Home Spoutible - Initial Thoughts

Spoutible - Initial Thoughts

Even Captain Ahab would let this one go for now. Syndication

This post has been updated to include new information regarding the Bot Sentinel feature that was announced on March 1, 2023 via Twitter.

Instead of pointing out all of the cracks we know exist with the infrastructure and features that are missing, I want to look at why the platform was created and the unique features that may or may not be good ideas for Mastodon.

The platform is built on ideas that we all should strive to meet with a social media platform:

  • Fierce guards against harassment, especially for POC and LGBTQIA+ communities.
  • Giving a staggering amount of control to the creator to moderate their content.
  • Free speech - just don’t attack other people.
  • Privacy and security focused features, but with some head spinning mistakes along the way.

Sound familiar? But there’s some big differences.

Controlling your audience

This is actually an interesting concept, a bit different that presented in Mastodon.

When you create a new post, you can decide who can participate in the discussion. You can choose to allow everyone to reply (Public), only people I follow, or only people i explicitly mention. Oddly missing is the option to post to only people that follow me. You can also change the audience AFTER posting.

Spoutible options for controlling your audience at post-level

Mastodon gives us the option to publish a post to a specific audience as well but curious if the way Spoutible presents the options is more understandable. One of my peeves right now in Mastodon apps is that there is no common naming conventions for actions.

Spoutible gives the creator all kinds of options to control their conversations.

  • Deleting replies
  • Controlling the audience as mentioned above (both at post time and after)
  • Setting the minimum “Bot Sentinel” rating criteria that must be met before someone can reply or contact you
  • The obligatory Report and Block User

Here’s the options you can set at your account level:

Bot Sentinel rating criteria for replying to your spouts

AI is in your Spouts.

The creator of Spoutible runs a Twitter analysis tool called Bot Sentinel. It reviews content on Twitter to determine the likelihood that an account is a bot, or that the account was created for the sole purpose to harass an individual. A score is applied to those accounts. The higher the score, the more likely you are a bad actor.

One oddity here is that there’s really no explanation of what the score means, or how it’s calculated…it’s just there on every profile….

In addition - it IS on every profile. And if you’re at 0%, it has no real meaning. So the user is just going to begin ignoring that piece of UI because it’s mostly the same across all users. It should really only display if a user is NOT at 0%. I assume they wanted something to display since it’s their unique aspect.

Bot Sentinel Rating Display on a profile

Users can set preferences to ignore individuals with a higher Bot Sentinel score, removing them from their timeline and not allowing them to interact with their own posts. Once again, since no one was NOT “0% Normal”, I couldn’t test this.

Update to Bot Sentinel Feature Information

On March 1, 2023, the CEO of Spoutible noted that this feature has yet to be turned on because it needs a certain number of posts for an account before it can do the analysis. It’s expected that this feature will be enabled in mid-March.

He noted that it was “well known” that this feature wasn’t turned on because he announced it in a few Twitter Spaces, of which I did attend a few. Those live events lasted for at least an hour or two each, covering many topics. Note that Spaces are only available when using the phone apps, and recently a feature that creates transcripts of recorded spaces has been removed by Twitter. There are no transcripts of the conversations and no documentation was found on the recently added Knowledge Base for Spoutible, either.

I have to respectfully disagree that it was well known. I even asked on Spoutible itself if other users had seen any non-0% accounts and one person did respond that most users didn’t have enough posts yet to complete an analysis, but no one knew that it wasn’t even turned on yet.

I also find it alarming that the replies to this Tweet (I rarely dive into Twitter anymore) were full of praise and adulation and not one concern about a user assuming an account wasn’t a bad actor because their Sentinel Rating was 0%.

At the very least, this feature should have been marked as “Coming Soon” or removed from the UI until it was running.

You have the power - to delete replies.

Users are also given power to delete not only their OWN posts, but also replies to their posts they deem as hurtful.

