"Where to start" function

It would be cool, and hopefully useful, if you on an author page could get the works sorted by your "Will you like it?" scores.

For a random example, if I go to C. S. Lewis's author page and start clicking on the works and checking "Will you like it?", I'll find that LT thinks I'm somewhat more likely to enjoy Out of the Silent Planet than The Lion, the Witch, and the Wardrobe. It'd been neat if I could find out this directly from the author page, along with where they rank among all the rest of Lewis' books (that are widely enough catalogued to have WYLI scores, of course).

(A possible refinement could be to prevent latter volumes in series from being ranked above earlier ones; it'd be silly to suggest to anyone to start with The Two Towers even if the algorithm thinks they're more likely to enjoy it than The Fellowship of the Ring. Ideally a toggle I guess; there's far less reason to read say Discworld in any particular order.)

I would like that too.

CS Lewis has over 400 works. Of those many - probably at least a quarter have over 100 copies. How much computer effort would it take to compare a hundred books or more to your library all at once?

If you have entered older science fiction, but not children's books, yes it probably thinks you will like Out of the Silent Planet better than LWW. It gives me the same score on both. I love the Narnia books, and find the Space Trilogy unreadable.

There's a problem I see with this request. The "Would you like it" score is only calculated when you click on it. At the feature's creation, Tim said it took too many cycles away from other important functions to make it a regularly calculated, stored value.

This would mean going to the author page, clicking on the change of sort then waiting while it went through and calculated all those values, then did the sort. For an author like CS Lewis, with over 400 works, you'd have a bit of a wait. Authors with 5 works might not be bad.

Perhaps if Tim revisited the Would You Like It feature to change the way it calculates?

There's also the problem that the measure doesn't actually take into account your or other member's book ratings so it doesn't actually say anything meaningful about whether you would like that book.

>5 jjwilson61:

I disagree, obviously. Let's not rehash the argument about whether ratings should be taken into account, it's been had enough times already; but please accept some of us do find it useful as is.

>3 MarthaJeanne:, >4 gilroy:

I realize it's likely to be computationally expensive, but I figure it's better to wish for a pony and perhaps not get it than to keep quiet and certainly not get it.

>6 AndreasJ: I just wanted to make sure that one of the limitations was mentioned, for when others read the thread.

(Someone mentioned Internal Chronological Order v. Publication Order in another thread, which of course would be a problem for the "refinement" - should it, to stick with the C. S. Lewis example, force PO on the Narnia books, or the ICO? In this particular example the latter strikes me as obviously silly, but others, IIRC Tim included, feel differently. Both currently exist as LT series, and the choice which to force would presumably have to be somehow made algorithmically.)

andreasJ (#8):

the choice which to force would presumably have to be somehow made algorithmically

Nope. The choice which to force was made by Tim, who chose ICO.

>9 lorax:

Please reread what I wrote. The choice cannot have been made yet, because the suggested functionality does not exist yet.

Tim made that choice for Series already, there's no reason he wouldn't apply it anywhere that Series are used.

>11 lorax:

As I'm pretty sure you know, there are two series for the Narnia books, one ICO, one PO. The function, if implemented, would have to somehow chose which to use.

Now maybe Tim would, if he implements this RSI including the refinement, try and algorithmically determine which if any of the series shared by a pair of books is an ICO one, but that would require something more than an preference for ICO. You could add a flag to series that say what they're organized by, or try and guess from series name (quick: find me every permutation of "publication order" used in any language on LT), but you can't just apply what's already done for series, because, well, he hasn't actually enforced ICO in any way, we can and do make series organized in other manners.

(I can offhand think of three widespread orderings used in LT series - PO, ICO, and offical numbering - but I'm sure there's others.)

>12 AndreasJ: Those two series should have been folded into one with the new series system since the new system allows a user to toggle between original publication and another order within a single series. Tim has made the other order, internal chronology presumably, the default, so it stands to reason he'd use the same default for other uses of series.

We are not presently combining publication order and chronological order

So you're keeping a publication order series which has a story/publication switch which when you select story will keep the order by publication date. Sounds confusing and I doubt it's the way that Tim intended it to be used.

>15 jjwilson61: It's been recommended to hold off on combining separate listings for Chronological/Publication order because the publication sort is still buggy.

https://wiki.librarything.com/index.php/New_Series_-_best_practices

You can find links to the discussions about this topic there, too :) .

>15 jjwilson61:

Tim's given us a series relationship "Is a reordering of", so it does seem to be intended.

Could we stop discussing series in this thread, though? The basic suggestion here doesn't depend on series, and there's plenty of other conversations on series around on Talk.