Comments (2)
  • Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

1 CindyVanEpps commented Permalink

Its interesting to me that we judge the world of continuous delivery by the Facebooks and Etsys and "born on the web" companies. Recently when I was unable to add someone to a Facebook group, rather than go to the trouble of reporting it, I just waited a few days and tried again. It was more annoying on April 15 when I couldn't submit my tax return with TurboTax. Then I only had the tolerance to wait 15 minutes each time I tried again. And it eventually worked. But I didn't report a problem; I assumed someone else would.
But will I use this same wait and see approach when my bank transfers money to the wrong place, or my stock trading system goes rogue? Absolutely not! So the nature of the systems you are testing is a key influencer.
I conjecture that reputational risk will become the major driver in demanding quality in all systems and our tolerance for "oops and loops" will drive our loyalty as customers.
I believe you are right, Monica, that we will have to look more at the risk of what we test and what we don't. But it will vary greatly by application and I the pendulum has to swing back a bit for the right balance to be achieved.

2 monica914 commented Permalink

Great points Cindy and it's why I'm also thinking about what's important to test. I totally agree we have to focus testing & the automation on the critical failure points. And the rules are different for systems of record (bank balances) versus systems of engagements. I cover some of that in this earlier blog: https://www.ibm.com/developerworks/community/blogs/rqtm/entry/what_does_testing_mean_in_the_context_of_continuous_delivery?lang=en