That’s right, without submitting a request for support, the original author can just delete any post from their thread.

I asked around to see if someone would respond and let me delete their message, and plenty of people took me up on the offer.

After deleting a bunch of replies, the original post shows a little trash can icon and the total number of replies I deleted.

More concerning is that at least as the original author, I could click on a deleted reply and still see the original offensive content - so it’s not really getting deleted, and is it just hidden from view but still included in the HTML? Oh my.

Spoutible post showing that I have deleted replies from the original post

Encrypted DMs

Yep, they have encrypted direct messages. There’s nothing much to show - it just looks like a DM interface. No explanation of what type of secure encryption is being utilized, but it is supposed to be. So taker their word for it…right?….

Spoutible’s Blue Checkmark - An odd choice

Because I was on the waitlist, I got in a bit earlier than others, and immediately went about setting up my profile - profile photo, background, introduction, hashtags - again - eerily similar to our favorite little pachyderm.

Then I noticed an option to “become verified”. Ooook….how does this work? I was intrigued.

You’ve got to admit, the Mastodon verification process isn’t easy for your Aunt Loretta up in Amarillo to pick up on. Maybe we can learn a thing or two. But I was, unfortunately, wrong.

This was the biggest shock for me. When I opened the dialog for completing a verification process, it asked for the following:

  • Create a video less than 30 seconds
  • Show your full face and say your name and your handle
  • Hold a PASSPORT…or GOVERNMENT ID in your right hand and display it in the bottom right corner of the video…..

I forgot to take a picture when I got to this screen, but I pulled this from Reddit - I can verify that it is exactly what I saw as well.

Spoutible Verification Modal, seen on first day of pre-launch

There was no statements about how this would be secured, where it was going, what steps would occur next - nothing like that.

Because of the explicit instructions, I assume once the “verify me” video is uploaded, it’s reviewed by an automated process, and if passed allows you through? Or sends to the next level of verification?

Then - that option disappeared and I never saw it again. Later the creator mentioned it was available by mistake. But some people had already submitted - what happens to their videos with their driver’s license / passport being displayed?! 🤦

At this point I stepped back a bit to decide how much private content I wanted to have associated to this platform. Not much.

Ads? Subscription Service? Open Source? It’s not clear.

I assume ads will start appearing somewhere, somehow - or my data will be utilized in some way. Or there will be “premier” features. But none of that is clear at all right now.

Algorithm…are you there?

I never really got a sense that some AI was trying to push me into a specific direction with posts it displayed or hashtags it presented to me. I’m not sure if that was because there just wasn’t enough CONTENT out there yet, or if Spoutible is going the Mastodon route and not using an algorithm to determine what to show you.

However, I have to say it felt more disjointed - I almost wanted an algorithm here to help me find people because it wasn’t happening naturally.

A Spout too soon

Even Captain Ahab would let this one go - for now.

This was rushed to production. It’s got all kinds of problems. It should not have been released.

But the alterna-twitter universe should take heed of what’s happening here and not make the same mistakes.

I do believe the spirit of the platform is good. But it could have used another 6 months of internal / beta testing.

This post is licensed under CC BY 4.0 by the author.

Contributing to Open Source - The Mastodon Ice Cubes App

Mastodon Apps - The Next Generation


Total Interactions: 5
2 Likes 0 Links 0 Posts 0 Re-posts 0 Bookmarks 3 Replies


Posts, Re-Posts and Bookmarks

No webmentions were found.


  1. @box464 this is a very fair take. I like Spoutible but I think it needs more work to function smoothly

  2. Respectfully, why show the rating in the UI if it's not enabled? Show "Coming Soon", hide it, make it clickable and show a "This feature isn't implemented yet." message.

    Currently, users could be interacting with accounts that are not good actors based on the displayed rating.

  3. Update..the last question, which had Mastodon mentioned, has been removed “because so many people didn’t know the other platforms”