The Messy Middle · Scott Belsky · 2018
The Messy Middle is a book of common sense - there’s an incredible amount of wisdom in here if you’re just starting out. Less gold for more experienced folk.
Key Highlights
The Messy Middle isn’t chronicled because its unappealing and too revealing. To disrupt and industry be a thesis driven outsider → and stay alive long enough to become an expert. You need to develop a source of renewable energy. Without any customers or evidence of progress, without a steady stream of rewards, you will feel empty. You have to manufacture optimism.
Endure
You can get the important stuff right and still lose by not enduring enough. Few people can stay loyal enough to a strategy to have their vision realised. That's the hardest part.
- Be an energy giver not a taker, leave people with more energy than when you started your conversation.
- Your weird is the source of your power. Don't try to fit in.
- Break longterm goals down into chapters. Chapters are clear goals - underscored by why they’re important - teams do tactics.
- The easy path will only take you to a crowded place. There's often an easy option and a best option. Your competitors can follow the easy option.
- The best way to complain is to make things. If you’re actually willing to do the work, you’ll have more influence than those who simply do their jobs.
Optimise
- Why did that work? How do you do it again? Optimise the hell out of everything that gets traction. Optimisation stems from a conviction that you can do better.
- Resources become depleted, resource-fullness doesn’t.
- Hire for Initiative more than experience.
- Sometimes you’ll need to suppress the team’s immune system to get change through.
- Build risk-taking into your culture - you need psychological safety to experiment.
- Foster apprenticeship - find a way for the new to work with the experienced.
- Shed the bad to keep the good - balance the needs of individuals with that of the team and the mission. People need to admire their team members and see their commitment.
- Culture is created through stories - the early ones are very important (Amazon using doors for desks). Helps team members work out if something is uniquely x, or not... and right from wrong.
- The person who did the work should present the work. Be wary of false attribution.Present your ideas don’t promote them.
- Never outsource your competitive advantage.
- Your process should be about helping your team solve problems, not quelling your anxiety. Give teams problems, let them choose how to work through them. Delegate - entrust - debrief - repeat.
- Post-Mortem everything.
- The greatest risk is taking a shortcut in the one area that distinguishes you the most. Every product has a few differentiating attributes, don’t take shortcuts, rush, or strip down the process of creation for these features. Speed through the generic stuff - take time to perfect the few things that you’re proud of.
- Tap into the knowledge of the people around you, but don’t give tough problems to committees.
- It’s hard to make a simple product, it’s even harder to keep it simple. Make one subtraction for every addition.
- The greatest cost of trying to sustain too many initiatives is having too little to thrust behind each goal.
- You don’t want all the customers straight away. You want to right group to start iterating with. The ideal customer is … willing > forgiving > viral > valuable > profitable.
- Effective design is invisible, good design is as little design as possible. Beware of creativity that compromises familiarity.
- Build your narrative before your product.
- Best to market > first to market.
- Identify and prioritise efforts with disproportionate impact.
- Engagement drivers (longterm) interest drivers (short term).
- Mine contradictory advice and doubt to develop your own intuition.
- The danger with measures is their gravity, and how quickly they dictate our daily actions. Measure each feature by its own measure - determine what each measure is intended to achieve and measure it accordingly.
- ‘People don’t use that’ is not deep analysis. We don’t want people to use password reset functionality but we should have it. Tow hooks aren’t meant to be used all the time - but they need to be discoverable and effective when needed.
- Avoid having too many metrics - the more you track the fewer you’re focusing on changing. Boil the business down to one or two core metrics. Keep evaluating their efficacy and alignment with your longterm goals.
- Active Commitments: investments of time, energy and resources in areas you willingly choose to love and pursue. Passive commitments: commit to doing something that doesn’t align with your interests. Operating out of guilt rather than intention. Everything you do should be an active commitment.
The Final Mile
- Keep repositioning the ultimate goal as being the furthest away point that you can see
Subscribe Button In the News
Amazon saved 4500 developer years of work by using AI assistants to upgrade applications to Java 17 (Article). Coding with Cursor is blowing people away (Article). Which begs the question… will generative AI replace software engineers? (Article).
Quick Links
- The three pillars of observability (logs, metrics, traces) · Article
- Lenny on time management techniques that actually work · Article
- Harvards 6 guidelines to mastering the art of negotiation · Image
- Scaling Product Discovery by Tim Herbig · Article
- Brian Armstrong on action producing information· Short Video
- Use these hooks in your ads to boost your sales · Image
Amazon.com Recommendations. Item-to-item Collaborative Filtering. Greg Linden, Brent Smith, and Jeremy York 2003
At Amazon.com, we use recommendation algorithms to personalise the online store for each customer. The store radically changes based on customer interests, showing programming titles to a software engineer and baby toys to a new mother. The click-through and conversion rates — two important measures of Web-based and email advertising effectiveness — vastly exceed those of un-targeted content such as banner advertisements and top-seller lists
Amazon were pioneers of early recommender systems. The author reduces compute required vs greedier methods and navigates data sparsity issues. Much of the calculation to happens offline but recommendations can remain responsive in realtime.
/w=3840,quality=90,fit=scale-down)
Book Highlights
Starting at the end is like being handed the keys to a time machine. If you could jump ahead to the end of your sprint, what questions would be answered? If you went six months or a year further into the future, what would have improved about your business as a result of this project? Jake Knapp · Sprint
Likert’s most important contribution from his famous paper was that he demonstrated that the simple method of getting a score by adding up all the answers to the individual response formats was as good as the more complex methods that his predecessors in scale development had proposed, such as getting a set of judges to assess each statement for its potential contribution to the overall scale. Caroline Jarrett and Steve Krug · Surveys That Work
We’ve gone from many people to one computer, to many computers per person. The next wave of computing will make demands on us in terms of privacy, security, bandwidth, and attention. We can no longer design technology in the way we designed for desktops. We need to think about how we’ll design for the next 50 billion devices. Amber Case · Calm Technology
Organising by actionability counteracts our tendency to constantly procrastinate and postpone our aspirations to some far-off future. Tiago Forte · Building a Second Brain
/w=3840,quality=90,fit=scale-down)
Quotes & Tweets
Imposter syndrome: I don’t know what I’m doing. It’s only a matter of time until everyone finds out. Growth mindset: I don’t know what I’m doing yet. It’s only a matter of time until I figure it out. The highest form of self-confidence is believing in your ability to learn. Adam Grant
Tiers of web scraping (easiest to hardest):1) You already have the URL, parse HTML (beautiful soup)
2) You already have the URL, javascript spaghetti, selenium)
3) You don't have the URL, clicks needed (playwright)
— AI Zone —
4) Few clicks needed, but not sure which (multi-on ish)
5) Few clicks + associations needed (LangGraph)
6) Many clicks, many sites, deep research needed (multi-on + GPT researcher) Greg Kamradt