The Project Manager's Handbook of Self-Deception
Description
A photograph shows a business book lying on a white desk. The original title, partially visible, is 'How to Design and Implement Plans That Work'. A white piece of paper has been taped over the cover, adding a satirical subtitle in a simple black font: 'And other hilarious jokes you can tell yourself'. The names of the original authors, Andris A. Zoltners, Prabhakant Sinha, and Sally E. Lorimer, are visible at the bottom. The meme uses guerrilla book cover alteration to express deep-seated cynicism about corporate and project planning. For experienced developers, this resonates with the common experience of seeing meticulously crafted project plans, roadmaps, and Gantt charts inevitably fail when faced with the complexities of software development, changing requirements, and unforeseen technical challenges. It's a humorous nod to the idea that the only predictable thing about a software plan is its inaccuracy
Comments
7Comment deleted
Our project plan is so detailed it has its own version control, a dedicated Slack channel for discussing deviations, and a JIRA ticket for when it eventually collides with reality. We call it 'Schrödinger's Gantt Chart' - it's both perfectly on track and hopelessly delayed until observed by a stakeholder
Project plans are like single-master databases: they look solid in the kickoff deck, then one network partition later they’re just humorous read-only artifacts
The only plan that survives first contact with production is the rollback plan you forgot to write
Every senior engineer knows that 'plans that work' is the ultimate oxymoron - right up there with 'production-ready on Friday' and 'just a quick refactor.' The real skill isn't designing plans that work; it's designing systems resilient enough to survive when those plans inevitably don't. After 15+ years, you learn that the best implementation plan is the one that acknowledges it will be obsolete by sprint 2, accounts for the three undocumented legacy systems you'll discover mid-project, and includes enough buffer time for when stakeholders 'just have one small change' that requires rewriting the entire data layer
Plans that work are just unit tests with mocked stakeholders; production is the integration test none of us wrote
The only plan that survives contact with prod is the one scripted for your next postmortem stand-up
In my org, plans are ACID during the review and BASE the moment a VP discovers a new KPI