Wow thanks for the reply. I noticed it mainly (possibly exclusively) occurs when the navbar comes into view, or goes out of view (so scroll up, release,then scroll down). It gets annoying because the nav bar reappears if you switch tabs, or leave chrome and come back later.
Typekit understandably wants to make sure that no one else uses the font bundles, so they check the referrer. If you strip referrers it'll block the response and eventually it'll timeout, and show using the default system font.
The choice is a delay for the few people who block referrers or a flash of unstyled content for everyone else.
"You retain ownership of all intellectual property rights in your User Content. Medium (and/or other third parties) retains ownership of intellectual property rights in all Content other than User Content."
Here's the Firefox Developer Tools output for Net, on a page reload immediately after the original page load (which seems to cut down drastically on requests due to caching which makes it easier for me to copy/paste the only 3 lines)
[02:27:36.182] GET https://medium.com/ [HTTP/1.1 200 OK 538ms]
[02:27:40.346] GET https://use.typekit.net/c/ec2d65/ff-tisa-web-pro:i4:i7:n4:n7,freight-sans-pro:n5:n7.Vdl:K:2,Vdq:K:2,Vdk:K:2,Vdp:K:2,TJD:K:2,TJJ:K:2/d?3bb2a6e53c9684ffdc9a98f7195b2a621af3706c43fd27fb1c2ac857895f9c1aa6e0a150c34a58a8cc268e093140e6a1c68d195f68c7ab5421d37c9f8037314acfccd341c5549953c3a5458d2153cf8598355c0169dde8a5e929e0abaef018efc3172b52ac4ce90d2e52930933d272c500a79829545c0b6ce81c15e104d6e141cf6a16eac595caa6382b0e1d176e5adc9bf2c5223ee5a452e892ca4cc2ddd59f9a101d5108046fb341684043b4aaa0808e0352746b2dd67d4d08453cb5a64ec80e36d899fbceb49c43ffc6ad03f5d251dc23736866 [HTTP/1.1 403 Forbidden 12ms]
[02:27:44.154] POST https://medium.com/_/stat/ [HTTP/1.1 200 OK 41ms]
Bingo. That's the problem. Sorry, I figured starting Firefox in safe mode reverted my custom config. I had disabled referrer via about:config's network.http.sendrefererheader. Changing that value back to '2' (send referrer headers always) eliminates the problem. I guess only amongst HN readers can you count on people not sending this header. Anyhow, if TypeKit could not require this, that would be great. Thanks for looking into this, even though it might only help .001% of users.
Thanks. My comment got some upvotes so I figure others are having the same problem. It loads perfectly in Chrome and Safari though. You might share your results with the Mozilla community if you figure it out.