/dan/ danish example football loads_and_does_save loads_but_cant_save soccer statisticsII
This is a statistics notes for a class. It's real, so I need this tool to work right.
But I'm going to talk about the tool just as much as I talk about my obsessive need for good grammar when I type. And there will be some actual statistics work too.

The main point is to type the notes and the stats. I want to rely on the tool. So a bit of insurance is required. For one thing, the changes you make save into your local store as you type, so a freakout by the browser won't cost you more than a moment.
The Save button has a fault; you want to push command-s. And that ain't right. So it might be better to replace it with a file icon with options about how the server is communicated with in terms of saving. The default needs to be to upload automatically, of course. The server can return a number indicating how long it would like you to wait before your next autosave. You're not supposed to really consider the work to be physically limited to your own computer anyway if you ask me, and the best tool is leeeaaan.

I wouldn't crowd the user with features, escpecially menus called Features. The templates gear should take care of that. I'd switch the search to after the lock and html. A feature for typing like 'enter box' and 'leave box': really <box> and </box>. Sometimes though pressing 'return' should get you out of a single line tag, like H1 or BUTTON. And the math, when it is indisputablely so, should be done in front of you with reasoning that helps.

<h>chapter 17

A population would have a true percentage that has a true property (which of these balls are properly inflated? which of these voters will cast a ballot?)
We call the proportion p, out of the true population. If the population is perfectly split it is 0.50, or if it's one-yes-to-three-no then p is 0.25
If you sample a truly random subset of the proportion, you're going to get a sample (of size n) but it's not going to be exactly the same value most of the time. It will differ by a certain amount, which is measured in units of deviance. Your sample landing within one deviance away of the real amount happens 68% of the time. Landing within two deviances away of the real amount (which includes everything within one deviance) happens 96% of the time.
After that point, a sample starts to provide strong evidence that you weren't even right about the true population in the first place. Your assumption can still be correct. It's not even correct to say it's "less likely" since it's an assumption.

<box>Having typed the beginning of the box, I'd like it to start boxin' my text! And a return char at the end, with no other specific context, ought to break me out. But right now I have to use the box button...<box>

This box works! And safari is good about letting me edit it. But not perfect! Deleting the last line even seems to drag the line after the box into it, styles and all. And I lost the style on the initial bit of html, which kills the situation dead anyway. It's fair to say that "edit the shit directly" mode sucks. But the individually-managed-sectors approach showed a lot of promise, as long as the selection is respected, and the sectors are properly maintained. Augh! How do I get OUT of the box now?? There's no ending tag!</box></box> no good and the contenteditbale doesn't let me either. Oh no, I copied and pasted the box and it deboxed the text. Let's try more: