Hey nighttime Knitters!
Blog stuff
Yesterday I blogged about Knit for the first time. It’s an interesting feeling to publish something you’ve put a lot of energy into and be able to see the reactions in realtime. The best part is probably just having done it; the next time will definitely be easier.
E-mail sign ups just barely broke double digits. The tweet got picked up enough that my account got some new follows (some aren’t even bots!) including a VC. One Hacker News upvote :)
I’m happy to have included a survey (suggested by a reviewer like you!), because I got to see this feedback:
Beneath the iceberg stuff
The blog post took a lot longer to put together than I had planned. It turns out I had to learn a bunch of new things. Which is great! In case you decide to launch your TikTok for dogs app idea soon, here were some things I wish I knew.
Video editing
A lot of times you can just record a screencast or Loom. But this can be tricky if you’re trying to choreograph a more complicated demo, or if you want to edit your content to be as short as possible. I picked up a little bit of Kdenlive, which is an NLE (non-linear editor). It’s actually pretty cool once you work out the kinks.
Analytics
Whoops I should’ve put in Google Analytics. Mainly this would’ve been helpful for tracking referrers. I have some page analytics from Cloudflare, but they’re more privacy conscious (so less useful).
At least I uploaded the video to YouTube, because I wanted to see video engagement metrics. It’s nice!
Forms
The main goal was to collect e-mail signups, but I also wanted to ask optional survey questions. I still wanted the e-mail signup inline in the blog post, and couldn’t find any way to do that with Typeform, Mailchimp or Webflow. I got pretty close to what I wanted by using a Google Form with a pre-filled form answer.
If you want your web page to look good in a tweet, it needs OpenGraph metadata. I set this up afterwards, and it does not apply retroactively.
RSS feeds
It maybe wasn’t worth the effort to set up initially because no one uses RSS. Fiddling with metadata can be a little annoying, in part because online RSS aggregators like Feedly cache bad metadata indefinitely.
Static site generation
Substack (how you’re reading this) is a pretty good starting point, but I wanted a more customizable blog. Static sites are super easy to serve and debug. Hugo is easy and lightweight, but takes some effort if your site doesn’t exactly fit the template structure.
Hosting
Cloudflare Pages is great for static sites. It has a 25 MB file size limit, so I’m actually not using it right now (the drafts served the video file directly instead of from YouTube).
Future stuff
Being back on Twitter is probably bad for mental health, but it’s a pretty good medium for distributing new ideas. We’ll see how this all plays out. I’ll probably set aside some time every morning and evening for lighthearted doom scrolling.
Tentatively the next milestone is a live demo site. Need to figure out some basic security stuff, and fix up a bunch of hacks so it’s usable.
I’m thinking of keeping the blog active in the ballpark of an article a month. Someone described this type of content like “a pitch deck for a new way of working with data.” Some potential working titles:
- Everything, Everything, All at Once…with data
- What is a data flow?
- DAGs should use Merkle trees
- The moral failures of public data
Nitey nite!