I can't deny there are good suggestions sometimes, but most that I see are bad.

#26

I suspect all this is neither here nor there, but of my first ten favourite authors listed only Lewis Carroll and Lenny Bruce currently have not-their-name suggestions--'Bracken Books' for Carroll, two 'other people' names for Bruce.

>28 LolaWalser:
Do any have valid suggestions?

Tim, when you get a chance please explain why split Plinio from Pliny, or Seneca from L. Annaeus Seneca etc. and how are we supposed to get all the editions on the same combine-separate page if 'last name' author name versions are split from other versions.

Right now there's this obnoxious message half the time I want to combine his works:

'Combine list crosses authors. Make sure you're doing the right thing.'

Crosses authors? No, it doesn't:

Moral Letters to Lucilius, Vol. 1 by Seneca (1) — 2 copies
Moral letters to Lucilius Volume 1 by Lucius Annaeus Seneca — 1 copy

#29

Valid suggestions for what? I asked questions and I'm waiting for Tim to explain stuff to me.

You said that "but of my first ten favourite authors listed only Lewis Carroll and Lenny Bruce currently have not-their-name suggestions" - I was asking if any of those authors had valid combination suggestions.

#32

No, those three (one for Carroll, two for Bruce) were the ONLY suggestions. See here:

Lewis Carroll

Lenny Bruce

I agree that the "combine with" suggested authors should be removed. I was just thinking yesterday that I haven't seen a valid suggestion in that section for years.

Much to say here--probably in multiple messages.

The suggested author combinations are almost always really bad ideas

I would say, however, that the combination suggestions have become less useful because so many useful ones have been done. Maybe it's outgrown its usefulness as a default option.

I agree that the "combine with" suggested authors should be removed. I was just thinking yesterday that I haven't seen a valid suggestion in that section for years.

I'd want to spend some time looking at this. We all look at certain sorts of books and not others. I'd worry, for example, about books by new authors.

I overspoke when I said "Could that block just be removed from the page?" It's just the list of suggestions that should be removed; the option to search for authors names to combine with should still be there.

How about if we moved all of this to a secondary page?

Pliny and Seneca are split authors and so cannot have dissimilar names combined into them.

"I thought different language versions of authors' names get combined as a matter of fact." Sure, unless there is a SPLIT AUTHOR involved.

Yeah, this is the wrong principle that's been invented, and which has caused the plague of aliasing between splits, to splits and all round.

There is no reason why split authors shouldn't be combined into. You combine and you split--it's simple.

Tim, when you get a chance please explain why split Plinio from Pliny, or Seneca from L. Annaeus Seneca etc. and how are we supposed to get all the editions on the same combine-separate page if 'last name' author name versions are split from other versions.

No, I think you're entirely right. Just because the books under Plinio are all, say, Pliny the Younger, is no argument against combining Plinio into Pliny, even though Pliny is split into Pliny the Younger, Pliny the Elder and go-knows-who-else went by the name of Pliny. The rule that splitting an author should shut down combination is silly, and yields complexities we don't need.

Seneca. Apocolocyntosis (combine)
Lucius Seneca (combine)
senecatheyoungerluci (combine)
Seneque (combine)
Sénèque (combine).

All good suggestions, I suspect.

Pliny and Seneca are split authors and so cannot have dissimilar names combined into them.

They can. The system allows it, and it's the best way to handle these situations.

We all look at certain sorts of books and not others. I'd worry, for example, about books by new authors.

That's why I looked at random authors. Actually, I think new authors are particularly bad. They tend to get suggestions for collaborators or editors.

>36 timspalding: How about if we moved all of this to a secondary page?

Good idea!

>38 Collectorator: In this example Michael S Smith (1) & (3) are shown as Michael S Smith on their books' covers. Michael S Smith (2) is shown as Michael Stephen Smith. The difficulty of deciding for sure how an author wishes to be known has already been noted above but I am content to regard a cover name as the best guess in most circumstances. So, for Michael Smith (6), why not edit the primary author to reflect these different forms and pull all the three authors' works off the Michael Smith page? It looks like all three authors - or maybe their publishers with or without their agreement - wanted to differentiate themselves from the other Michael Smiths. Maybe they would all answer to Mike Smith in the pub but so what?

I've done a lot of nevering. If an author appears regularly in anthologies, you get a long list of editors and other contributors as suggestions. Right now, there are five editors suggested to combine with George R.R. Martin. One can only imagine how many have been nevered already.

In addition, the nevering process requires a lot of clicks and page reloads, so if there is a long list of suggestions, it's very tedious.

>41 PhaedraB:

the nevering process requires a lot of clicks and page reloads, so if there is a long list of suggestions, it's very tedious.

It's extremely bad when you are nevering suggestions on a split author page, because you keep getting sent back to the disambiguation page.

Okay, I need help:

Last call: Work-combination improvements
http://www.librarything.com/topic/244878

#36

Thanks, Tim. There are a bunch of authors, then, whose ''other'' (i.e. non-English) versions of names have been faultily split (or never combined), between various languages.

>45 LolaWalser:

Clearly I need to get to the bottom of the dogma of "don't combine into split authors."

I wrote elsewhere: "There are other areas which we could raise - I've seen people discussing the 'Strathmeyer' (sp?) syndicate where many authors publish under a single name. There are also organisations that publish lots, but have many contributing editors. Is there a case for a level above author? (And is that what disambiguation pages are, in some regard?) "

So, disambiguation pages (and their splits) represent hierarchies of name-forms (where a name-form is a collection of name-strings). E.G. Michael Smith > Michael S. Smith > Michael Steven Smith. At any one lowest level, it helps to see the birth/death dates by the side of that split. If a split showing on a disambig page references a level that is also split, can that be shown on the higher page?

Or, can the whole of the hierarchy be shown on a single page? In that instance, all the works that are unassigned across all levels can be shown. Picking up the show/hide feature I suggested for work combination, that could be used to reduce complexity. Also - and here I'm stretching a bit - could you drag/drop from unassigned into the correct bucket?

Also, when splitting an author which has CK, can you choose which split the CK gets re-assigned to?

>47 Noisy: "...when splitting an author which has CK, can you choose which split the CK gets re-assigned to?"

OMG yes, a thousand times yes.

>48 PhaedraB:

Yeah, but, strictly speaking there's no such thing as splitting authors. You just separate out individual names on by one. I guess the option could be "do you want to take the CK with the split-off?" or something.

>That would work. The cut'n'paste CK option is especially tedious if someone has entered loads of stuff for one of the authors on the page.