Moravec's paradox: "We're more aware of simple processes that don't work well than of complex ones that work flawlessly"

2 replies

0 recast

5 reactions

Lessons in there ⤵

0 reply

0 recast

0 reaction

"Pay more, get less. Pay less, get the actual thing"

0 reply

2 recasts

7 reactions

Today's PSA: Always account for platform risk and plan for alternatives and backups. If you rely on a single provider for anything, you're putting yourself in a situation where they can, and likely will hold you and your work hostage at some point. Reminder that 2 is 1 and 1 is none. https://blog.openreplay.com/github-alternatives-2026/

1 reply

2 recasts

16 reactions

"Deep curiosity is a luxury good. It is expensive; not in dollars, but in cognitive effort. It requires time, freedom from short-term incentives, and the willingness to question accepted truths. These costs act as a barrier to entry." https://platforms.substack.com/p/humans-as-luxury-goods-in-the-age

0 reply

0 recast

11 reactions

This is a very common take. If an LLM can technically do the task, that means it’s bad news for the person who used to do the task because they are presumed to be rendered obsolete. The reality is more complex. 1) Not everyone knows how to prompt what they want to conjure onto existence. 2) Most people treat LLMS like they would a google search, so the dominant way most people prompt is a vague ask more often then a detailed one. 3) Don’t conflate what you see as ubiquitous knowledge. Even though the barrier to entry has been lowered, few will know that they can replicate something they see this unless they’re really paying attention. (Which few actually are). 4) Those that are paying attention from the standpoint of using new tech are mostly comprised of people who experiment and have an open mind to tinker with it. 5) The person hiring you for a creative job seldomly cares about the tools you use. They care about the output. 6) Those that have no domain experience are more likely to fumble and assume they can do what someone who has experience can do. They will inevitably paint themselves into a corner because they don’t know the questions to ask the LLM to get them out it. 7) All rumors of a skill/tool’s death are greatly exaggerated for clickbait purposes. 8) Established tools will transform to meet the demand of features that impress the masses. 9) Not all tools/products will incorporate them successfully 10) AI won’t take your creative job. It will convince someone who doesn’t do your job that they can do it. Inevitably, they will get frustrated and ask for help/hire someone who’s proven their experience. Position yourself accordingly.

1 reply

2 recasts

18 reactions

"Everything downstream depends on the quality of your thinking." This overview of Gas Town is easily one of the best things I've read around running multiple agent workflows. If thats you, then this is a must read https://substack.com/home/post/p-188005341

1 reply

0 recast

4 reactions

I also suffer from the speed of my thoughts either moving too quickly to be captured on paper/keyboard, or not being in an ideal location to capture them all. The game changer for me was leaning into audio notes, followed by transcription. Most phone OSs have the ability to transcribe locally, so I just utilize that feature, followed by a shortcut that automatically sends it to Obsidian where I flesh the idea more when I have time. It's reduced my stress and increased my output considerably. Win-win. There are also lightweight services that do this such as Audiopen, which I'd recommend if you want something that works out of the box: https://audiopen.ai/

0 reply

0 recast

1 reaction

@atown has the perfect formula going on with @tryemerge 1) Remixability Easy to prompt, easy to share. Constantly generating fresh outputs that can bring attention and usage back to it 2) Impulse buy territory Few people would question a 25cent purchase. Toy vending machine/ blind box logic works extremely well here 3) Ability to pay with alt coins This is the future, its just not evenly distributed yet https://farcaster.xyz/miniapps/pmZbrBBIA6wT/emerge

0 reply

0 recast

2 reactions

Who else here is using n8n for workflow automation? I recently built an automation pipeline that: * Watches a google drive folder new media * Transcribes the audio of each new clip * Using trained examples of the client's previous posts, it then writes new social posts based off the transcripts, and keeps formatting consistent. *Copy is written to a Google Doc so they can review before signing off and posting *Pings their team on Slack. So far it's saved them 12 hours a month.

0 reply

0 recast

6 reactions

Following 37 Signals move to open source Campfire, their Slack/Discord alternative a few weeks ago, Small Bets, an online community centred around supporting entrepreneurs followed suit by open sourcing their modified fork, which they've been running for over a year. This fork elevates Campfire beyond Slack and Discord IMO, and makes it an alternative communities should experiment with if they're looking to move away from platforms that either come with hefty seat subscriptions at scale, or are a privacy nightmare. The trade off is that you have to host it and take on the responsibilities that come with that, but it's worth it if you're looking to build a more self sufficient home for your community. For those who are concerned about the future of channels here on Farcaster, it's worth noting that it is technically possible to incorporate one inside Campfire via webhooks. I'm writing a larger piece around how this can look and operate, possibly being a "best of both worlds" client that allows channel operators to hedge against any uncertain future they might have on Farcaster. Overall I think its important to show what a hybrid client can look like. Especially one that allows for communities to grow at their own pace, without changes being foisted on them by external devs and VCs. https://github.com/antiwork/smallbets/blob/master/campfire-mods.md

1 reply

0 recast

7 reactions

A reminder that @felt helps to make sure any account you want to see more of doesn’t get buried by the algo. Personally a big fan of how how mini apps work on desktop, allowing you to scroll multiple feeds at once. The mobile experience isn't bad either.

2 replies

0 recast

8 reactions

Where there is a will there is a way

0 reply

2 recasts

5 reactions

When it comes to coding with A.I agents, one prompt that has cut my error rate down considerably is asking it to check its context window before it applies any revisions. Night and day results

0 reply

0 recast

4 reactions

It's thinking like this that gives Bluesky the competitive edge for brands to invest their time in long term. Farcaster had the right idea with channels, but since they dropped support to improve them they sadly now sit in this limbo state. It all leads to unnecessary confusion and friction for those who want to plant a flag here and engage with this ecosystem. Having said that, the Merkle team changes things up every few weeks around here, so maybe they're re-evaluating how they approach brands outside of an airdrop first perspective 🤷‍♂️

1 reply

0 recast

0 reaction