#48 | Big Work Change, Calling BS On Data, Covid Testing
|Benjamin Joffe||Oct 6, 2020|
Ben’s irregular newsletter mixes work and play.
Unsubscribe at the bottom / Sign up here
Find me: Twitter - LinkedIn - Medium - Website - Podcast
Howdy! It’s been an eventful month, with a big change for me.
1. WORK: On Leave, Fresh Content, SOSV Most Active VC in Q2
2. EXPERIENCES: More Coding, More Thieves, Friending the dead
3. CULTURE: Octopus Love, Calling BS on Data
4. THOUGHTS: On Games
5. COVID LIFE: Art, Nomads, Testing
After 7 years with SOSV I’m taking a leave of absence to explore startup projects by participating in the Entrepreneur First (EF) program in Paris. I don’t have a team nor project yet, but felt the startup itch and decided to scratch it.
If you’re not familiar with the EF model: they select ±60 highly motivated tech and biz people, and help them form startups. If EF likes the team and project after a few months of that, they write a first check.
I’ll be focused on that and somewhat available until year end, so you’re welcome to say hi — online or offline!
After a series of interviews of deep tech investors on the Lab To Market podcast, I’ve asked some industry experts and entrepreneurs to give a kind of ‘crash course’ about their sector.
The latest episodes cover Mental Health, Fashion & Textile, and Robotics.
SOSV was ranked the most active seed VC globally in 2Q2020 by PwC/CBInsight, right during the peak of the pandemic.
Opentrons, a lab automation company in the SOSV portfolio (and in which I am an angel investor) signed a contract with NYC to process up to 20,000 tests per day by November, for only $28 per test (see article in the NYT and NYC website). This will boost significantly the testing capacity, and might be a model for other cities.
SOSV’s IndieBio program has been running several live events on key biology-related topics, from epigenetics to plant biology and food systems and of course, Covid-related things. The videos are here (I ran one on Startups against Covid-19 a few months ago with investors from Khosla Ventures and Fifty Years).
IndieBio’s colleagues Po Bronson and Arvind Gupta (now partner at Mayfield) just wrote a book titled ‘Decoding the World’. Buy it or read the first chapter on Amazon and/or join the launch talk on Thursday online here. Biology will save the world!
Coding and the Risks of Data
I’ve completed the Automating The Boring Stuff With Python course on Udemy, and built some throwaway things. It was very satisfying.
It also got me interested in playing around with image recognition and look into machine learning and data science. (thanks to S. for recommending other Udemy courses — Andrew Ng’s course I had started on Coursera didn’t use Python and felt outdated).
I read about gradient descent algorithms and was happy to see I could handle the maths (partial derivatives of polynomial expressions). Note that most people skip the maths and simply use the libraries.
It also got me to understand better the problems of local minima, overfitting and Simpson’s Paradox (more on that later — this one was quite important).
Almost Robbed (Again)
In my previous newsletter, I told the story of how two thieves stole my phone: one stepped on my foot to distract me, while the other lifted it from my back pocket. I shared how I caught the thief red-handed and retrieved it (note: those ‘pros’ generally give it back easily, to avoid involving the police), as well as the thoughts that ensued about how best to react (still debating), and my approach to prevent it from happening again: I sewed my short’s back pockets.
I am happy to report that more sewing work awaits: I am pretty sure I was almost robbed again recently.
I was in the subway in a transfer corridor when I noticed two young people at the bottom of a flight of stairs.
They seemed to be the only ones chatting and not walking to anywhere.
I started to walk up the stairs. Half-way, I decided to speed up as I was in a hurry.
This is when I noticed the man behind me, who stopped and walked back down to his friend.
That’s when I put two and two together.
Of course I can’t tell for sure, but I’m glad I seem to have improved my awareness — maybe I spent too long in East Asia … (I’ve lost and found back valuables several times in Japan, for instance, while I’ve heard countless stories of theft in Paris).
Our 4th lendmeyourcat resident went back to his home. I was inspired one evening and decided to paint the beast resting in its lair.
The process went like this: fail with direct watercolor => draw raw sketch => use gouache => cat moves => regret not having drawn a better sketch, and not using masking tape.
The result wasn’t too shabby, especially in terms of mixing colors and being bolder with paint. As my favorite YouTube painter James Gurney says: 'Painting is not about putting down the perfect stroke, it's about correcting the ones that are wrong'. That surely applies beyond painting!
A friend reached out to get help for his son’s Ivy league application, as the latter was struggling with his ‘personal statement’. I had never done anything like it, and learned about this odd ‘humble bragging’ exercise, and its cottage industry. This kid has already done many coding projects, speeches, and even started a company. The only weakness is that he hasn’t faced as much struggle as seems popular in such essays. Whatever the result, I think any college would be lucky to have him!
JiuJitsu Stop & Go
The gym had barely reopened that it closed again due to Covid. I had just enough time to sprain my ankle, after a successful bout where I baited my opponent, grabbed his back and won by submission.
I thought all was fine after training at 2pm. By 7pm I couldn’t walk. Fortunately, a few days with Rest, Ice, Compression and Elevation (‘RICE’) did the trick … then the gym closed again. Ah!
Friend The Dead
I was going through friend requests on Facebook (I barely use Facebook) and accidentally accepted an old request of someone whom I know passed away 2 years ago. It gave me a strange feeling. Maybe future wills should cover our digital legacy as well?
Cinemas are open in Paris. It might sound either irresponsible, bold or … so 2019. Anyway, I went to watch a movie.
Family Romance, LLC**
This is a ‘stylized documentary’ by Werner ‘Carry Bolt-Cutters Everywhere’ Herzog. It focuses on a real Japanese company where people can hire ‘actors’ to hand difficult social situations (apologizing at work, replacing an absentee father, etc.). I had only read about this company in the news a while ago, but Herzog packed his stuff (and 2 family members I think), went to Japan, and shot a documentary. I didn’t find it outstanding but I always enjoy opportunities to use my Japanese language skills.
Overall, such service is not cheap, but should probably exist everywhere. Is Japan ahead of the curve, or — like often — exploring a future nobody else will experience?
The Good Fight***
My favorite black-owned law firm show is back on Netflix. The first episode was pretty surprising, though the rest felt less interesting than in previous seasons. Still, the acting remains top notch.
My Octopus Teacher***
Sometimes Netflix suggests something intriguing. This is the story of a man who falls in love with an octopus, and live a year-long romance until the latter lays eggs and dies eaten by a pyjama shark. This documentary has beautiful cinematography and a slow, charming pace that highlights the strangeness of this underwater world.
The Spanish Apartment**
This is the first opus of a tryptic by French filmmaker Cedric Klapish. This one felt a bit dated and cheaply produced, but the story of this student who goes to Spain to prepare for a government job and shares a place with a bunch of international students had some charm.
The Goat (‘La Chevre)**
A French movie starring Gerard Depardieu and another well-known comedian. The latter plays a chronically unlucky guy sent on a mission to find the chronically unlucky daughter of his boss, by following his bad luck. Cute though dated.
I watched this animation series about a ninja school. For those familiar with Naruto (the manga sold >220 million copies so it’s quite something), Boruto is his son. The dad was an orphan ostracized as his body contained a demon, and his son is mostly burdened by the fame of his absentee dad, who became the boss of the ninja school protecting the city.
The novices complete missions in teams of 3, and test various teams to find the best combination. I found this had eery similarities with the approach of Entrepreneur First to team formation … Eventually the anime uses classic ‘hero’s journey’ tropes, but the challenges, grit and lessons the kids go through were quite enjoyable.
Calling Bullshit on Data****
It takes 10 times as much effort to clean up the bullshit than to create it.
— Brandolini’s Bullshit Asymmetry Principle
I watched a great series of lectures on YouTube titled Calling Bullshit: The Art of Scepticism in a Data-Driven World, by professors at the University of Washington Information School (h/t to B, who recommended the book).
The talks go over many important issues, to help sort the vast amounts of BS we are facing every day across all media. Here is what I found the most interesting:
Are we bullshit-neutral? Check your social media sharing and what you repeat around you.
The importance of statistical significance (p-value) and its widespread neglect.
The problem of false positives and false negatives.
The lack of reporting of negative results of experiments in science, and the incentives of scientists to seek recognition.
The most shocking learning for me was Simpson’s Paradox. In short: statistics can be totally misleading if you don’t know the data sets and the domain they come from. I urge you to read it as it could literally mean life or death (one example is about how to choose a cancer treatment).
I was already familiar with quite a few of the concepts and biases thanks to the classic How To Lie With Statistics, and from my recent studies of machine learning and data science, but it was still very much worth it.
The Psychology of Money***
My friend B. recommended this book, and I’ve been listening/watching some of the corresponding talks.
For most of us, our way of thinking about money results from the exposure to ideas we had in our early years (up to our mid 20’s?) mostly via our parents, our social class, and the economic climate and opportunities at the time.
But what worked or didn’t in the past at different times (S&P, real estate, bonds), might not work the same today and in the future.
There are several talks online by this author.
Rick Beato on Pop Music***
Rick Beato is a musician and producer who analyzes famous songs and does all sorts of interesting commentaries. I watch this channel on occasion, though I stopped playing guitar a long time ago. His recent take on why boomers hate pop music was educative: his analysis was that songs are quite ‘formulaic’, have little to no chord and rhythm variations, and use autotune and Midi drum sounds from the 80’s. He even mentioned a similarity with nursery rhymes. That said, he wasn’t entirely negative and highlighted a few of his favorite pop songs. In another of his videos, I liked the take by Joe Satriani (a bona fide guitar hero) who said each generation has to make their own music.
After playing many rounds of Pandemic (mobile and board game versions), I gave a shot at another board game called Wingspan, where you’re some kind of bird specialist, and have to collect bird and lay eggs to win (!).
Playing made me think about game mechanics. I now put them in 3 broad categories (boardgamegeek has its own take on categories):
Competition: it’s a direct confrontation, or a battle for resources (e.g. Monopoly, Catan, Chess, Risk, Diplomacy).
Cooperation: everyone is allied to reach a goal (e.g. Pandemic).
Corridor run: every player does its thing independently and scores are compared at the end (e.g. Bingo).
Then there is the mix of luck and skill. Too much luck and you’re a mere ‘passenger’; too much skill and the fun goes away if levels are not matched, as an inferior player sometimes has zero chances of scoring points, let alone winning.
Note: It has even been observed that among some animal species (dogs, rats…), larger animals would occasionally let their smaller play-mate win so they keep playing. I also don’t think it’s ok to pretend to lose (and lie) with children — maybe starting with a handicap would do?
Now you can think about your favorite games with Comp/Coop/Corridor and Luck/Skill. Then look into your playing style according to Bartle’s taxonomy: Killer, Socializer, Achiever or Explorer, as not every player plays to win.
What I noticed:
Competitive games often create tension, and can turn unfair quickly if, for example, two players team up against a third. Some make those dynamics an integral part of the game: Diplomacy is almost pure negotiation.
Cooperative games are pretty rare but generally quite interesting and much more peaceful. It also feels good to work on a common goal.
Corridor games can be more interesting when you can observe how others play in parallel, and learn new strategies.
Finally, all those games have to face the problem of learning the rules and general onboarding. Some take a few minutes, some can take up to an hour (or more).
Few games have solved this ‘learning curve’ problem.
Some onboard with a simplified version of the mechanics, and gradually complexify it gradually.
The duration of a game can also matter.
Note: I co-designed a card game with my nephews this past summer using similar cooperative mechanics: each card is an activity with ‘fun points’ (fun for kids) and ‘bothering points’ (annoying for adults) e.g. ‘stuff your mouth with food’ is +2/+4, ‘play with kickboard’ is +6/+2, ‘take a walk’ is 0/-3. The goal is to get as many fun points as possible before the ‘bothering points’ reaches 10 (= adults get angry). The design phase included interviewing their grandparents about what annoys them the most. My goals there were:
(1) Show my nephews they could create a game.
(2) Raise their awareness about their daily actions.
(3) Give them perspective on what makes a game fun.
What I didn’t expect is that the game we made turned out — after some tuning — to be quite playable, to the point they sometimes volunteered to play it!
Tommy Two Misses
I came across the story John T. Thomson, most famous for his invention of the Thompson submachine gun or ‘Tommy Gun’ after WWI. The war was over and instead of the army, he sold to law enforcement. A few years later sales were low and he was replaced as a director. He died in 1940 (aged 79), before the US entry into WWII, which lead to large orders.
I have heard conflicting views among doctors, within governments, and in the media — and there was the Lancetgate. I’m not a doctor and only sharing what I think in good faith — hoping to be ‘bullshit negative’ or neutral. Everyone is welcome to make their own opinion, and send me useful sources.
The most viral thing about Covid might be the artwork. Read the story here on how two medical illustrators at the CDC created it (h/t Kickstarter). This art has been seen billions of times!
Spoiler: it’s not a faithful microscopic virus image, it’s a rendition. Just like many others before it, it is designed to ‘put a face on the enemy’. Everything, including the colors, have been carefully chosen for this purpose.
Do you know some people who relocated because of Covid? I know a few who packed their bags—sometimes overnight—to flee a gloomy confined life.
Will that trend continue beyond 2020? I would think so, as remote work seems here to stay, and opened new options for knowledge workers. Some entrepreneurial types are even considering building independent cities (floating or not) to welcome the displaced (elites).
My current view is:
Immunology tests (= detect the presence of antibodies in the blood) are quite unreliable. According to this study across 38 studies, early detection (<2 weeks) with those tests is very bad, and has lots of false positives.
PCR tests are much more reliable in principle. The process implies taking a swab, then ‘amplifying’ the virus genetic material (doubling it dozens of times). The real problem is: how much is enough, and how much is too much? Is it 30? 35? 40? and what is the right threshold to say someone is ‘positive’? The New York Times wrote about this.
“In three sets of testing data that include cycle thresholds, compiled by officials in Massachusetts, New York and Nevada, up to 90% of people testing positive carried barely any virus, a review by The Times found.”
“In July, the lab identified 872 positive tests, based on a threshold of 40 cycles. With a cutoff of 35, about 43% of those tests would no longer qualify as positive. About 63% would no longer be judged positive if the cycles were limited to 30.”
If you combine this issue with the fact that most countries expanded from testing mostly symptomatic people to now a majority of asymptomatic people, it seems it can only lead to a massive number of false positives.
Of course, having some false positives is acceptable to prevent the spread, but too many has important downsides (see this recent Lancet publication).
The falling number of deaths seems to support this idea of false positives, and possibly of a lower fatality rate with the current virus strains.
As a reference, deaths in France between Jan 1st and Apr 30th was about 25% above 2019 and 17% above 2018. A large difference. The number of deaths in France in 2020 since May 1st are almost identical to 2019 and 2018 (<2% difference).
It’s not an easy public health and political exercise, but I hope we see soon more reason in policies and resources.
A parting thought for Autumn 2020:
‘Fall with masks, shorts turn long.’
(applies to the weather and the stock market)
Until next time!