Open sgoodreau opened 4 years ago
Wow, these are GREAT comments!
OK, I trimmed the language down, removed some of the apologies and side notes.
I also added a bold main point on the network tab. The suggestion was to have it not on either tab but at the end as a summary point, but I couldn't figure out how to do that. And I actually think this works well, because you get those points and more when you read the full text tab. But it guides those who are only looking at the network straight to the point we want them to see, in common language.
Everything else will have to wait for another day, or, in many cases, for the research-level version.
OK, I trimmed the language down, removed some of the apologies and side notes.
I also added a bold main point on the network tab. The suggestion was to have it not on either tab but at the end as a summary point, but I couldn't figure out how to do that. And I actually think this works well, because you get those points and more when you read the full text tab. But it guides those who are only looking at the network straight to the point we want them to see, in common language.
Everything else will have to wait for another day, or, in many cases, for the research-level version.
The summaries are great, but it might be worth trimming the top text to allow the whole tab to render without scrolling:
1st section
the 6th section
OK, I trimmed the text in the Good Ol' Days - how does that look? (in the repos SocNetDist.html, not online yet)
Hard to do for Section 6, since the text is already so short.
In the end I think it's hard to guarantee that any particular amount of text fits on one screen across all of the different platforms and screensizes. Probably the best thing that could be done is to make the plots smaller? But that would have downsides as well. Thoughts?
OK the main formatting points in the thread are implemented. The rest are feature additions that will probably have to wait for funding or the next pandemic.
Steve,
This is super-impressive. Especially impressive that you came up with this while working I'm sure more than full time on epidemiologorific analyses. You do a great job of introducing the model and gradually adding complexity until you can make your point.
I have a few comments that I hope are taken as constructive.
1) A tiny UI point: I was actually not clear that I needed to click on 2.2 to get the explanation. I think being a bit more obvious the first time would be helpful.
2) The description of 3-paths is great. And it's great that you say a higher number of 3-paths makes it more likely there's transmission. I feel like it would be helpful in one of those paragraphs to make it more clear: even if you are SUPER-careful about your interactions with person B, you could end up getting the virus from person B' in a different 3-path instead. So if you have ten paths, or a HUNDRED, even if you're careful, your risk goes way up.
3) The idea of having a quick text intro for each chapter followed by a network picture, and then a (let's admit it) optional text explanation, is cool. It means some of your readers may go through the doc not reading the explanations, but they could still get something from the intros and pictures. Was that a constituency you were aiming at? That said, I feel like they ought to get a sentence or two of conclusion text on the main page (i.e. not in a subchapter) too. Like; "Let us be very clear here: some people are going to get infected, and some people are going to die, because of these connections. It’s that simple. But these connections are so essential that we as a society are willing to make that trade-off. Without them, many more people would die of other basic things besides COVID-19: starvation, freezing, crime, other diseases."
4) In your bury-the-lede email, you wrote, "It involves a lot more text than I was first imagining. I guess that's always the way." This is one of the greatest struggles with in my career: I've always got so much super-useful stuff to say! I've gradually learned - though I certainly haven't won the battle (Just look at my emails to the roommates.) that shorter can often be better. So even if draft 1 is pretty long, I would encourage you to try to shorten it even a few percent for draft 2, even though it's really hard to do. That's a general comment, with some specific notes below.
5) I guess it depends on your audience. Clearly, the buzzfeed folks won't get past the first paragraph, if even that far. But I don't think you're aiming for them. I think you're aiming for people who're at least a bit intellectually curious. Still, most people, espeicially teens, will balk at walls of text. Perhaps just by splitting 4.2 into one or two more chunks, it would look less intimidating. And if people decide to skip one of the subchapters, they can still get your main point. Especially if you put a concluding sentence on each chapter.
6) Someone on FB suggested that you try simpler language. I could imagine a version of this where you didn't even mention R and modeling and stuff, and removed a few of the harder conclusions. It could work, but I don't know how easy it would be to do. Maybe you'd need to have the home page let the user choose between "Just give me the summary!" and "I'm curious! Can you explain?" I guess this again is about your audience and your goal. If your goal is to convince people they should study epidemiology in college, then this is great. If you're trying to show random teens around the world some pictures to explain why they should stop going to their friend's house, it might be a bit too highbrow. But who has the energy to write two versions of this?!
7) Relating to the "different audiences" thing, I did feel that there were a few places where you were either explaining epi jargon or justifying the model. It seemed to me that you were doing the latter because you're a careful scientist, and you don't want your paper to get rejected. But I don't feel like it's necessary here, and it could obscure the main message a bit.
7) You mentioned that multiple 3-paths are more dangerous than single ones. But I think it would be really cool to SHOW that. For example, you could have a network where you show all the 3-paths, and then when you click a button you say, "Let's say the chance of passing sickness along each edge is just 1 in 20, because people are careful. How many people get sick?" and then you run your model and light up in red all the folks who end up infected. Then click a different button at 1 in 50, to say, "Even if you're REALLY careful, with enough 3-paths, a lot of people still get sick."
8) Even cooler - but possibly WAY more work - would be to have a slider where people can choose how many people are leaving the house, chance of infection along each edge, etc. and then you could run a simulation, or even run more than one with different seeds, and see how you do. I hear all the cool kids are doing great things with R Shiny these days. But maybe you have a day job.
Oh look: way more text than I originally planned. Sorry!
We told Noam a couple nights ago that he shouldn't visit his girlfriend any more, even though none of her siblings were seeing anyone. It was a tough conversation. Thanks for giving us the nudge to do it.
Thanks for putting in a lot of effort to educate laypeople and maybe even reduce those transmissions a bit!