How to fix friends and inference people

Medicinal articles to treat a bingo board of bad bosses and misinformed coworkers

Let’s face it. If you work in data, you’ve met that coworker who makes you want to facepalm (or even facedesk?) on a daily basis. Whether it’s the pointy-haired boss who is surprised to learn that you can’t do data science without data, the statistical inference person whose pedantic nitpicking gets in everyone’s way, the become-a-data-scientist-in-3-months “hotshots” who have no idea that they have no idea, the science-obsessed leader who thinks numbers can’t lie, or the salesperson whose pitch overshoots science fiction to sound more like I’ve-never-even-heard-of-science fiction… you’ve likely been tempted to scream into a pillow.

Frustrated souls, my first newsletter is for you. Here’s my cunning plan: I offer you a set of 30 blood-pressure-raising symptoms (in no particular order) that I’ve encountered during my career - never all in one person, since the list runs the gamut from ignoramuses to pedants to their mutant offspring - and invite you to play bingo to see how your experience compares with others.

Where’s the cunning part? Perhaps one of these is causing your team a special headache. Let’s fix that! Each symptom comes with a link to an article that you can send the offending entity to help cure them. We’ll try to be sneaky - the official cover story is that you’re just sharing these articles to “entertain” your friends... that’s why I tried to make them as entertaining as possible.

Be sure to use the original links I’ve provided, since they sneak you past the Medium.com paywall.

Ready? Let’s play bingo!

Count how many of these you’ve encountered during your career and let me know your score here (link repeated at the bottom).

Symptom 1: They have no idea what AI is, but they talk about it anyway

Treatment: bit.ly/quaesita_sbucks and bit.ly/quaesita_simplest

Symptom 2: They believe machines + data + math = objectivity

Treatment: bit.ly/quaesita_aibias and bit.ly/quaesita_tiger

Symptom 3: They make you cringe whenever they say “because”

Treatment: bit.ly/quaesita_experiment and bit.ly/quaesita_correlation

Symptom 4: Their ignorance of the data science job market is impressive

Treatment: bit.ly/quaesita_bubble

Symptom 5: They expect machine learning to solve all their problems

Treatment: bit.ly/quaesita_fad

Symptom 6: They wouldn’t know a data-driven decision if it punched them

Treatment: bit.ly/quaesita_inspired

Symptom 7: They want to apply statistics *everywhere*

Treatment: bit.ly/quaesita_pointofstats and bit.ly/quaesita_saddest

Symptom 8: They think analysts are “lesser” data professionals

Treatment: bit.ly/quaesita_hero

Symptom 9: They never split their data… and it shows

Treatment: bit.ly/quaesita_sydd

Symptom 10: They get excited about obviously-doomed ML/AI use cases

Treatment: bit.ly/quaesita_island and bit.ly/quaesita_parrot

Symptom 11: They’ve missed the point of hypothesis testing

Treatment: bit.ly/quaesita_damnedlies

Symptom 12: They’re constantly second-guessing the person in charge

Treatment: bit.ly/quaesita_incomp

Symptom 13: They think algorithms are the most important part of ML/AI

Treatment: bit.ly/quaesita_fail

Symptom 14: They say “data science is the sexiest job of the 21st century”

Treatment: bit.ly/quaesita_22

Symptom 15: They trust ML/AI solutions that other people have built

Treatment: bit.ly/quaesita_donttrust

Symptom 16: They don’t see what’s so hard about getting useful data

Treatment: bit.ly/quaesita_phillla and bit.ly/quaesita_provenance

Symptom 17: They insist that “numbers can’t lie” and “facts are facts”

Treatment: bit.ly/quaesita_confirmation and bit.ly/quaesita_scientists

Symptom 18: They’re obsessed with hiring PhD researchers

Treatment: bit.ly/quaesita_roles

Symptom 19: They’re only willing to trust AI if “it can explain itself”

Treatment: bit.ly/quaesita_xai

Symptom 20: They think leading a team of data scientists is easy

Treatment: bit.ly/quaesita_dsleaders

Symptom 21: They take an analyst’s “insights” seriously

Treatment: bit.ly/quaesita_versus and bit.ly/quaesita_datasci

Symptom 22: Their presentations are gooey with jargon and equations

Treatment: bit.ly/quaesita_noeqns and bit.ly/quaesita_speaking

Symptom 23: They’re scared of AI for the wrong reasons

Treatment: bit.ly/quaesita_ethics and bit.ly/quaesita_genie

Symptom 24: They expect a data pro to know “the everything” of data

Treatment: bit.ly/quaesita_universe

Symptom 25: They nitpick constantly and can’t get enough of “rigor”

Treatment: bit.ly/quaesita_battle

Symptom 26: They don’t see the point of doing validation (properly)

Treatment: bit.ly/quaesita_idiot

Symptom 27: They think unsupervised learning is unsupervised

Treatment: bit.ly/quaesita_unsupervised and bit.ly/quaesita_drugs

Symptom 28: They start ML/AI projects by calling in the nerds

Treatment: bit.ly/quaesita_first

Symptom 29: They won’t shut up about how terrible p-values are

Treatment: bit.ly/quaesita_needles

Symptom 30: They tell lies with data and call it “data storytelling”

Treatment: bit.ly/quaesita_inkblot

*************************************************************************************

So, what’s your score? How many of these are painfully familiar? Let me know here!

Connect with Cassie Kozyrkov

Let’s be friends! You can find me on Twitter, YouTube, and LinkedIn. Interested in having me speak at your event? Use this form to get in touch.