16 skills · 152 min

Start simple, add complexity when pain is measurable.

Not dogma, but defaults. Violate with documented reasoning.

2. Framework Choices

Use frameworks when they provide genuine leverage.

The question isn’t “framework or not” but “does this framework serve the thing we’re creating, or are we creating something that serves the framework?”

3. Human-Legible Systems

Configuration

• YAML/JSON are fine - the format isn’t the problem
• Problem is: 500-line configs with nested conditionals
• Good: Config a new team member can read and modify in 10 minutes
• Document non-obvious settings inline

Error messages that teach

• What happened
• Why it happened
• What to do about it
• Link to docs if complex

Logs you can understand

• Structured logging (JSON) for machines
• Human-readable format for development
• Timestamps, context, severity
• Searchable without specialized tools

Documentation lives WITH code

• README in each significant directory
• API docs generated from code
• Architecture decisions recorded (ADRs)
• External wikis for onboarding/process only

4. Local-First Where It Matters

Not “never use cloud” but “don’t require cloud unnecessarily.”

State should be inspectable

• Serialize state to file for debugging
• State machines explicit, not implicit
• Reproducible from snapshot

Sync as enhancement

• Local is source of truth where possible
• Sync failures don’t break the app
• Conflict resolution explicit, user-controlled

5. Composition Mindset

Libraries over frameworks when:

• You need one capability, not an ecosystem
• You want to control the architecture
• Exit cost matters more than speed

Frameworks over libraries when:

• Team expertise exists
• Time-to-market critical
• Convention over configuration is valuable
• The framework’s opinions align with your needs

APIs expose primitives

• Convenience methods are fine
• But power users can access lower levels
• Don’t hide the machine

Minimize exit costs

• Data exportable in standard formats
• Avoid proprietary lock-in where practical
• Document the exit path even if you never use it

Cloud & Infrastructure

When Cloud Makes Sense

Cloud Pragmatically

• Serverless for spiky, unpredictable loads
• Edge functions for latency-sensitive operations
• Managed databases when ops overhead > cost
• Self-hosted when control/cost/compliance require it

The question: Does cloud serve your users, or does it serve your assumptions about scale you don’t have?

UI/UX Philosophy

1. Immediate Feedback

<100ms for user actions, honest progress for longer operations

• Optimistic updates where safe (can rollback)
• Progress indicators that reflect actual work
• Spinners are fine - they indicate honest work
• Skeleton screens for predictable loading patterns

Loading states should:

• Show what’s happening
• Estimate time when possible
• Allow cancellation for long operations
• Never fake progress

2. Visible State

User always knows what the system is doing

• Status visible without digging
• Background processes surfaced
• Errors prominent, not hidden
• System explains its decisions when non-obvious

No black boxes

• User can understand why something happened
• Audit trail for important actions
• State inspectable in dev tools

3. Spatial Consistency

Things stay where you put them

• No layout shifts after load
• No rearranging “for the user”
• Muscle memory works
• Consistent component placement

Predictable navigation

• Back button works
• URLs are bookmarkable and shareable
• State survives refresh
• Deep linking works

4. Undo & Recovery

Implemented at the data layer, not just UI

• Soft delete by default
• Versioned state where valuable
• Recovery path documented
• “Are you sure?” is not a substitute for undo

Destructive actions

• Confirmation for irreversible operations
• Grace period before permanent deletion
• Clear communication of consequences

5. Respect Attention

Notifications

• User opts in explicitly
• Meaningful, not engagement-driven
• Batched where appropriate
• Easy to adjust or disable

Modals & Interruptions

• User-initiated, not system-initiated
• Dismissable
• Don’t trap focus unnecessarily
• Keyboard accessible

Autoplay

• Never for audio
• Video only with explicit user intent
• Motion respects prefers-reduced-motion

Defaults over customization

• Good defaults eliminate settings
• Power user options available but not required
• Complexity progressive

What This Rejects

Derivative Thinking

• “X but for Y” without asking if Y needs X
• Features because competitors have them
• Patterns because tutorials use them
• Architecture because FAANG does it

Cargo Cult Engineering

• “Best practices” from different-scale companies
• Microservices for 3-person teams
• Kubernetes for single-server loads
• OAuth for internal tools

Premature Complexity

• Abstraction layers “for future flexibility”
• Scale architecture before scale problems
• Features before foundations work
• Real-time before single-user works

Process Over Thinking

• Scrum ceremonies replacing actual thought
• Documentation for compliance, not clarity
• Meetings about meetings
• Roadmaps pretending to predict

Application

When Reviewing Designs

First-Principles Check

• Not done: What new thing does this create? (Not “what existing thing does it extend?”)
• Not done: Why does this need to exist?
• Not done: What becomes possible that wasn’t before?

Simplicity Check

• Not done: Is complexity earned or assumed?
• Not done: Can a new developer understand this in an hour?
• Not done: What’s the simplest version that solves the core problem?

Tool Fitness Check

• Not done: Do tool choices serve the creation, or does creation serve the tools?
• Not done: Is the framework justified by team expertise + problem fit?
• Not done: Are cloud dependencies necessary or assumed?

Human-Legibility Check

• Not done: Can someone read the config and understand it?
• Not done: Do error messages teach?
• Not done: Is documentation where developers will find it?

UI/UX Check

• Not done: Is feedback immediate or honestly progressive?
• Not done: Can users see what the system is doing?
• Not done: Is everything recoverable/undoable?
• Not done: Are interruptions user-initiated?

When Generating Solutions

Start by asking:

1. What genuinely new thing are we creating?
2. What’s the simplest architecture that enables it?
3. What complexity is earned by real constraints?

Default to:

• Simplest tool that works
• Framework if team knows it and it fits
• Local-first where possible
• Cloud where genuinely needed

Add complexity when:

• Pain is measurable, not theoretical
• Team agrees on the tradeoff
• The path back to simple is documented

Threshold Triggers

When to upgrade from defaults:

Justified Exceptions

Complexity is acceptable when:

• Frameworks: Team expertise exists AND problem fits framework opinions AND time-to-market matters
• Cloud dependencies: Multi-user collaboration OR heavy compute OR compliance requires it
• Microservices: Teams can’t coordinate on monolith OR genuine scale isolation needed
• Heavy tooling: Build time investment pays off in development velocity

Document the reasoning. Future you will thank present you.

Pragmatic Defaults

Start simple, add complexity when pain is measurable.

1. Begin with the simplest architecture that could work
2. Wait for real problems, not imagined ones
3. Measure before optimizing
4. Document why you’re adding complexity
5. Ensure the path back to simple exists

Premature complexity is technical debt with interest.

Anti-Dogma Clause

These are defaults, not laws. Violate with documented reasoning.

Every principle here has valid exceptions. The goal isn’t purity - it’s intentionality.

Valid reasons to deviate:

• Team expertise strongly favors different approach
• Business timeline requires faster path
• Regulatory/compliance requirements
• Measured performance needs
• User research contradicts assumption

Invalid reasons to deviate:

• “Everyone does it this way”
• “We might need it someday”
• “The tutorial used this”
• “It’s best practice” (without understanding why)

When you deviate, write down why. One sentence in a comment, ADR, or README.

Quick Reference

The Core Question

When designing anything, ask:

“Am I creating something new, or commenting on something that exists?”

Renaissance architecture isn’t about rejecting modern tools. It’s about using them to build genuinely new things - not just another variation on established patterns.

Medieval scholars could only write commentaries because they believed truth was revealed in the past. We have no such limitation. We can create.