A NON DEVELOPERS GUIDE TO UNDERSTANDING AND USING THE AT PROTOCOL, AND WHY THERE IS HOPE
I’ve kept up with all the public information I could find about Bluesky’s intentions into open source and open protocols, but it wasn’t until I was able to login for the first time and see it for myself that it all clicked.
THE PROTOCOL, THE COMPANY
The philosophy of an open protocol and the cultural good it can bring is a topic dear to my heart, and I’m pleased to write this article about my findings, since I conclude the AT Protocol seems to be a surprising and brilliant design, perfectly positioned to become adopted, and that it could level us up safely out of the Web 2.0 silo buzz feed hell we just escaped out of alive, circa 2023, post-pandemic recovery.
Please bear with any shortcomings in how I attempt to describe what is boiling up, I’ve not been so blown away in a while and it’s still sinking in for me what has been built by Bluesky so far.
The beauty of an open protocol is that it’s just what it sounds like: an open method for doing something. Nothing more.
It’s a way of saying, with what we have, how about everyone do it this way? Anyone can introduce a protocol in any form or fashion, but getting others to adopt the protocol is typically the difficult part, for there must be an *incentive*.
When Chris Messina introduced the hashtag concept, it was a quintessential example of the introduction of a simple open protocol. He explained the rules (use the # in specific ways), said that these are the rules he would follow, and he invited anyone else to do the same. It wasn’t forced, it was the other way around. It was an open idea that anyone could use. If someone wanted to come along and use the # themselves, or change the protocol to two ##, go right ahead. The incentive in the case of adoption with the hashtag was undeniably simple practicality.
The AT Protocol has a set of instructions for everyone to follow that are obviously exponentially more complex, but it’s the same open concept throughout. Specifically, the AT Protocol provides a method for how to use different languages to handle things like pushing or pulling the titles and text bodies of posts back and forth between two points (e.g. between my website and Bluesky data, or posts between my website and yours), or how to get the number of followers of someone given their handle, how to get replies, authentication for the login – all the stuff you need to make your own social media app or whatever tool you want, but essentially just to connect up, get all the data and then make your version available in an open exchange. It’s notable that the bar to entry is so low for someone to create somewhat of a copy-cat app to Bluesky and if something bad happened to Bluesky, people could move to the other cool person’s app.
As I hope to show below, it’s quite a bit smoother and way more powerful than that, with a lot of reasons why you would use multiple apps, and how that might look, considering there is already a Bluesky-inspired app store, and you can join it for free, all you need is a browser. No one controls the store, its called https://
To clarify then what’s involved, the AT Protocol (Authenticated Transfer Protocol) is the name given to the new open-source protocol that is being introduced and built by Bluesky. Bluesky is a different thing, it’s a for-profit company. It just so happens the company is gifting the protocol (the instructions with their vision in mind), and they are saying what Chris said, which is that if you want to do it this way, go right ahead, this is how they are going to do it. They built it their way and unleashed it into the wild. Now their private company Bluesky which uses it, acts as an example of what you can do.
THE FIREHOSE
The firehose refers to the single real-time stream of all the users, posts, and their metadata, and it’s easy to visualize as the name suggests the reality. It’s what your feed would look like if you followed every single public user on the platform, and then set your timeline to start streaming it all including all the replies and media in real-time.
The Bluesky app is said to have ~ 60,000 people atm, which would be a complete blur just as the water stream of a firehose would be with that kind of flow and pressure you would have up against your glasses, and yet still minuscule enough for someone at home to ingest and work with it off a laptop, where they are, now, writing code doing cool stuff with it.
To see what’s going on statistically, for example, if you wanted to know about hate speech, you wouldn’t need a tiny sample to then infer what your conclusion might be at scale, you could come up with a definitive, exact answer if you have the whole data set from the firehose.
You can also verify that a platform like Bluesky is continuing a good business. If they say they are not throttling, but the data isn’t in the firehose, well how would they explain that? I’ve always noticed that access to the firehose worked well as a gold standard for determining what is actually open.
And especially you can dip in a grab specific things when needed, like just the history of one public account on the internet, like going to a website and reading a page.
TAKE NOTE. History, culture, and finance suggest the firehose will not be open for long, for Bluesky (the company) is a business, so they are most likely to turn it off and sell access to their data. It would be an unusual philanthropic gift if they do keep the firehose open. It would make them one of the greatest social good organizations of all time if they left it free even if they had some public-good rules they required that you follow to use it. It would for sure be a global sensation if it remains open, and a town square signal would emerge clearly, as clear as it is today on Bluesky in a town of just 50,000. If, and more likely, when they turn it off, it won’t necessarily be de-facto evil and under the conditions we are talking about here, can be okay without spoiling the greater dream. It’s a business model step that, with bad intentions leads to enshitification but being crummy in this environment leads to extinction and more support for the developers who are supporting something worthwhile.
If ever it was worth stopping for a moment to consider a proposition, I think now is the time to point out, with both foresight and in retrospect, that Elon Musk’s biggest mistake with twitter of all the mistakes was killing the API for in so doing, he did the most damage in destroying the ecosystem. It was akin to turning the Amazon forest into the Sahara Desert in three weeks and then running the exact opposite way that we are going here.
We’ll see what happens with Bluesky. Upon turning off the firehose and going mainstream with the population base, they are likely to open up a limited API as a replacement to their firehose to offer more of a drip instead. They’ll likely offer a free level of API service for the young at heart and turn the spigot on a bit more for Fortune 500 enterprise sales. Then the big guns and I do mean guns like Microsoft and Facebook have to use a phone for the “Call Us Now” prices.
In so doing, in the longer run, small companies and developers would be priced out of the big data set, and for example, when it comes to getting the data (say for LLMs), the big companies like Google & Facebook would be able to afford to say they are justified to boot because they are regulating us. That’s one of the material factors that would lead Web 3.0 more quickly back in the wrong direction of centralized proprietary silos.
It certainly would be disappointing if Bluesky does that for we can regulate ourselves with elected officials we can hold accountable, we shouldn’t have to have forced bros. But it’s completely expected the firehose will be off soon but in a way that doesn’t necessarily trap us into that destiny of evil because another outstanding feature of the potential behind this setup Jack has orchestrated is that it’s already bigger than him now.
THE WEB APPS
If Bluesky enshitificates, we can simply not log in over there anymore. Just log in to one of your other apps you’ll probably have soon. For example, check out Flat, a minimal and clean Bluesky client that is using the AT Protocol:

