Skip to main content


When I read about these things I always think about some of the writing of @pluralistic on graceful failure modes. A product (system) is not defined by its success but by how good or poorly it fails. I've been teaching students that not considering (poor) failure modes is a huge liability.

arstechnica.com/gadgets/2024/1…

#failure #scalability #devops #management #governance

in reply to Koen Hufkens, PhD

Many of these problems originate from shifting failure modes from a focus on weak-link problems to strong-link problems, to increase profit.

Weak-linked problems are defined by their worst performance, while strong-linked problems are defined by their best performance. Problems don't reside strictly in either category, but when dealing with infrastructure (which isn't an easily replaced discretionary purchase) the focus should not deviate too far from a weak-link assumption.

in reply to Koen Hufkens, PhD

i used to have an electric bike for which the only way to turn on the builtin lights was via an app. I got rid of that bike.

(UPDATE: the bike was a Blix. Don't buy a Blix bike. For more reasons than just that)

This entry was edited (3 weeks ago)
in reply to steev hise

@detritus Seeing #bicycles being infected by this kind of technological approach hits deep 🥺
This entry was edited (3 weeks ago)
in reply to Jorge Toledo

@eldelacajita @detritus This is worth the read within the context of #enshitification and the cycling industry (only talking shifters). What is the cost of convenience, and what are we willing the bear to not have things fail on us in unexpected ways. Many who are informed are now considering these questions. I will probably never use electronic shifting.

bikepacking.com/plog/cost-of-c…

in reply to Koen Hufkens, PhD

Not only is this poor failure handling on the engineering side, it highlights a further issue: the attitude of "it doesn't matter, we'll refund the users if it happens" forgets that failures are likely to lead to consequential losses and even physical danger - or ignores that because of a click-through get-out clause
in reply to Koen Hufkens, PhD

Had a discussion about this with a fellow veteran-engineer this weekend, about how designers are neglecting the basic #UX concept that systems should fail to manual.

I have a touch-activated faucet in my new home which, when batteries fail, just doesn't let you have water. Fail-to-manual is a much more sensible approach than locked out of your car, or can't have water.

@pluralistic