Public posts preserved on my own site.
IDEA: anonymous mastodon client, but when you can follow and chat with Eliza, the non-smart chat bot and optionally have Eliza like and reply to all your posts, which would be only written to localstorage.
Fully supporting anonymous users is hard, but I'm going to keep trying.
MASTODON™️ "Your employer probably didn't block all the instances"
Just imagine: anonymous access Mastodon client with a menu of starter packs to pick from to start up with like
- "Economists" follow all, create list
- "Economics tags" follow all
- "Mastodon cliches and stereotypes" follow all, create list
- "The good tags" follow all
Just trying out Mastodon would go from, like a 1 semester college course to ~3 -4 clicks, well within the 3.46 minute attention span of a modern human.
This is so cool. So this link dumps you straight into the anonymous user experience.
I got one starter pack going, maybe if there were 20-30 more to pick from.
Where oh were to look to find more starter packs/collections. I know, I can do 2,648 API calls to find the collections of my friends.
Brain size and nerve count goes up with animal size, but large animals are not proportionately smarter. This makes me think that a lot of brain volume is for handling the increase in the number of nerves in the body. Insects imply that LLMs can get very efficient with parameter count if there is pressure to do so.
Biggest missing mastodon-to-other-platform integration
Archived poll
1 voters
Platforms grow when there are easy to use cross platform account correlation tools, i.e. "import my friends"
Seems like there used to be a website/browser plugin/frames thing where you could browse the web and leave graffiti, comments on other websites. It would only be visible to other people using the same frame/browser plugin.
And cross platform commenting (screenshot + comment on the *other* platform), this is the closest that you can get to anonymous replies.
IDEA: Anonymous mastodon replies, where you fill in your comment and the computer plays it back at maximum volume in case the target is shouting distance.
Anonymous replies, where you type up your comment, print it out and mail it on a postcard.
Anonymous replies, where the comment is posted to paste bin
Anonymous replies, but it just clears the text box after clicking send.
Anonymous replies, but when your click send, it begins a brute force password guessing attack accounts until it gets one, then deletes all the existing content and, removes the bio and then posts a reply.
IDEA: anonymous access mastodon, but when you post, it redirects you to a google search. I think we should just give Riker what he wants.
IDEA: popup after you post something on mastodon that says, "Hey, your post will get the same reach as anyone else, you can't pay to get more attention"
Well, I broke my mastodon client and maybe found a new feature. "Write-only mastodon". You can post, but you can't be distracted by content. Perfect for people who weren't paying attention anyhow. Oh wait, that's the same as those autoposters that post to mastodon, but the author never visits mastodon to read the comments. Everything has already been invented.
I'm not making typos, I'm proving my humanity.
About 60% of the twitter/bsky/mastodon experience can be done anonymously. But client never surface an anonymous/no-account mode.
You only need an account when you want to interact. But ALL the major sites, the first experience is "You know why you're here, create an account" and you can't see anything of the normal experience. (Trends and a chaotic federation feed is not my experience and that is a bad experience)
"Matt, that's impossible! The good API endpoints require auth!"
Nope, follow, main feed, algo, search, lists, favorites can all be done client side.
What you can't do is notifications, chat, replies/reactions.
"But Matt, Twttr has been around since 2007, everyone knows what it is!"
My wife hasn't used any of the twitters ever. Most people at my office don't have bsky or mastodon.
We lose these people my making them login before feeling any of the regular experience.
Tags are such a weird bird.
- kind of like a friend because you follow them and they show in your feed
- kind of like a saved search, for when you are monitoring a topic
- kind of like "bold" to tell readers what is important about this tweet
Here is a calculation that can be done but only with a large chunk of the total dataset: reach.
- Your follower count - minus inactive accounts
- Get follower count of each retweeter - minus inactive
- Get waking hours for each account - minus everyone asleep during the time posted
And 10,0000 API calls later, you have a good proxy for reach.
Mentally I see 2 kinds of twitter/mastodon clients
- *current data*. For a dozen API calls you get an experience. Works great if you have already found your peeps, your tags, and so on.
- *all data*. You need a large dataset to do analytics, search or anything that is computationally expensive (like an algorithmic feed that uses the full platform's data to find the good stuff, or stuff you might like, etc)
Another dimension is how connected it is.
- Musk twitter is completely disconnected
- Mastodon is connected, but only really to other activity pub
Oauth clients could be connected. Obvious example, all the apps to post to 10 different platforms all at once.
It rocks.
We're not going to have an RSS revival when the best way to get the RSS feed is to view source and search for RSS.
And there isn't going to be an RSS renaissance while the RSS feeds have CORS blocks on the xml endpoints! Because nothing says user friendly like needing to setup a GHA proxy for every !@#$ feed you want to follow.
