Will you like will you like it?

So, does anyone use the "Will you like it?" feature on work pages? Example:



I find it fun and not inaccurate, but opinions clearly vary, and space is at a premium. What do you think? Do you use it? Is it accurate? Is it a fun treat, or a dumb waste of space?

I do use it from time to time. I've found it more accurate than not, and I am entertained by it.

I use it sometimes and it’s fun.

I never noticed it before. And it doesn’t interest me. The couple I just checked were inaccurate, too.

I use it occasionally and find it mildly accurate but not very helpful.

I use it sometimes, it’s fun and fairly accurate, and I have used it to help me decide whether to add a title to my TBR when I’m on the fence.

I don't use it. Checking books that I've recently read, it's not terribly accurate.

I find it entertaining, but I am not entirely sure how much stock I put in it. I'd be sad to see it go, though.

I wasn’t aware it existed so won’t be bothered if it goes away. I never trust that kind of thing as I’ve never found them accurate (I take most people’s recommendations with a grain of salt, too).

I like it and would be sad to see it go, but it's not a feature I can't live without.

I use it occasionally, but it's only moderately accurate.

Never seen it before. Checking some recently added books, apparently I won't like them - confidence very low. What's the algorithm based on?

I've consulted it occasionally. It's fun, but I don't find it very reliable.

I use it when canvassing for new books to read, but usually the answer is that I will like it, with a low confidence. So maybe not that useful.

I use it sometimes, and it's more reliable than not. I wouldn't be bothered if it went, though.

It tends to be fairly accurate when there is other information to help make a decision - ie when you don't need it. When that other information is missing it is very hit or miss. Maybe more miss.

I have looked at it quite a bit, but only 'used' it if it agreed with what I already thought. If it disagreed with my personal opinion I just ignored it.

I hadn't used it until now. So I checked about twenty books as a sample -- and it didn't give me an opinion on any of them. Roughly half the books I checked had the message "There are too few copies to analyze whether you will like it." That's in a quite random sample; I would expect that to be about the rate for my whole library. But, to repeat, even the ones that didn't have that message didn't produce a rating.

Unless there is a setting to turn it on, it's pretty useless for someone with a library as obscure as mine. Which doesn't surprise me much. :-)

//Correction to my original post: I see that I have to click on it to get the recommendation -- though I still have that 50% "too few copies" problem. Which makes it not very useful, because my library is getting more obscure by the day. :-) Based on the results that I did get, though, I can make an observation: I value a book based on three criteria: subject matter, quality of the writing, and level of scholarship such as footnotes and bibliography; a good book requires all three. Example: James IV by Norman MacDougall: Good subject, excellent scholarship, reads like a lump of lead. I am forced to slog through it, but once I'm done, I will happily never look at it again. Yet the algorithm expects I will love it. Based on my revisited sample, the "Will You Like It" algorithm appears to be good at subject matter, no good at level of scholarship, no good at quality of writing. Not sure what can be done about that, but the subject of a book is the easiest thing to learn about; the problem is to determine level of scholarship and quality of writing. The algorthim is only telling me the part I already know. So I guess my revised opinion is that it's a fun toy but not to be trusted. I kinda like it even so, but only as a toy, not as an actual recommendation engine.//

I was aware of the feature but have not used it. Looking at a few recent purchases it is reasonably accurate but I really won’t miss it if it goes.

I use it quite often. I don't really trust it, though.

I've not used it in ... years. I found it funny, fun, and inaccurate.
No real loss if it goes away

It's always nice to have a suggestion, so that you can do the complete opposite.

It doesn’t really give enough information to use it as a basis for a decision. If it said “you won’t like this, because we know you have a prejudice against books by people who put their academic qualifications on the front cover,” or “you’ll love this, it’s full of juicy sex-scenes,” it would be disturbingly intrusive, but a source of useful advice.

As it is, we don’t know whether it’s telling us we will or won’t like something on reasonable grounds, or just because of stray stuff in our libraries we haven’t looked at since we were teenagers.

I like the idea of it, but in practice I haven't used it really at all.

Come to think of it, I haven't used it in such a long time I can't even find the feature now.

I used to use it occasionally but it was never accurate except in the most obvious cases, so I haven't used it in years.

I'll join the it's fun, but not useful group.

I won't miss it. In my recollection of the incidental use I made of it years ago the feature was fairly inaccurate and/or useless ("too few copies"), and I haven't bothered with it since.

The terminology it uses seems odd to me: With "very high confidence", you "probably will like" this book.

