Even someone as successful and established as Anil Dash struggles with this: https://t.co/r4ZjnkQdvH
— Rachel Thomas (@math_rachel) October 26, 2018
Even someone as successful and established as Anil Dash struggles with this: https://t.co/r4ZjnkQdvH
— Rachel Thomas (@math_rachel) October 26, 2018
This is a very real challenge. It's helpful to have people you trust that you can get feedback from directly, and to mute/avoid toxic channels, but there's no failproof solution. https://t.co/wIx6W16jOl
— Rachel Thomas (@math_rachel) October 26, 2018
“Ask for help. Don’t feel like you ‘should’ know something because you have a certain level of experience.” Great point by @kaelen_medeiros at #noreastr18. There will always be parts of data science and #rstats new to you, but you’ve got a friendly community here to help. pic.twitter.com/QqiwEoubD1
— Emily Robinson (@robinson_es) October 26, 2018
Lesson 1: Megaprojects go better when you prepare thoroughly. Alas, Brexit has - for understandable but purely political reasons - been rushed. Zero preparation before the referendum - as bad as it gets.
— Tim Harford (@TimHarford) October 26, 2018
Lesson 3: Pick a highly experienced team. Alas, our highly experienced negotiator quit early in the process.
— Tim Harford (@TimHarford) October 26, 2018
Similar to how, say, you can always reinvent the Pythagorean theorem on the fly if you think about geometry through the lens of vector products, or how you don't need to memorize the quadratic formula if you understand what an equation is and the general process for solving them
— François Chollet (@fchollet) October 26, 2018
The same is true of backprop in deep learning -- knowing how to code up backprop by hand gives you no useful knowledge wrt deep learning, and inversely, developing powerful mental models for deep learning does not in any way require knowing the algorithmic details of backprop
— François Chollet (@fchollet) October 26, 2018
Many people in engineering believe that to understand something, it is necessary and sufficient to have a low-level mathematical description of that thing. That you need to "know the math behind it". In nearly all cases, it is neither sufficient nor at all necessary - far from it
— François Chollet (@fchollet) October 26, 2018
The compute and data moats are dead
— Sebastian Ruder (@seb_ruder) October 25, 2018
This is a thoughtful article by @Smerity that makes a super important point people tend to forget: Compute and data advantages rarely matter in the long term. You can contribute to the field even with limited compute.https://t.co/xCAaYXAHnH
“Natural language is the ultimate compression”—Pete Warden https://t.co/kUeDjduaOo
— Stanford NLP Group (@stanfordnlp) October 25, 2018
"It is easy to lie with statistics, but it is easier to lie without them." Frederick Mosteller #statistics #biostatistics #Bioinformatics #DataScience
— Edward Tufte (@EdwardTufte) October 25, 2018
This is the other reason I worry about an MCMC-first Bayesian resurgence: it's unclear to me that the benefits relative to "just optimize the log likelihood and use Fisher information" justify the >100x performance costs. https://t.co/ksSdGX39dZ
— John Myles White (@johnmyleswhite) October 24, 2018