Rob Moffat
Rob Moffat Author of Risk-First Software Development. Developer. Working in the UK.
  1. Failure Modes
  • Understanding Failure: what exactly does it mean to fail?

  • Personal Failures

    • CapsLock: complexity, not using tools.
    • Configuration Tool (Complexity, feature fit, bugs in hibernate (dependency risk, then dead-end risk), difficulty mapping domain model)
    • Wide Learning (Funding, but also complexity), did we know what we were building? Agency risk
    • AreAye - needless complexity XMLBox
    • Agora: Notes / Typing. (Complexity Risk) Archipelago
    • PDC: website redesign. funding. i.e. schedule risk
    • Hawk: complexity risk in the software. but actually, they made it work. offshoring.
    • Dark: market/feature fit?
    • J10: marketing / market fit / Complexity in spades. algorithmic complexity
    • DSL: complexity (code generation). complexity = layers. team dynamics.
    • REF: complexity. agency risk. failure of goals. m&t.
    • REF Testing: complexity risk. communication risk?
    • HSC: Trader Comments: feature fit.
    • HSC: Takeover of Symph: Complexity (of change)
    • TT: Feature Fit
  • Boehm.

https://wwwx.cs.unc.edu/~welch/class/comp145/media/docs/Boehm_Term_NE_Fail.pdf

https://www.worksoft.com/top-software-failures-of-2017-so-far

https://sites.hks.harvard.edu/m-rcbg/ethiopia/Publications/Top%2010%20Reasons%20Why%20Systems%20Projects%20Fail.pdf

JC Example 1: Compensation Workbench, rules to uplift payrolls annually based on certain definitions. e.g tied into performance gradiing, min wage. Was done off s/s, out of HR system. Analysis done on S/s. But wanted to do it in the HR database. It took a lot longer than it should. Underestimated the complexity. Excel could be infinitely complex, but if you standardize, you lose that. The customer hadn’t got experience of big IT projects - scope creep. No real sense of prioritisation and focusing on what was iportant. PM was nice, but didn’t understand delivery management. (i.e. managing risks and issues). Had short timescale, lead delivery resource underestimated. 2m exercise. Lasted 5m in the end, but not with full functionality.

Add Your Star On GitHub to receive an invite to the GitHub Risk-First team.