Goji Says

Writing is not transcribing and talking is not recitation. Both are acts of creation.
The kitchen container paradox: you always need one more container than you have.
You may have heard that most published research is false (Ionnadis). But what you probably don’t know is that most corporate data science is also false.
If people weren’t lazy and would infer the quality of the article from the article alone and not the journal, the rush to submit to prominent journals would abate. Fix the supply by demanding academics carefully read articles they cite.
Data science operations in most companies are like paid apprenticeship programs except that the blind lead the lame.
Statistics in corporate decks fails silently.
The returns on noise and confidence are absurdly large.
The four major axes of sanity are physical health, good mental habits, opting into good people, and work.
Anything worth doing is worth automating. (This is my take on the G.K. Chesterton classic.)
Here’s one despairing summary of life: “You hop on one leg. You hop on another leg. And then you die.”
It is easier to mislead with data (‘science’) than to lead with data science.
Imagine like a king, execute like a janitor.
Statistics is the discipline of converting lots of numbers into fewer numbers.
Statistics is the art and science of optimally measuring distances.
The canonical insight from Computer Science is: automate repeated actions. The canonical insight from Data Science is: optimize repeated actions.
The fact that infinite precision doesn’t exist in nature—unit right triangles exist even though we cannot write down the length of the hypotenuse—is provoking. At the risk of ridicule, it leaves open the door that we are living in a computer simulation.
Data science doesn’t begin with data. It begins with a question.
Quality ~ f(effort, intellect, competence, attention to detail, number of independent points of view)
If you are a ‘central’ technical (ML, engineering, etc.) team, there are only three kinds of businesses you should be in: a) the pipeline business—taking away abstracted out pain of multiple teams, b) public goods business, and c) ‘Moses as a service’—what is the ‘best’ way of doing X.
People mistake well-cast roles for good acting.
The goal in science is to learn the right thing, not to be proven right.
Wealth from owning land is mostly wealth from regulation, except where land is used for agriculture or mining.
White-collar work is not a market for lemons. It is a market for back scratchers.
Far better to be proven wrong than to be wrong.
The past is here — it’s just not very evenly distributed. A take on William Gibson’s saying, “The future is already here — it’s just not very evenly distributed.”
One way to know if you ‘know your data’: Can you report the means of key variables of the top of yr. head? How about s.d.s and cor.?
The key limited resource is attention. Spend it freely if you want to produce something sensible, lasting.
Picking nits in your writing is important. It prevents readers from scratching their heads.
The first rule of dealing w/ bureaucracies and customer service: if the first person doesn’t give the answer you like, talk to the next person.
In Silicon Valley, there is a myth that you (only) win by serving users’ interests the best. No, you “win” by creating addiction too.
High-level programming languages have robbed much meaning from the aphorism: “Unless you have programmed it, you don’t understand it.”
Q. for agnostics: Do you consider yourself an agnostic about tooth fairies and Santa Claus? If so, god help you. If not, what’s so special about god?
The 1st rule of citing research in science = tl;dr.
The 2nd rule of citing research in science = if it supports what you conclude, tl;dr, otherwise state why you can’t trust the results.
The ns don’t justify the means.
Sometimes people distinguish between the academic and the real world. They don’t understand that academic production is ‘real’ — too real.
The world is shaped by energetic people, only a small fraction of whom have good ideas.
Scientific production is like an Indian road—a bullock cart next to a stray cat next to an ‘autorickshaw’ next to a Volkswagen. And the Volkswagen is fiddling with its pollution meters.
The MAO of learning: Motivation, Ability, and Opportunity.
One way we overcome our own limitations — others.
What most people mean when they say ‘s/he is good’ is that ‘s/he is good to them.’
When ~3% of pop. produces all the basic necessities, there is a good chance that many, if not most, of the 97% are in the entertainment sector.
Aside from whatever differences we inherit, we are all sums of our own investment, and what others invest in us. The timing of the investment matters immensely.
Any communication platform where all information is born ‘public’ reduces to a platform for elites (marketing).
D.C. is the site of the largest run government welfare program. Except it is mostly for those earning over $100k.
Wide acceptance is no insurance against fundamental problems.
The only questions you get to ask in life: what can you do to improve your game, and what can you do to help others.
A great many people have the dough for happiness but never bake the cake.