Ben’s irregular newsletter mixes work and play.
Unsubscribe at the bottom / Sign up here
Find me: Twitter - LinkedIn - Medium - Website - Podcast
3 months since the last letter - how time flies! Also it’s #50 which made me dig out the first one from June 2014 (scroll all the way down here). Anyhow, here are my news:
1. WORK: Back to work
2. STUDY: Coding, Clubhouse gear, BJJ, Chess
3. EXPERIENCES: Ikebana & more gardening
4. CULTURE: In Treatment, Fleabag,
5. THOUGHTS: Things around science
6. COVID LIFE: Stats, Sewage, Stocks
My 3 months break during which I joined the Entrepreneur First program ended without a startup project. It was a very stimulating experience but falling in love with both a long-term (viable) idea and a co-founder in 3 months is not easy.
Participants coming from research were looking for a viable market, while those coming from software and business had almost ‘too many options’, and ideation was a challenge. I also wish I had been a bit more systematic in ideation, so I eventually looked deeper into it and wrote — a bit too late — How To Find Startup Ideas. Eventually, one third of the participants ended up getting their project funded by EF. Best wishes to them!
Back to work
After this break, I am back at SOSV. The pandemic has been difficult for many startups, but several in biotech, robotics, e-commerce and edtech have been thriving. I am now working on internal systems, marketing, PR, networks and more with Ned Desmond, who joined SOSV after many years as COO of TechCrunch.
I also put some of my coding study to good use and built an internal tool that will help with our PR efforts. Funny enough, I thought about using Python but eventually it runs entirely on Google Sheets thanks to its scraping functions. The fund had some great news in its bio, automation, edtech, e-commerce and blockchain/crypto portfolio.
We’re on Clubhouse
Look for ‘SOSV Startup Forum’ — our bio and Asia teams are already running weekly events. I might run some events soon too (follow me: @benjoffe).
Data Loves Comedy
I’ve been studyingPython on and off since last summer and got curious about text processing. As an exercise I picked one of my favorite standup routines, the ‘State Abbreviations’ by Gary Gulman to see whether I could figure out something from it. Here is the resulting medium post. I tweeted it to the author, who had nice comments and even retweeted it. The folks at The Startup (apparently the largest medium publication I never heard of) republished it.
I’ve been playing a bit with the latest Silicon Valley media app darling.
I came across conference-worthy talks, and some on the niche-est of topics. It’s like the indie radio era on overdrive, with social and access on top.
I had to get an iPhone to join and picked a refurb SE 2000 which did the job.
To play further, I heard the new Apple M1 chip was able to run iOS apps, so I got myself a Mac Mini (and bought some Apple stock after hearing a data center CEO saying — on Clubhouse — they bought hundreds due to their low heat and power usage). Sadly, Clubhouse doesn’t run fully on the M1 (you can’t search clubs or enter rooms), but it was still useful to experiment.
I kept the Mini for coding, which lead to finally getting a proper screen (a basic curved Samsung, as Xiaomi’s sexy gamer screen was sold out).
I got also a Logitech keyboard and mouse (the M570 trackball — update of a mouse I used 20 years ago — it takes very little desk space as you don’t need to move it around. No good for gamers, though)
…Then I got chair envy from a friend and ordered a SecretLab gamer chair (I’m no gamer).
This is how you end up spending $2k just because you’re curious about something…
With gyms closed nationwide I now take private lessons and practice with friends at home, which is great for fitness, fun and sanity. The ‘closed guard’ system from Danaher videos is finally starting to come together. Beware the pendulum sweep and the top lock!
Watching “The Queen’s Gambit” reminded me I used to play chess in a club as a teen and led me to try the Lichess app.
It was a humbling experience: the app gives you the benefit of doubt with a 1500 rating initially (the lowest is 600, highest around 2800, average 1550).
I started playing 5-min ‘blitz’ and quickly dropped below 1000 (!). Now I’m crawling back up to 1400. I guess I just wasn’t that good? (update: I do much better in 10-min ‘rapid’ and I am close to 1700 in that style — maybe I am just slow…).
One interesting idea that came out of now close to 1,000 games played is that action brings luck. A half-planned but sustained attack can uncover new opportunities. I think this also applies well to Jiu-Jitsu and startups: trying leads to either success or learning, and more confidence in trying more.
What to do when most everything is closed during lockdown? How about an ikebana class with your local Japanese florist?
