implementation
Pilots are tightly scoped interventions that we expect not to work (remember, we have to explicitly prove efficacy as a defense against confirmation bias), so we use small populations, focus on speed to market, and do them in an operationally dirty way. Besides being fun to say, “operationally dirty” just means that we’re shooting for minimal
... See moreMatt Wallaert • Start at the End: How to Build Products That Create Change
.implementation .modelthinking
Sharpen your intellect by making it a habit to do one thing at a time.
Kevin Horsley • Unlimited Memory: How to Use Advanced Learning Strategies to Learn Faster, Remember More and be More Productive
When an intervention doesn’t create the behavior change you want, you’ve got a decision to make. And like intervention selection, this one is something you ultimately just have to intuit: either you revise the pilot and rerun or you kill it and return to your pressure map and intervention design.
Matt Wallaert • Start at the End: How to Build Products That Create Change
.psychology .implementation
It’s what’s known as the classic cocktail party effect: one mention of your name, and neural systems that were sailing along snap into action. You don’t even have to do any work. Most things don’t have such nicely built-in flags to alert you to their significance. You need to teach your mind to perk up, as if it were hearing your name, but absent
... See moreMaria Konnikova • Mastermind
.implementation
Holmes’s trick is to treat every thought, every experience, and every perception the way he would a pink elephant. In other words, begin with a healthy dose of skepticism instead of the credulity that is your mind’s natural state of being.
Maria Konnikova • Mastermind
.implementation an excellent hook to think of everything as pink elephant
Since we can’t improve our people-reading skills that much, we have to focus our efforts on making others more readable.
Eric Barker • Plays Well with Others: The Surprising Science Behind Why Everything You Know About Relationships Is (Mostly) Wrong
Welcome to the post-training era for startups
Training LLMs can be divided into two major phases: pre-training and post-training. The pre-training phase is an extremely expensive process that involves training a general model from a large corpus of data. Even in the case of DeepSeek, a single run of training costs $6 million, while it’s estimated
... See moreEvan Armstrong • What Actually Matters (And What Doesn’t) for DeepSeek
To avoid impulsively “just shipping it,” he introduces deliberate friction through cooldown periods when reviewing designs or providing feedback. This intentional pause allows ideas to mature, fostering more thoughtful and meaningful responses.
every.to • Sailing Against the Current of Frictionless AI
When you choose your pond wisely, you can best leverage your type, your signature strengths, and your context to create tremendous value. This is what makes for a great career, but such self-knowledge can create value wherever you choose to apply it.