The summary.
MULTIPLY YOUR ESTIMATE BY THREE
Creative work estimates are unreliable and solo work has no slack. Choose a game whose content requirements you can actually meet.
Never defer saving — it shapes architecture and retrofitting it is painful.
AGREE OWNERSHIP IN WRITING WITH EVERY CONTRACTOR
Without explicit assignment you may hold only a licence. Start with a small paid test piece, and keep source files rather than exports.
RECORD EVERY ASSET'S SOURCE AND LICENCE AS YOU ADD IT
Reconstructing that later is impractical, and licences are checked at submission.
Make the distinctive things yourself; use purchased assets for background and secondary content.
MOST UNFINISHED GAMES WERE ABANDONED, NOT DEFEATED TECHNICALLY
The middle is where they die. Keep something playable at all times, record weekly progress, and protect a sustainable pace — exhaustion ends more projects than difficulty.
Restarting to do it properly usually means avoiding the unglamorous remaining work.
EFFORT ALREADY SPENT IS NOT A REASON TO CONTINUE ANYTHING
Not a feature, not a project. Set reassessment points in advance, so the decision feels planned rather than like giving up.
NEVER SHIP A HALF-IMPLEMENTED FEATURE
An obviously incomplete system damages perception of everything else.
