infra

We have been offline for a few days, as I kind of have a day job and other commitments, but I am going to do a sprint this weekend to get bugs fixed. Then, time to really push and get out there more! <3 to all of the alpha testers!

After what google and apple did this week, I am glad we decided to build a browser app instead of a mobile app. Look at what we can post:

Gulf of Mexico.

So far on the bug tracker:

Filtering after pulling up an individual account is glitchy.

Startup is a little slow, taking maybe 10 seconds to launch the feed.

The feed should filter out videos seen previously.

Skinny phones don't show videos right.

Not bad for an alpha launch. Will fix soon.

WRT our alpha, so far the service seems to be performing pretty well. Login and initial feed starts may take several seconds to load, but it is responsive once going. We may add more indexing on or allocate more resources to the database to speed it up a little more.

Alpha test is live! Probably a little buggy, could use improvement, but live! There are no limits on users except what the server can handle, no app that needs to be installed. Check it out!
www.infra-at.com/index.html

After some discussion, I think we are going to let users upload long form video up to maybe 500mb, since others don't allow this yet. This won't be pushed to the AT protocol until we set up our own PDS supporting larger files. And they will be categorized as adult until we have a good porn detector.

Also, the creator portal is not going to be active just yet. Why? Because I do not yet have a reliable model categorize adult content. For now, we just pull in everything that has been posted on BlueSky. Later, we will offer video editing and FX and host videos longer than BlueSky allows.

If we get a decent sized user base, I would expect that our server may need some upgrades and perhaps some horizontal scaling. Slow feeds may result if we get overloaded. That would be an amazing problem to have.

The layout may need work for different devices and resolutions. I did the dev work on a desktop browser and am in the process of finessing it to also fit well on mobile.

Also know that we do not store important user info. Your login (your BlueSky user and password) generates a token that is used to keep you logged in, but once you invalidate the token via BlueSky you get logged out. Passwords are not kept. Your posts are automatically cached in infra, but that's it.

I am using some interesting models to help with the feed, though I am considering letting users adjust the weights on the feed to their personal preference. This would allow for greater weight on recency, for instance, or like counts.

Better models will come the more we use the app as well!

So what can you expect from the infra launch? A relatively functional webapp.

Have I built a mobile app for it yet? Nope. App stores are, uh, rather restrictive. And I have not had enough time for that effort yet.

Is the app perfect? No, I will debugging the app in prod ;)

But, it should be fun!

Rather, 2 weeks of me working a bit on the weekend and in spare time outside of work.

Well, it looks like I ordered the wrong case for the new server. Who even builds a mid size tower that is not ATX compatible in 2025?

Regardless, the infra software is largely MVP-complete. Check out the screenshot! A few tiny odds and ends left todo, but 98% of the way there. 2 weeks dev time.

It's not going to be perfect right away, but I think it will initially scale to about 10,000 users on the $1200 server we are building. And in some ways I think we are an improvement over existing apps, even at launch and before our grand plans come to fruition.

How many devs and how many months (years?) do yall think meta had before they launched reels? I'm just 1 dev here working in my spare time after work, having a "hold my beer" moment to push out a video app.

Ordered hardware for our first server! Let's see how far a 4tb SSD and 16 cores gets us! We will be assembling as soon as it gets here and are looking to launch is about a week - no limits on sign ups, you just use your BlueSky to log in.

We are working on ways to do all of this. And if anyone wants to add ideas or feature requests on that front, we are all ears.

Stay tuned! We are planning to launch ASAP, running on a single server, hoping we get enough users to crash it before we scale up!

At infra, we are looking at ways to monetize content, with general ad offerings and with systems to do more specific contracts between businesses and creators. Ideally, we want to cut down on the labor it takes on this side of the business. The openness of the AT protocol complicates things, too.

As much as we are not particularly satisfied with how money oriented the world is, artists need to (and deserve to) be paid for their labor. And practically, big creators will not come to platforms where they cannot be paid, and without popular creators the rest of the ecosystem lags.

As we (the team is 1 software dev/hotdog enthusiast and 1 graphic designer/hotdog enthusiast) finish up the prototype app, I think we on the more open protocols need to have a discussion about monetization. I know BlueSky is already discussing it as well, and I'm not sure what their status is. Cont.

Getting really close to launching! And now we are posting a text post in order to have a test case for QA.

We all agree with @tenspd.bsky.social that hot dogs are better than politicians. And now we are using her video as a test case for infra!