90 minutes learning about the proportions you need between the three main branches (shin, soe and tai) in this particular style (shajitsu moribana) and the balance and angle of the flowers and leaves (balance the ashirai with mikoshi but don’t do shitakusa because that just looks bad — noob mistake). You even get to take them home!
The live flowers have now made space for dry things and there was a notable drift in the rules, but the spirit is there.
More Happy Gardening
There was suspicious white dust on my new screen. The Xiaomi air purifier was indicating an alarming 400+ in air quality (zero is best, >100 is not good). Had someone smuggled in some 2008 Beijing air?
It took me a while to connect the dots: we have plants at home, and some are not happy as the air is too dry, so I bought the Xiaomi humidifier which vaporizes the water filled in.
We use tap water, which is said to be pretty ‘hard’ in Paris (>15°f), between 20 and 30°f, where 1 °f (degré français) corresponds to 10 mg of calcium per liter (in fact, CaCO3). The brave humidifier vaporizes the calcium with the water, which then floats in the air, settles on my screen and likely in my lungs, too.
So a battle started between the silent poison of the humidifier, and the air purifier on full blast.
Because of the noise I have to use headphones, and end up speaking quite loudly, which is not to everyone’s liking.
Eventually I had to have the appliances come to an agreement: the humidifier is now on only intermittently, and only in tandem with the purifier, ideally with windows open. I wanted to organize this mess but Xiaomi doesn’t have documented APIs for their devices… Update: I just found this open source Python library and might try it.
Movies & Video
This is a show about psychotherapy: we follow several patients during their sessions with one analyst. The original version came out in Israel in 2005, the US did a local version in 2008, and I watched the French version (‘En Therapie’) that just came out in 2021. France is often a bit behind…
The sessions gradually peel the onions of why people behave the way they do, helping them recognize and solve some issues. While it goes much faster and the analyst talks much more than in real life (it’s TV), the method is quite faithful and might help your own thinking. I might watch the US version later on (and maybe the Israeli one?).
Fleabag (Amazon Prime series)****
A 30y.o. woman and her dysfunctional life, family and relationships. As if Tim from The Office (UK) and Carrie from Sex & The City had a millennial baby girl. If your life is difficult right now, it might help you feel better. The show was originally a one-woman standup show.
Buster Keaton movies****
All his pre-1925 movies are now in the public domain. I was impressed by the camera work, and the physical prowess and creativity of his stunts in The General, Sherlock Jr.and One Week. Jackie Chan mentioned him as a major inspiration. Here is an additional interesting interview.
Seinfeld*** on Amazon Prime
I finally watched a good chunk of this famous 90’s show. While entertaining, I am still partial to Curb, by Seinfeld’s co-author Larry David.
Sherlock (Netflix series)***
I watched distractedly but it was well produced and acted by Tim/Bilbo as Watson and Turing/Smaug/Cummings as Sherlock (Benedict Cumberbatch apparently struggled with the pace and memorization of his Sherlock character). The editing is trying to reflect Sherlock’s mind — don’t watch if you’re prone to epilepsy! Note that the ‘memory palace’ is a well documented memorization method that a journalist tried for himself and won the US memory championship.
Don is the man if you’re looking for something indie and very strange in animation. interesting works. I bought his 3 shorts on Vimeo that talk about weird time travel (a bit a la Dark). Another is David O’Reilly (check his free works Please Say Something and The External World), who worked on the video game sequence in Her.
The Imitation Game***
Alan Turing and his friends make a machine to break German codes. The movie covers Turing’s demise due to his homosexuality, illegal at the time in the UK. Overall it was good but all fairly predictable. The contribution of Polish code-breakers was mostly ignored and made Poland angry. If you like genius stories check out A Beautiful Mind or Good Will Hunting.
The Great Hack (Netflix documentary)**
I heard of this documentary about Cambridge Analytica on a podcast with the French AI scientist Luc Julia, co-inventor of Siri, talking about the number of data points that will have on any teenager by the time they reach adulthood. The documentary focuses on two defectors / whistleblowers, and a professor who requested (unsuccessfully) to retrieve his personal data. Are we ‘persuadables’?
It strongly reminded me of the (even more impressive) ‘How to start a revolution’ documentary on the ‘nonviolent revolution’ expert Gene Sharp. His methods have been used globally to overthrow governments (in the Middle East, Eastern Europe, and more) and generally serve the Western/US agenda.