Why get rid of DST when you can get rid of time zones too?
Let's switch to internet time.
Fixed two more things
- they're tweets. I'm not learning a new set of words for each platform.
- Every user should see a blue check on their profile because you know that you exist, you know you are sentient, relative to you, you're Verified.
Mixing bsky and mastodon posts... easy. They got similar temporal distributions.
Mixing bsky+mastodon+rss - uh, if they get no extra prominence, they disappear in the flood. If you give then high prominence, they show up day after day & now I'd have to track "unread" status.
I was going to post some guidance on how much $ to give the mastodon network and your instance, but a single user is like $0.10 a year, so any donation at all and you're mostly paying for free riders.
Micropayments don't make sense, the transaction cost is a multiple of the donation amount.
On Mocking Bird (http://mawkingbird.com) Get a mastodon account and a bluesky account and you can follow most of the people that left Twitter.
https://mawkingbird.com/
I'm still working out how to get more peeps off twitter or at least follow them from other platforms. With the API nerfed and Nitter being directly attacked by the last 2 engineers at this website, it is difficult.
I got an idea for doing account correlation so you can find you twitter friends elsewhere, like debirdify did, but that is on the roadmap.
A neutral/positive sentiment algorithmic feed is hard. My first attempt even after some primitive sentiment filtering, still is mostly anger.
Anger also drives up retweets, likes and replies. Maybe, "many likes/retweet and few replies" == positive sentiment? And getting ratio'd is a sign of controversy so likely angry comments in the feed and someone said something inflammatory.
How long can the AI bubble go on? Well, past actual ponzi schemes lasted ~16 years
Rail Mania was ~16 years https://en.wikipedia.org/wiki/Railway_Mania
2 hours for line goes up/NFTs?
4 hours for star wars hotel?
phah! amateurs
9 hours for Fermat's Last Theorem for dummies.
Taco Bell wasn't the cause, Taylor Farms was. Taco Bell was the only organization paying attention now that the CDC is gutted.
In Republican America
- Taco Bell == CDC
- Waffle House == FEMA
Why I had a bot write Mockingbird, an homage to the good twitter.
Okay, by popular demand, an non-evil Algorithmic feed
- The good stuff
- Filtered and sorted from content you said you wanted to see (people you follow, tags)
- Primitive negative sentiment filtering
- Cheap in terms of API calls.
Does not inject random content from across the platform because you didn't ask for that.
What an algorithmic feed would look like if it was consumer centric and not trying to drive up engagement..
- show me the good stuff. Content I already asked to be shown, sorted by quality signals (likes * retweets * replies)
- Let me toggle a filter platform content vs friends, and some sort of sentiment filter to deal with the fact that engagement can also just a signal of inflammatory content & rage
I don't need "trending tweet" if it really is, it will get into my feed organically.
Hey, what's the best feature of the good twitter that never made it to bksy or mastodon?
IDEA: give the infinite mastodon feed an end and tack bookmarks onto the end. This creates a natural workflow for revisiting bookmarks.
Because come on, really, how often do you revisit your twitter, bsky, mastodon bookmarks? Like every Friday at 4pm when you update your TODO and calendar. HA HAHAHAHAHHAHAHAHAHAHAHHA. Yeah right.
The smoke has arrived in DC. Canada gets revenge for the US violating its treaties and trade agreements. Irritates the eyes to just walk in it for 10 minutes.
I got to the bottom of the feed:
Algorithm may take some tweaking, as implemented I think I accidentally banned looking at deep history. I got another mastodon client (mimb) that does a better job at deep history anyhow.
People we should let back on to Mastodon
Archived poll
8 voters
Hacking on mastodon clients and I gotta hand it to 2018 twitter, they did really good product design. Even after you subtract out the addictive parts, it is still really good product design. Twitter clones should use it as the baseline standard to improve on.
TODO: probably 50% of the people I follow have an RSS feed to their blog and short of running a tool to harvest an OPML file, mastodon clients do nothing to make it easy to read those feeds. See, Twitter, Threads etc, exist to expand their platform, and RSS is a competitor so they'd never promote a competing platform. i.e. the open web.
My controversial hot take:
We should stop using 1998 computers for our daily driver. Everyone has been using 1998 computers without upgraded for 30 years and *this* is why climate change has gone wonky. See no one has upgrade to 2026 machine that are literally 1000x faster by CPU, 1000x more memory, 15,000x faster network bandwidth.
So in conclusion, maybe we should upgrade our computers.
Watch for my hot take tomorrow, "Why we should upgrade from vacuum tube computers"
TODO: add the mememaker UI to the mockingbird client