"Design It Twice"?(16elt.com)
16elt.com
"Design It Twice"?
https://www.16elt.com/2024/09/14/design-it-twice/
7 comments
I like this idea, but this felt like half an article. I thought the second section was setting up a detailed explanation of "the time [they] got it wrong" and I was really interested to learn about it, but instead it goes straight into a conclusion (and a plug for their telegram channel!) without offering any real evidence or examples to support the idea.
I’m more of a fan of “do it again”.
Start a project, get to 80%, rewind and start over from the same beginning.
Several times this has led to much superior solutions and of course the second implementation happens at a much faster pace.
Start a project, get to 80%, rewind and start over from the same beginning.
Several times this has led to much superior solutions and of course the second implementation happens at a much faster pace.
>>I’m more of a fan of “do it again”.
>>Start a project, get to 80%, rewind and start over from the same beginning.
This, definitely! I thought the original article will advocate just that but it argues for a silly thing: "Come up with at least two different designs even if you have to force yourself. By comparing these different design, you can find the best approach."
Well, news at 11: this is how EVERYBODY does it the first time and still it misses the mark. It's trying to guess the future by extrapolating the present and the results are "Cars of the Future": https://www.youtube.com/watch?v=RUWMH75THow
Only by living that future you can go back and re-design that software properly.
>>Start a project, get to 80%, rewind and start over from the same beginning.
This, definitely! I thought the original article will advocate just that but it argues for a silly thing: "Come up with at least two different designs even if you have to force yourself. By comparing these different design, you can find the best approach."
Well, news at 11: this is how EVERYBODY does it the first time and still it misses the mark. It's trying to guess the future by extrapolating the present and the results are "Cars of the Future": https://www.youtube.com/watch?v=RUWMH75THow
Only by living that future you can go back and re-design that software properly.
I agree. The design needs to hit a tiny bit of the real world sometimes to really figure out why it sucks.
Great read!
Design it exactly Twice. If you cant afford that design it once. Twice will also not be enough. Attempt to approach a local maximum because all design has competing values. Design for yourself, or someone you like more, or at least someone with a lot of money.
Dont fall in love with your first design and spend days weeks years developing it before getting feedback, or comparing it to the second design you never made. Also make sure the committee all gets a say and all their feedback is in the new design, to approach the global maximum design that eats all other designs.
Be sure to ship a design, but also a good one, that's better than the other one. On budget. With good taste.
Design it exactly Twice. If you cant afford that design it once. Twice will also not be enough. Attempt to approach a local maximum because all design has competing values. Design for yourself, or someone you like more, or at least someone with a lot of money.
Dont fall in love with your first design and spend days weeks years developing it before getting feedback, or comparing it to the second design you never made. Also make sure the committee all gets a say and all their feedback is in the new design, to approach the global maximum design that eats all other designs.
Be sure to ship a design, but also a good one, that's better than the other one. On budget. With good taste.
Comes out to the same as "Throw away the first one", doesn't it? Who was that, Fred Brooks?
well, carpenters "measure twice, cut once"...
also, when many people write an argument, then will also write a counter-argument to be thorough.
So "design it twice" might be ok.
Would there be other techniques that could help, like some sort more adversarial second design?
also, when many people write an argument, then will also write a counter-argument to be thorough.
So "design it twice" might be ok.
Would there be other techniques that could help, like some sort more adversarial second design?