To use it, you don’t need an app, just go to their website and log in with your Bluesky credentials, and you’re in, no setup or anything. Perhaps you might like Tokomeki.

You can simply go there and log in with your Bluesky credentials and get your stream in the form of the associated images. If you follow a lot of artists and photographers you might choose this site to log in to each day.

How about the Chrome extension already in the Google store?

It’s beautiful.

I didn’t log in to this one below but I have a good feeling it helps to illustrate what is happening with this ecosystem already:

If “The Blue” site above suits your purposes, wouldn’t if feel refreshing to pay them $8 directly on their website without them having to give half to Apple? The *incentives* for the developer ecosystem is insane right now, and the cost of entry is almost nothing to do what these sites are doing. It’s ironic that GPT-4 was cut-off in 2021 before the AT Protocol began and doesn’t know about it. But you can still prompt your way into building with it. This will incite even hobbyists to join in, and I think its well funded and well positioned to absolutely explode right now.
There are also tools you can use that are already built by people for example. If you have a blog, or want to use your Bluesky posts as a blog, check out Bluestream. Just drop in your handle and it will give you an RSS feed.


Here is the feed now on my sandbox site at Julia Set automatically:

If ever it was worth stopping for a moment to consider a second proposition, I think now is the time to point out, it’s absolutely in our favor as the sheep to not call the posts on Bluesky skeets. I get that its funny, and fun to rally a cause, and it’s an honest way to have some fun hopping on a fun train – though they are giving us an opportunity to remain agnostic and exist unbranded to them in a bigger way than Bluesky will be if they succeed. The word “post” is the most suitable term for an open world, whereas “Skeet” and “Tweet” are more propriety and temporary. A post is a universal thing that can be shared across all kinds of platforms including tweety platforms, but ultimately they will be *your* posts.
The opportunities and potential rewards, right now, for someone with a good idea who wants to start building, due to the scale this thing is now destined to take on, are of a once-in-a lifetime type significance. The ease one can build something right now, and the lack of anything out there known, will pay for itself if you have an incentive. At least one non-advertising business model is proven: Make it free and sell incentives. This is how you enshitificate if you lose your way. Your town won’t care, they’ll touch a button and be gone. Work for the town and the rewards could be more significant now, and less later as things become more saturated, also a good thing overall.
No more middle people in this scheme, build and they will come, fail and they will leave. It’s ironic all the stuff that Elon Musk is building in his do-everything app was a missed boat, thank goodness, for he almost controlled everyone, but with this design, he controls no one. He’ll have no choice but to offer an API to sync with the AT Protocol. Hats off to Jack Dorsey for gifting us to this opportunity, I think the best thing anyone could do besides build is spread the word and get builders excited about what we could do with this. If we move quickly, no company will be able to keep up with the development of the whole world. Also, I’m sold on the Bluesky company so far, the CEO is bopping around making all the right calls with the best energy possible, dealing with issues I could never foresee or handle.