I think @NateSilver538 confuses people because he’s legitmately, deeply, interested in what will happen in the future.
— David Robinson (@drob) November 20, 2019
He’s neither “Bernie will win because America is tired of lies” nor “Trump will win because America is easily fooled”
He’s actually interested in the question! https://t.co/bT6mUbrTXR
“Junior researchers affiliated with less prestigious institutions show the most benefits from coauthorship with a top scientist.”
— hardmaru (@hardmaru) November 20, 2019
So science is like the music industry?
I hope to see ‘indie science’ become more of a thing.https://t.co/eFPgtjSVCS https://t.co/DqCEM6v1Ay
There are 2 big cultural failures here:
— Angela Bassa (@AngeBassa) November 19, 2019
1) folks feeling comfortable with insufficient context: “It’s not my job to understand why this needs to get done”
2) folks not feeling safe enough to gather context: “I’m too scared to ask questions” https://t.co/bOsMz7hOhD
This is such an important and underappreciated point. Simple, rule-based approaches can be good for consistency, fairness, & transparency. https://t.co/6ReoUNs3Pn
— Rachel Thomas (@math_rachel) November 19, 2019
Hear how we cut $500,000 in cloud spend costs on the latest episode of The REWORK Podcast: https://t.co/5Fzpt2g8EG
— Jason Fried (@jasonfried) November 19, 2019
Personally identifiable information (PII) is the new plutonium:
— Brandon Rohrer (@_brohrer_) November 19, 2019
* difficult and expensive to store safely
* powerful if painstakingly refined
* valuable on the black market
* can destroy lives accidentally if mishandled
* can destroy societies if weaponized https://t.co/i0wMU6yE2S
“The world is its own best model. It is always exactly up to date. It always contains every detail there is to be known. The trick is to sense it appropriately and often enough.”
— hardmaru (@hardmaru) November 19, 2019
— Rodney Brooks (Roboticist)
Bing uses a 3-layer BERT-like transformer for every query. In addition, to model simplification and possibly distillation, they implement more efficient GPU code. https://t.co/9irfYz1DCw
— Leonid Boytsov (@srchvrs) November 19, 2019
3 factors that have greatly contributed to disinformation:
— Rachel Thomas (@math_rachel) November 17, 2019
- audience consolidation (gives propagandists a ready audience)
- personalized targeting
- gameable algorithms (helps disinfo jump from social media to mainstream press)#TechPolicyCADE @noUpside
Sampling bias occurs when not all elements have the same probability of appearing in the datasets. Police records are not a representative sample of crime due to variation in:
— Rachel Thomas (@math_rachel) November 17, 2019
- reporting rates
- police attention
- arrest rates (eg Black people more likely arrested for marijuana) pic.twitter.com/VWfEJW0pUJ
“People are usually afraid of change because they fear the unknown. But the single greatest constant of history is that everything changes.”
— hardmaru (@hardmaru) November 17, 2019
— Yuval Noah Harari
You can't just defer to data. You have to keep inspecting your results, looking for feedback loops, iterating, and visualizing in order to identify & mitigate risks. -- Deven Desai #TechPolicyCADE
— Rachel Thomas (@math_rachel) November 17, 2019
