ArgumentFrameworkSoftware engineeringLeadership
Speed and stability rise together. The trade-off is the myth
The idea
Accelerate reports several years of survey data on software teams and finds that the teams that deploy most often also have the lowest change failure rate and recover fastest. The received view, that moving fast means breaking things, is not what the data shows; the two rise together because the practices that make deployment frequent, small changes, automated tests, fast feedback, are the same practices that make changes safe. The Library calls it the only book on the shelf with a defensible empirical claim about what makes teams fast, which is why it is worth citing precisely.
The practices that make shipping frequent are the ones that make it safe, so do not trade one for the other.
When it works
Engineering management questions, system design discussions about release process, and any debate that assumes speed costs reliability.
When it may fail
Survey data shows association, and the teams that adopt the practices may differ in other ways. The claim is strong for software delivery and should not be stretched to every kind of work.
How to use it
Asked how you would make a team faster, name the four metrics and say which practice moves each: smaller changes for lead time, automated tests for failure rate, observability for recovery. Avoid the word 'careful' as a plan.
Where it came from
Book Accelerate · Nicole Forsgren, Jez Humble and Gene Kim
What the source says. The research behind deployment frequency, lead time, change failure rate and recovery time: the one book here with a defensible empirical claim about what makes teams fast.
Connected ideas
Says something similar
Complexity is what you must hold in your head to make a safe changeJudge a design by how much a reader must understand to change it safely, and prefer deep modules behind small interfaces.
Concepts
Put it to work on L3VLUP
More like this
- FrameworkCapital allocationBook
A chief executive is a capital allocator first and an operator second
Judge a chief executive by what they do with the cash, not by how well they run what produces it.
From The Outsiders · William N. Thorndike
- TechniqueSoftware engineeringVideo located
An index turns a table scan into a handful of page reads, paid for in write speed and disk
An index buys read speed with write speed and disk; add one for the query you run most and remove the ones you never use.
From B-tree vs B+ tree in Database Systems · Hussein Nasser
- FrameworkLeadershipLetter
Day 2 begins when a company manages to proxies and treats decisions as one-way doors
Check whether the metric is still the goal or has become the goal, and reserve the heavy process for decisions that cannot be undone.
From Amazon's 2016 letter to shareholders (Day 1) · Jeff Bezos
- FrameworkSoftware engineeringVideo located
Consensus is agreeing on one value when any machine may crash and the network may lie
Ask what a consensus protocol assumes about crashes and the network; its guarantee is only as strong as those assumptions.
From Distributed Systems 6.1: Consensus · Martin Kleppmann (Cambridge)