Best of Twitter - Week of February 15, 2021
These data are so fascinating... I remember when Byron Reeves was first starting this approach during my PhD.
But does 'screenome' & 'screenomics' signal the invasion of 'nome' to social sciences?
nature.com/articles/d4158…
ignore meta studies because the studies themselves are garbage
Daniël Lakens @lakens
^ https://darkblueheaven.com/spatialsoftware/
Introducing my unofficial Clubhouse client for Android, Windows, and Mac.
If you don't have a Clubhouse invite, You can
reserve your username
browse topics
and search for users and clubs
If you do have an invite, you can also join rooms as a listener!
hipster.house
->
@staringispolite It is that bad. It doesn’t require a token for you to join a room. So if you’re signed in you’re a real user, otherwise you can just fill in the info in the http request for your user account and join (there is no auth). This is because the rooms don’t require auth.
^ wtf lmao
Wow, South Korea TFR was 0.92 in 2019. I knew it was low, I didn’t know it had further dropped off a cliff from the low ones. Japan is 50% higher! People don’t pay enough attention to culture here.
virgin netflix: $m’s of IP, sfx, exposed to free market, harvests pixel-level data on human attention to create algo-gen max engagement, gets average 4.8 on IMDB
chad BBC: makes same cop drama w/ same 5 actors for 40 years, state monopoly, “what’s cinematography?”, 8.5 on IMDB
I think about this Steve Jobs quote every time I eat great salami. Not quite software ratios, but easily 10:1.
I'm sure part of the success of deep learning is its ability to maintain the practitioner engaged through watching training curves go down. The nets are actually training us to train more nets and propagate their existence.
More
Following up on the selection effects tweet from last week, Edward links to his great essay exploring the theme more broadly and discussing how it leads us to misinterpret the data we look at: https://marketingbs.substack.com/p/marketing-bs-selection-effects-88c
Correction to last week’s tweet about chimpanzees’ working memory (h/t Stephen Malina and Gleb Posobin):
Hello! Primatologist and evolutionary anthropologist here to share this PSA:
THIS TAKE IS INCORRECT.
Because, as it turns out, the result was better explained by training (the chimpanzees had tons more experience with the task) than species-level differences. Pls share 1/n
Brian Roemmele @BrianRoemmele