A prompt-a-thon for those who get it doneA prompt-a-thon for those who get it done

Today in Austin, TX, a few of us got together to put some random ideas to work and build any anti-spam solution that improves our lives.

About the event: Come to the bitcoin++ clubhouse this Saturday, May 16th for a friendly competitive 'prompt-a-thon'. Who can make the best 'anti-spam' project in 4 hours?About the event: Come to the bitcoin++ clubhouse this Saturday, May 16th for a friendly competitive 'prompt-a-thon'. Who can make the best 'anti-spam' project in 4 hours?

Winners will take away a baseball cap prize:Winners will take away a baseball cap prize:

2nd place gets bitcoin++ and 1st place choose Core or Knots

I contributed in my way: being a fly on the wall, sussing the vibes, and reporting back to you. I asked participants a short list of questions including:

1. what’s your name/nym
2. how many hack-a-thons have you participated in before?
3. what AI frontend and/or model are you using today?
4. what problem are you targeting for your anti-spam solution?
5. what’s the best anti-spam solution for the long-term: ...filter spam out? make spam expensive?
   a. and this question led us most often here: does the responsibility of finding smart solutions to spam fall to the individual user or to the platform?

Here's what they had to sayHere's what they had to say

plebdev (@bitcoinplebdev)plebdev (@bitcoinplebdev)

15 hack-a-thons

Codex app, GPT 5.5 low fast

Bitcoin - because it’s an interesting problem and maybe a funny, adversarial solution

Curate signal - web of trust - fighting spam is an always losing game. But you can always define what is timeless to you and what you want to curate - that’s more of a winning game. Focus more what you want to see vs what you don’t want to see.

JakeJake

2 hack-a-thons(these were Austin BTC++ mempools and this one)

ChatGPT

Large transactions on blockchain

You can’t pin down what spam is. BIP110 is about picking what people use now, but the problem is, when you patch those out, people just come up with new ones. Patching up all the holes is the wrong approach. Instead, incentivize the behavior you want.

stutxostutxo

6 or 7 hack-a-thons

Codex ChatGPT 5.5

Solving the debate: is it spam? or is it a monetary transaction?

Proof of human or make it expensive, Twitter is doing it - you have to buy credits to call the API. Some way to tell who you’re talking to is human, but this is the huge problem.

JackieJackie

2 hack-a-thons (these were at an internship and one internal at a startup)

Codex with Superpowers

Bookmark manager for shopping - organizing stuff I wanna buy

Slop detection might be most reliable, could be automated. If a human can tell what's AI-generated, then an LLM could. A platform is free to filter if it improves their product, but depending on the approach that could be bad. Mostly the user ultimately decides.

Nifty (@niftynei)Nifty (@niftynei)

more than 10 hack-a-thons, my first using AI tools

ChatGPT 5.5 default fast, first time using ChatGPT usually use Claude

I’m trying to make it easy to see who you’re getting email from and how to unsubscribe, keeping the user in the loop.

No solution other than building better tools. Email is easy because of the concept of unsubscribe. Whereas elsewhere, it’s more difficult to curate. It really depends, people’s definition of what counts as spam is different. Like I’m interested in some things that could be ‘spam’, I do want to see them, just less frequently.

ReflectionReflection

Over the last year, I pretty much turned my ears off to the endless debates on spam inside of bitcoin circles. It lost meaning for me; I just don’t have real world touch points with the subject. But speaking with everyone about it briefly today, I realize an answer to spam in general as a concept is pretty clear, and we all basically agreed on this point: Spam is subjective and it’s your job to deal with.

That’s pretty clean to me. I don’t have any particular problem looking at it this way. Just like anything else in my life that causes me pain, I am ultimately the best person to address it.

ResultsResults

Peer-ranking determined the winners. Each participant had 5 minutes to show off what they made.

Winner: subbot.me by nifty - a service that analyzes who you get email from where you can unsubscribe in one click

Second: Bitcoin Bouncer by plebdev - a living filter for bitcoin txs, replaces hardcoded filters with an agent and spam dectectoor - sends pending txs to a twitter poll where the people decide if it's spam or nah

Everyone put together something that worked. The sprint was successful!

Thanks for reading.