It's really interesting to be on the other side of this.
Because I'm around programmers, I usually hear programmers explain how some other field is simple, but in this thread we have the opposite, someone from another field explaining how programming / engineering is simple
Because I'm around programmers, I usually hear programmers explain how some other field is simple, but in this thread we have the opposite, someone from another field explaining how programming / engineering is simple
https://twitter.com/jimnotwell/status/1478847543647752196
IMO, one of the most common mistaken beliefs I see is the belief that, outside of one's own field, the world is understood.
It's easy to see that, in one's own field, the world is not understood but, for some reason, people don't realize this is also true of other fields.
It's easy to see that, in one's own field, the world is not understood but, for some reason, people don't realize this is also true of other fields.
An example from the hardware world is Intel's Copy Exact methodology, where they tried to make every fab identical.
It's not obvious this is a good idea because it's very expensive to do this since conditions (land, etc.) are different in every location where you have a fab.
It's not obvious this is a good idea because it's very expensive to do this since conditions (land, etc.) are different in every location where you have a fab.
Also, people at other companies (AMD, etc.) argued that making every fab a copy of "fab 0" and having people train and fab 0 and copy the exact process elsewhere would stifle innovation and creativity since people at other fabs wouldn't be able to innovate.
With the benefit of hindsight, we can see that Copy Exact worked well and, despite recent stumbles, Intel wiped the floor with their competitors on the fab side of things for decades.
Copy Exact was a key enabler of this. Here's a story I was told that illustrates why:
Copy Exact was a key enabler of this. Here's a story I was told that illustrates why:
When Intel opened their first fab in AZ, yields were low and efforts to fix it didn't work. Eventually, someone realized they failed @ Copy Exact because the AZ fab didn't run its dehumidifiers.
When they injected humidity into the air and ran the dehumidifiers, yields improved.
When they injected humidity into the air and ran the dehumidifiers, yields improved.
I believe it was later theorized that the reason yields improved was that the dehumidifier system was removing ammonia from the air, but with Copy Exact you don't need to know why something works, you just need to make sure you replicate the exact conditions.
BTW, I made this same error in one of my first posts: danluu.com/hardware-unfor….
I hadn't yet worked in software and I argued that hardware had a bunch of nearly impossible to convey implicit knowledge, but of course once I started working in software I saw it wasn't so different.
I hadn't yet worked in software and I argued that hardware had a bunch of nearly impossible to convey implicit knowledge, but of course once I started working in software I saw it wasn't so different.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
