We really should talk more about why Nate Silver is not particularly liked by a lot of people who actually work with data.
"I would bet 60-40, 70-30 on lab leak probably"
This is just throwing random numbers on his hunches. He does this _all of the time_ and doesn't generally disclose (or even necessarily know to disclose) which pieces are his "hunches."
He mixes things that he puts into a model and fails to mention any validation beyond topline as well
hachyderm.io/@hrefna/113331563…
Hrefna (DHC) (@hrefna@hachyderm.io)
"Data journalism" equals traditional punditry except you put arbitrary numbers on everything.Hachyderm.io
))




Matt Campbell
in reply to Matt Campbell • • •Matt Campbell
in reply to Matt Campbell • • •Matt Campbell
in reply to Matt Campbell • • •Matt Campbell
in reply to Matt Campbell • • •Matt Campbell
in reply to Matt Campbell • • •Diversi-Tune Apple IIGS demo video
divtune.comPeter Vágner reshared this.
Matt Campbell
in reply to Matt Campbell • • •I suppose the whole process, and my own mind, is still contaminated because I initially used an LLM. I can think of the following things that remain from my usage of Claude:
- The use of FFV1 as a lossless video codec. Presumably I would have found that in the FFmpeg documentation if I had really looked.
- The pixel format conversion to yuv420p to fix compatibility problems. Claude gave me that one for the MP4 version, and then I also applied it to WebM.
- The use of 256 Kbps for AAC.
Matt Campbell
in reply to Matt Campbell • • •Matt Campbell reshared this.