One question for readers: for the gwern.net master version, would it be effective to present it in 2 versions, the first version with zero links, and then immediately afterwards, the version with all the original links? Or would people miss the point and only read the link-less version and not bother with the referenced version?
What I’m thinking is less about the distraction factor and something along the lines of helping defamiliarization by providing first an experience where the reader thinks repeatedly “that is super fake and made up technobabble gish galloping, gwern, nothing remotely like that does or even could exist, just making stuff up doesn’t make a good hard takeoff story” and then on the second time through, repeatedly goes “huh, that’s weird. oh, I missed that paper, interesting… I hadn’t thought about this one like that. yeah, that one is a good point too. Hm.” But of course that depends on seeing the second version and checking the occasional link (or annotation, more accurately), which I think I might greatly overestimate the probability of such dedicated readers.
I would naively expect something like a 10:1 ratio of skimmers-to-double-readers, though perhaps you have a better UI in mind than I e.g. if you had a cool button on-screen called “Toggle Citations” then reading and toggling it to predict which things were cited could be fun. Of course that 10:1 doesn’t include weighting by how much you care about the readers. It’s on-the-table that the few people who “get to be surprised” are worth a bunch of people not seeing the second version.
Thinking more, I actually quite like the idea of “Here’s the story” followed by “AND NOW FOR THE SAME STORY AGAIN, BUT WITH AN INCREDIBLE NUMBER OF CITATIONS AND ANNOTATIONS”. That sounds like it could be fun.
I didn’t have a toggle in mind (although I’m sure Said Achmiz could whip some JS up if I really wanted to do that), because with toggles it’s even harder to get readers to realize it’s there & use it. While before/after is extremely obvious and transparent if the reader wants to read both versions at all. Perhaps side by side in two-columns? We don’t have much two-column text layout support (just multi-column lists) but that might be a nice feature to implement regardless.
And yeah, that’s the question here and why I’m asking: how much loss do people think is acceptable for the gain of the one-two punch? And how big of a gain does it sound like it would be? It’s just an idea I had while thinking about how the story works, I’m not committed to it.
Right. I’d have to do a few user tests to feel confident (e.g. send the two pages to different people with a Google form for asking who got more out of it). But I’ve personally changed my mind and now think the 1-2 punch sounds really fun to read. So I change my vote to double!
After discussing a bit with Said (two-column layout: not too hard; table layout: very easy but bad idea; toggle with JS: harder but doable) about possible paradigms relating to the toggle hiding/showing links, I came up with the idea of ‘reader mode’ (loosely inspired by web browser reader-modes & plugins). We want to hide the links, but keep them accessible, but also not require toggles to make it work because first-time readers will approximately never use any features that require them to opt-in & a toggle would be tantamount to always showing links or never showing links. How to square this circle?
In reader-mode, most of the default gwern.net UI would be invisible/transparent: link underlines/icons, footnotes, sidebar, metadata block, footer, anything marked up with a new hideable class, until the user hovered (or long-pressed on mobile) over a hidden element and they would be rendered again. Reader-mode is disabled per-page when the user scrolls to the end of the page, under the presumption that now all that metadata may be useful as the reader goes back to reread specific parts. Reader-mode would cater to readers who complain that gwern.net is just too much of a muchness for them and they need to disable JS to make it bearable; other readers might enable it on pages like the ebook pages like Notenki Memoirs where there is a lot of useful annotation going on, but the reader may just want to read the plain text without distraction as if it were a print book.
With such a reader-mode implemented, then the story page can reuse the reader-mode functionality for a subtle one-two punch effect by adding a small special-case feature, such as a CSS class which triggers reader-mode by default (rather than being opt-in as usual). With that class now set on the Clippy story page, the reader loads the page and sees just a normal plain unhyperlinked SF story—until they hit the end of the page (presumably having read the whole story), at which point reader-mode terminates permanently and suddenly all of the links become visible (and now they suddenly realize that the technobabble was all real and can go back and reread). Or, if they are curious enough while reading, they can hover over terms to decloak the hidden formatting and reveal the full set of links. (They will probably do so accidentally, and that is how first-time reader-mode users will discover it is enabled.)
And then at the end, I can also throw in a link to the ‘link bibliography’, which is a page just of all of the links + annotations, intended to let you read through a page at the link level instead of having to hover over every single link. (I don’t think many people ever bother to use that particular gwern.net feature, but that’s at least partially due to the link bibliography being in the metadata block, and as we all know, no one ever reads the metadata.)
This seems a good deal more elegant than copy-pasting versions, and would be useful on a few other pages, and address a small but voluble contingent of readers’ minimalism need.
(I don’t think many people ever bother to use that particular gwern.net feature, but that’s at least partially due to the link bibliography being in the metadata block, and as we all know, no one ever reads the metadata.)
I don’t have any idea whether people use that feature or not, but I definitely love it. One of my fav things about browsing gwern.net.
I was directed to the story of Clippy from elsewhere (rabbit hole from the Gary Marcus vs SSC debate) and was pleasantly surprised with the reader mode (I had not read gwern.net for months). Then, I came here for discussion and stumbled upon this thread explaining your reasoning for the reader mode. This is great! It’s a really useful feature and incidentally, I used it exactly the way you envisioned users would.
FWIW I never read anything on your site because the links bug the hell out of me. They wouldn’t be so bothersome if it weren’t for the in-frame hover pop-up.