>30 cpg: There's basically two axes:

1. Will you like the book?
2. How confident are we?

In theory, we could plot it on a x-y grid, but it seemed fussy.

>31 timspalding: In theory, we could plot it on a x-y grid, but it seemed fussy.

I don't think there is any problem with the terminology; "confidence" is an easy concept, and so is the scale of how well one likes the book. The only question is, is the result accurate? If it isn't, it's just another horoscope. :-p

And it does seem as if a lot of people find the results inaccurate. The flip side being that we probably remember errors better than we remember places where the prediction is accurate.

Perhaps this calls for an actual experiment: Pick some people and have them examine, one at a time, their results for (say) their last forty books, and see what the correlation is.

Of course, in my case, the correlation is with "not enough copies." :-)

Much more useful is the LibraryThing Recommendations module. I use that all the time.

>33 2wonderY: Much more useful is the LibraryThing Recommendations module. I use that all the time.

You obviously don't get an average of about 135 recommendations per week, about 90% of them based on books you don't like much. :-)

I think I've used it a few times, but only for a laugh. I trust my own judgment and take risks with books that are edge cases. If it disappears, I won't miss it.

>34 waltzmn: I’m referring the the one on the work page.

>31 timspalding:

Yes, but saying "probably" instead of "a little" conflates the two axes.

The pretty diagram is very confusing because the arrow always seems to be at half or lower, even when the result is "probably will like The Grand Sophy (prediction confidence: very high)".

>1 timspalding: I have never used it before this day, but I am charmed and delighted by it now that I'm paying attention.

Mine shows as a plain gray bar until you click the "Will you like it?" link, which I didn't realize until now I needed to do, so I've spent the past however many years thinking it just didn't ever have enough data for the books I was looking at....

>36 2wonderY: I’m referring the the one on the work page.

OK. That one isn't very good for me either. But, again, that's because I have a strange library and, seemingly, strange standards. (E.g. I don't think one should recommend a nineteenth century history book to someone who is reading an eighteenth century history book.)

>40 melannen: Mine shows as a plain gray bar until you click the "Will you like it?" link, which I didn't realize until now I needed to do, so I've spent the past however many years thinking it just didn't ever have enough data for the books I was looking at....

This was my problem too. But I still have about 50% of my books that it doesn't have data for. :-)

The curiosity is that, if it doesn't have enough data, it tells you that, but if it doesn't, it shows nothing. That's weird enough to approach the status of bug: Either hide it every time or show it every time.

>32 waltzmn: I went back through some of my most recently read books, and ... it's not good. It really feels like a caricature of me versus a tight genre. I've got a lot of 70s sci-fi, so I will love Bloodhype (actual opinion, 3, 3.5 stars.) I don't have as much modern sci-fi, so I will probably like The Atrocity Archives (actual opinion, 4,4.5 stars.) Caught Dead in Philadelphia is 80s mystery, so love, and Glory in Death may be mystery & sci-fi, but it's also romance so probably like (despite the fact I have 3 reads recorded for it and for most of the 50 book series). It's worse for my dip into Japanese manga, which got "probably won't like" or flat "won't like", for a bunch of books I tended to find fun. Doctor Galaxy got a "will love", despite the fact it's probably my most disliked book of the year on my read list. (I don't tend to finish books that aren't fun, so not completely damning, but a portrayal of humanity that I find implausible that lets the bad guys win is not fun.)

>42 waltzmn: I still have a few with no data, but way more than I was expecting.

It is very weird, though. I understand if they don't want to take time to do the calculation if most people don't need the data, but the "Will you like it?" link definitely parses for me as "click for more info on the feature" not "click to make the feature load in". Even changing it to something more button-like would help a lot.

I just pulled up a book from my library that I gave 5*'s to and this is what I got:

LibraryThing thinks you probably will like The Invention of Nature: Alexander von Humboldt's New World (prediction confidence: very high)

Probably so.

I've never seen it, but that's probably because it's been years since I've done anything in my LT catalog. And that's not going to change anytime soon, either.

It entertains me. Is often accurate; sometimes amusingly & wildly inaccurate.

ETA: Wow! I just tested 10 of my top-rated recent reads & the algorithm predicted some form of liking for all of them. One book I rated 4.5 stars, the algorithm predicted I’d love — I didn’t even know that was a category! My respect for that little bar has grown.

I use it frequently , almost always. Some the time it is very accurate, sometimes not. But mainly I do find it is accurate. I like it. I would miss it.

>34 waltzmn: I feel the same way about recommendations. I rarely look at that.