Just sent out some updates to the new Classic editor (which I’m using to compose this post). It includes:
- Image width fix
- Support for horizontal rules and blockquotes
- Parsing fixes for Markdown images
- Layout fix when editing posts
I’ll bring those updates over to WriteFreely next.
Finished the manual moderation tools I mentioned yesterday, and I'm pretty happy with the result.
I tested it out on a queue of about 160 posts from the last 24 hours, and it went smoothly, once I switched from my original design (quick form submissions and a
JOIN with the main
posts table — bad idea, it's too big) to async requests and denormalized data.
I'm thinking about what this would look like in #WriteFreely, and I'm not yet sure if exposing all of our spam detection logic will make it easier for spammers to work around our roadblocks. I don't know if they put that much effort into it. But I know instance admins would be very welcoming to a feature like this. So I'm thinking of bringing it over soon.
Working on some moderation tools today. As I've written about before, we see a ton of spammy SEO / link-building content published every day, and have had to put some automated filters in place to keep it at bay.
Besides these loose rules that take automated action on users and posts, we also have a manual side, where a second tier of suspicious-looking posts are emailed to us to review and optionally take action on. But these reports come in through individual emails and make moderation very, very tedious.
So to speed things up, I'd like to add a post moderation queue on the admin dashboard, to make it easier to watch for widespread spammy behavior and quickly stop it. Then, as I know WriteFreely instances see plenty of spammers too, I'll port it over there once it works well.
In the beginning, Remark.as was meant to be an answer to “How do I interact with people on my blog?” I'm wondering now if we expand its scope to include information about who you're interacting with.
I didn't want it to merely become a “comment management” platform, where people comment on your posts, and you display those comments publicly. I wanted more conversation and more social context. Those goals alone mean more potential paths for this product.
For more conversation and context, I primarily intended to provide more communication channels, for private replies and public commentary. We're going to dip our toes into this shortly, with the added ability for Write.as email subscribers to reply directly to your emails (private replies). This will be a very minimally-viable way to communicate today, but it'll open the gates to richer interactions down the road, perhaps facilitated by a web application instead of your email client, for example.
What kinds of interactions will people have? How will this feature be used? I'm not sure yet; I think we need that functionality in place, out in the wild, before we have answers.
Working on giving authors access to their email list, a frequent feature request I hear. I'm thinking about this from a few different angles:
- Seeing who is reading my work
- Keeping in touch with my readers
- Knowing I can take my list elsewhere if I ever need to
It'll start off bare-bones, and we'll add only what's needed. As a bonus, this will lead to visibility into your fediverse followers, too.
I just fixed a long-standing bug that prevented people from following blogs from a Mastodon instance with Authorized Fetch enabled. The fix is deployed on Write.as, and will be merged into #WriteFreely soon. Write.as also sends out
Notes instead of
Articles in some cases. You should see that for this post, if you're following @email@example.com in the fediverse!
Working on a “welcome” email for one of our ongoing migration projects. I've always seen these as unnecessary, from my limited viewpoint and personal anti-automated-email bias. But over the years I've heard from a few people that expected to get an email like this after signing up — if anything, just to remember their username, and that they created an account in the first place.
Now that I'm implementing it and writing the email in the helpful way I would love to see as the receiver, I'm coming around to the idea. We'll roll it out for some very specific signups in the near term, and then expand it to all new signups if it works out well.
People who are good at designing and building incredibly complex systems are bad at recognizing their complexity and abstraction for what they are. If they did, they might be able to break them back down into simple devices.
A lot of things are coming together at the moment.
The CRM-ish thing I've been mulling over for a few months likely won't come out as a brand new product in the near future, but as an extension to one side to Write.as: interacting with your readers / subscribers. Instead of building a product solely for me, hoping for wider utility, it'll be built for everyone, and I'll happen to need it, too.
We're also going all-in on our multi-user side of Write.as. Unlike with our attempt at “Write.as for Teams” (a completely new moonshot product), this is a gentle, natural evolution from our current product. You won't have to reconfigure the way you write, or set up a new writing environment tied to your hosting configuration. You'll just write — but now with other people.
With this, we'll also be adding optional complexity to our minimalist platform. Users will start with a “just write and publish” experience as always, but we'll now offer a rabbit hole for advanced publishers who need to do more. I'll leave this intentionally vague for now.
The more of what you want algorithm is always bad. It's great for wasting an afternoon watching videos you may or may not actually care about. It's great for putting on repetitive music in the background. But it's bad for serendipity and discovery; bad for active listening; bad for critical thinking.
I'm reminded of this today as I listen to Kind of Blue on YouTube Music. I don't listen to this album, or jazz generally, every day. But when I listen to it, I'd like the algorithm to feed me more of that music instead of more of what I already like.
This is not a promise the algorithm can make me. I'm not being aurally guided by a human DJ, who can gauge my mood and play off of it. I'm being led by a pre-programmed robot who only knows what I already like, and doesn't have any preferences of their own. So instead of more jazz, I'm literally hearing indie folk artists I like next to vaguely genre-adjacent tracks I've already heard a million times.
I'm sure this is what the people in charge of these products think the masses want: more of the same. How boring is that, though? Why not tune the algorithm toward inspiration instead of tedium? Why not use this powerful position as product designer to break filter bubbles, instead of building and reinforcing them?