The 37% rule

The 37% rule
2 minutes by Michał Poczwardowski

There is a mathematical strategy to help make better decisions when choosing from a sequence of options you cannot revisit. It works by spending the first 37% of options on observation only, then picking the next option that beats everything seen so far. This gives you roughly a 1 in 3 chance of finding the best option. The key lesson is simple: calibrate long enough to set a benchmark, but not so long that you miss your best choice.

[Webinar] Can you prove AI is working?
sponsored by Unblocked

AI is in your engineering workflow. While the token spend shows it, the throughput doesn't. The human is very much still in the loop, and that's a context problem. Join live on Aug 19 (FREE) to learn how to track AI productivity, how context maturity affects where teams stall, and what it takes to get real value from your agents.

Perfection is not over-engineering
4 minutes by Var Zero

Over-engineering means solving the wrong problem, not caring too much about quality. A perfect solution exists when requirements are clear enough that only one answer fits. The real culprit behind bloated systems is poor requirements gathering, not ambition. Get every constraint on the table and the right solution becomes obvious.

Effectiveness comes from developing both leadership and management skills
6 minutes by James Samuel

Leadership and management aren't opposites — they're both necessary. Leadership creates direction and inspires movement, while management turns that movement into real results. One without the other leads to either vision without execution or busy work without purpose. The good news is that you're already practicing both, since every time you influence others or make smart use of limited resources, you're doing exactly that.

Make no assumptions
9 minutes by Will Larson

Codebases often build up distinct layers over time as new developers add code without fully understanding what came before. AI tools are making this worse, and the same problem is now spreading to how teams reason and make decisions. AI-generated analysis can look polished while being built on flawed data, leading teams to solve problems that don't exist. The fix is to inspect every layer of reasoning rather than accepting it at face value.

Avoid the trap of magical thinking
4 minutes by Bjorn Roche

Good leadership isn't about inspiring speeches or chasing productivity hacks. It's about understanding real business challenges, removing bottlenecks, and focusing teams on what actually matters. Magical thinking lets leaders avoid responsibility, so when things fail, they blame effort rather than direction. Strong teams built on clear goals and good communication outperform collections of star individuals.

industry

• AI Companies Are Trying to Hide a Staggering Amount of Debt
• Judge approves a $1.5B Anthropic settlement over books used to train Claude
• Kimi K3, Qwen 3.8, and Anthropic's (potential) Unravelling
• Who’s Afraid of Chinese Models?
• AI advice made people three times less accurate but twice as confident

security

• OpenAI’s accidental cyberattack against Hugging Face is science fiction that happened
• What a Law Firm’s Ransomware Nightmare Can Teach Your Startup
• Hacker wipes Romania's entire land registry database

And the most popular article from the last issue was:

• Engineering leaders day-to-day activities