I’ve started toying around with setting up my story at its permanent website, and have put a test page of the first chapter at this page (NSFW due to image of tasteful female nudity). I find myself faced with all sorts of options—font? line width? dark on light or light on dark? inline style or CSS? spot colors? hyphens or em-dashes? straight or curly quotes? etc, etc? - and I don’t have a lot of evidence to base any answers on.
As a preliminary set of answers, I’ve drawn on Butterick’s Practical Typography, even though I don’t have any particular reasons to favour that set of advice over any other, other than that it’s a concentrated dose of a /lot/ of advice. I don’t know how to set up formatting for multiple viewing devices, I’ve never touched CSS, and my budget for fonts or professional advice is pretty much zero.
Does anyone reading this know where I can find evidence that any changes I could make to my preliminary formatting would do any better than what I already have?
You could just answer it experimentally with A/B tests measuring time-on-page or leaving a review or something like that.
I see you’re running off Apache on Dreamhost, so there’s no doubt plenty of libraries to help you there, but there’s other strategies: static sites can, with some effort, hook into Google Analytics to record the effect of an experiment, which is the approach I’ve used on gwern.net since I didn’t want to manage a host like Dreamhost.
If I knew when I started what I know now, I would have begun with a large multifactorial experiment of the tweaks I tested one by one as I thought of them. It sounds like you are aware of the options you want to test, so you have it easy in that regard. (With the Abalytics approach, testing all those options simultaneously would be a major pain in the ass and probably hamper page loads since all possibilities need to be specified in advance in the HTML source, but I suspect any Apache library for A/B testing would make it much more painless to run many concurrent interventions.)
It’ll probably take a few thousand visits before you have an idea of the larger effects, but that just means you need to leave the test running for a few months.
Formatting stories—any good evidence?
I’ve started toying around with setting up my story at its permanent website, and have put a test page of the first chapter at this page (NSFW due to image of tasteful female nudity). I find myself faced with all sorts of options—font? line width? dark on light or light on dark? inline style or CSS? spot colors? hyphens or em-dashes? straight or curly quotes? etc, etc? - and I don’t have a lot of evidence to base any answers on.
As a preliminary set of answers, I’ve drawn on Butterick’s Practical Typography, even though I don’t have any particular reasons to favour that set of advice over any other, other than that it’s a concentrated dose of a /lot/ of advice. I don’t know how to set up formatting for multiple viewing devices, I’ve never touched CSS, and my budget for fonts or professional advice is pretty much zero.
Does anyone reading this know where I can find evidence that any changes I could make to my preliminary formatting would do any better than what I already have?
You could just answer it experimentally with A/B tests measuring time-on-page or leaving a review or something like that.
I see you’re running off Apache on Dreamhost, so there’s no doubt plenty of libraries to help you there, but there’s other strategies: static sites can, with some effort, hook into Google Analytics to record the effect of an experiment, which is the approach I’ve used on gwern.net since I didn’t want to manage a host like Dreamhost.
If I knew when I started what I know now, I would have begun with a large multifactorial experiment of the tweaks I tested one by one as I thought of them. It sounds like you are aware of the options you want to test, so you have it easy in that regard. (With the Abalytics approach, testing all those options simultaneously would be a major pain in the ass and probably hamper page loads since all possibilities need to be specified in advance in the HTML source, but I suspect any Apache library for A/B testing would make it much more painless to run many concurrent interventions.)
It’ll probably take a few thousand visits before you have an idea of the larger effects, but that just means you need to leave the test running for a few months.