| I'm not an oncologist. But I'm married to one. So she tells me these rules on an almost daily basis. Also, many of her cancer patients don't make it - and then she tells me - this person didn't make it because he/she broke this rule #xyz in that list. I mean, ok, I get it. I work in ML in an enterprise. So everyday we have our usual litany of complaints. This item is not ranked correctly. Boosting isn't working on that item. Human overrode the item recommendation and now the faceting is broken. Search is running out of memory. P2 alert!!! program crashed and we are losing $10000 per hour wake up the programmers in the middle of the night... So I can also play this rule game. 1. Don't use computer. 2. If you must use computer, don't program. 3. If you must program, please let it not be in Java. 4. If it is Java, please don't call Python ML code from it. 5. Better yet, don't use ML at all if you have humans overriding recs. 6. Don't cache everything to reduce latency. You'll run out of memory. 7. If you run out of memory don't wake up the fucking programmer in the middle of the night when he's trying to get some action. and so on...I mean, what purpose does it serve ? You can't get programmers in 2024 to stop using hadoop and sprintboot and whatever ancient enterprisecrap. You expect to get Americans to eat less meat, drink less, maintain healthy weight ? With an official obesity rate of 39%, the non-cookedup version hovering in the 70s ? Fat chance. |
I see what you did there.