One question for readers: for the gwern.net master version, would it be effective to present it in 2 versions, the first version with zero links, and then immediately afterwards, the version with all the original links? Or would people miss the point and only read the link-less version and not bother with the referenced version?
Links don’t hurt it for me, mostly they help and make it feel more grounded! So I vote for single.
What I’m thinking is less about the distraction factor and something along the lines of helping defamiliarization by providing first an experience where the reader thinks repeatedly “that is super fake and made up technobabble gish galloping, gwern, nothing remotely like that does or even could exist, just making stuff up doesn’t make a good hard takeoff story” and then on the second time through, repeatedly goes “huh, that’s weird. oh, I missed that paper, interesting… I hadn’t thought about this one like that. yeah, that one is a good point too. Hm.” But of course that depends on seeing the second version and checking the occasional link (or annotation, more accurately), which I think I might greatly overestimate the probability of such dedicated readers.
I would naively expect something like a 10:1 ratio of skimmers-to-double-readers, though perhaps you have a better UI in mind than I e.g. if you had a cool button on-screen called “Toggle Citations” then reading and toggling it to predict which things were cited could be fun. Of course that 10:1 doesn’t include weighting by how much you care about the readers. It’s on-the-table that the few people who “get to be surprised” are worth a bunch of people not seeing the second version.
Thinking more, I actually quite like the idea of “Here’s the story” followed by “AND NOW FOR THE SAME STORY AGAIN, BUT WITH AN INCREDIBLE NUMBER OF CITATIONS AND ANNOTATIONS”. That sounds like it could be fun.
I didn’t have a toggle in mind (although I’m sure Said Achmiz could whip some JS up if I really wanted to do that), because with toggles it’s even harder to get readers to realize it’s there & use it. While before/after is extremely obvious and transparent if the reader wants to read both versions at all. Perhaps side by side in two-columns? We don’t have much two-column text layout support (just multi-column lists) but that might be a nice feature to implement regardless.
And yeah, that’s the question here and why I’m asking: how much loss do people think is acceptable for the gain of the one-two punch? And how big of a gain does it sound like it would be? It’s just an idea I had while thinking about how the story works, I’m not committed to it.
Right. I’d have to do a few user tests to feel confident (e.g. send the two pages to different people with a Google form for asking who got more out of it). But I’ve personally changed my mind and now think the 1-2 punch sounds really fun to read. So I change my vote to double!
After discussing a bit with Said (two-column layout: not too hard; table layout: very easy but bad idea; toggle with JS: harder but doable) about possible paradigms relating to the toggle hiding/showing links, I came up with the idea of ‘reader mode’ (loosely inspired by web browser reader-modes & plugins). We want to hide the links, but keep them accessible, but also not require toggles to make it work because first-time readers will approximately never use any features that require them to opt-in & a toggle would be tantamount to always showing links or never showing links. How to square this circle?
In reader-mode, most of the default gwern.net UI would be invisible/transparent: link underlines/icons, footnotes, sidebar, metadata block, footer, anything marked up with a new
hideableclass, until the user hovered (or long-pressed on mobile) over a hidden element and they would be rendered again. Reader-mode is disabled per-page when the user scrolls to the end of the page, under the presumption that now all that metadata may be useful as the reader goes back to reread specific parts. Reader-mode would cater to readers who complain that gwern.net is just too much of a muchness for them and they need to disable JS to make it bearable; other readers might enable it on pages like the ebook pages like Notenki Memoirs where there is a lot of useful annotation going on, but the reader may just want to read the plain text without distraction as if it were a print book.With such a reader-mode implemented, then the story page can reuse the reader-mode functionality for a subtle one-two punch effect by adding a small special-case feature, such as a CSS class which triggers reader-mode by default (rather than being opt-in as usual). With that class now set on the Clippy story page, the reader loads the page and sees just a normal plain unhyperlinked SF story—until they hit the end of the page (presumably having read the whole story), at which point reader-mode terminates permanently and suddenly all of the links become visible (and now they suddenly realize that the technobabble was all real and can go back and reread). Or, if they are curious enough while reading, they can hover over terms to decloak the hidden formatting and reveal the full set of links. (They will probably do so accidentally, and that is how first-time reader-mode users will discover it is enabled.)
And then at the end, I can also throw in a link to the ‘link bibliography’, which is a page just of all of the links + annotations, intended to let you read through a page at the link level instead of having to hover over every single link. (I don’t think many people ever bother to use that particular gwern.net feature, but that’s at least partially due to the link bibliography being in the metadata block, and as we all know, no one ever reads the metadata.)
This seems a good deal more elegant than copy-pasting versions, and would be useful on a few other pages, and address a small but voluble contingent of readers’ minimalism need.
I don’t have any idea whether people use that feature or not, but I definitely love it. One of my fav things about browsing gwern.net.
I was directed to the story of Clippy from elsewhere (rabbit hole from the Gary Marcus vs SSC debate) and was pleasantly surprised with the reader mode (I had not read gwern.net for months). Then, I came here for discussion and stumbled upon this thread explaining your reasoning for the reader mode. This is great! It’s a really useful feature and incidentally, I used it exactly the way you envisioned users would.
/sheds tears of joy that someone actually uses the link-bibliographies and noticed the reader mode
FWIW I never read anything on your site because the links bug the hell out of me. They wouldn’t be so bothersome if it weren’t for the in-frame hover pop-up.
You can disable those, you know… just click on the little “eye” icon in the top-right of any popup.
Thank you!