Fun fact: Gene Sharp was nominated 4 times for a Nobel Peace Prize, and while a favorite (?), he lost in 2012 to the European Union (!). Of course there are hundreds of nominees each year (see nomination process), and the lists include Hitler, Stalin, Mussolini and Michael Jackson. Some members of the European Parliament suggested Donald Trump for 2021, with an interesting list of reasons. The committee will decide!
Hold Up** on Netflix
This indie and crowdfunded French documentary was made to reveal the lies and corruption around Covid and recorded several million views within a few weeks. Their channel has now been removed by YouTube. The research and experts were very hit or miss. Still, I think ‘light is the best disinfectant’.
Coach Carter** on Netflix
The eternal story of a ragtag team of misfits and ne’er-do-wells that rise to glory thanks to a tough coach (is it one of the ‘Hero[es] with a Thousand Faces’?). It worked well enough, once again.
Snowpiercer Season 2** (note: it’s still frozen) on Netflix
I’m still watching this dystopian series about the remaining humans stuck on a 1000-cars train in a sub-zero world. I am yet to read the French 80’s graphic novel it’s based upon.
Sword Art Online Alternative: Gun Gale Online**
Another scifi-ish thing, this time about immersive VR games. I thought ‘Ready Player One’ was a bit simplistic, this Japanese anime’s setting is a bit better: Sword Art Online was an immersive VR game connected to your brain but the creator got crazy and locked everyone in the game, killing thousands IRL as they were dying in the game.
This sequel is about an unusually tall girl who looks for self-acceptance in VR playing a tiny and cute character, but ends up accidentally joining a massive Battle Royale game and becoming one of the top players. The show is ok but doesn’t go as far as Denno Coil (about AR) and Summer Wars (the better ‘Ready Player One’).
Cop** (French: Flic)
An undercover journalist became an official ‘assistant policeman’ after only 3 months of training. The ‘regular’ French police training used to be 12 months, but is down to 8 since June 2020 due to the lack of candidates. The duration varies around the world: 10 to 36 weeks in the US vs. 3 years in Finland and Norway (equivalent to a bachelor degree).
The author had done previous undercover investigations as a factory worker, a call center staff, etc. This case took him 2 years. A lawyer friend told me he now uses the book as evidence of cover-ups of police brutality. The book also shows the sense of meaninglessness due to arrests that don’t get prosecuted, the lack of recognition (both symbolic and financial), public hostility, and the necessity to display a macho strength. All recipe for systemic disaster.
In that environment it is no surprise that a recent bill trying to outlaw people recording of police officers caused a public outcry and was dropped by the senate. If anything, more transparency is needed to rebuild trust. And better and longer training, rather than rush more underprepared officers onto the streets.
Yann Le Cun Talk
Another French A.I. luminary, this Chief AI Scientist at Facebook, gave a talk on AI history. Most went over my head but I took note of a few things:
"Building your own tools can give you superpowers that other people don't have". One reason pioneers made progress was because they built tools. Open sourcing helps a lot now.
Some research themes are in fashion and some are not: when he got interested as a student in neural networks, he couldn’t find anyone to fund his thesis. Eventually his professor told him ‘I’ll sign but you’re on your own’. The next decades were also on and off with a 'neural net winter' in the late 90’s. I saw parallels with the ‘financing risk’ of startups: even great teams with great tech targeting great markets can get the cold shoulder from investors.
One anecdote: the French aviation pioneer Clement Ader, whose ‘Eole’ bat-inspired airplane flew in 1897, years before Wright brothers. Le Cun says it didn’t get recognized due to his high secrecy, and didn’t improve as he was ‘hypnotized by biology’ (bats with folding wings). This online discussion has additional elements and labels the Wright Brothers as ‘the first patent trolls in the history of aviation’.
This Le Cun conference triggered additional thoughts about ‘Science’. Today it is invoked almost like a magical word, sometimes with little substance. It also has religious aspects with its politically-approved priests, heretics and blasphemy.
In addition, it is one of those words, that looked upon closely, is not quite so clear: what is science? Theory? Experiments? Le Cun described the interplay between theory and empiricism (observation) and how you need both. More, what makes a science ‘soft’ or ‘hard’? Upon reflexion, I think it’s testability and repeatability (which social sciences, for instance, have trouble with, but also geology, etc.), but also time frames: the lifespan of a trend, a human, the Earth and our galaxy are quite different.
On the topic of science, the 2020 Nobel Prize in Chemistry went to two women (including a French) who discovered ‘one of gene technology’s sharpest tools: the CRISPR/Cas9 genetic scissors’. It is truly a remarkable discovery, and they very much deserved this prize. There has been almost as many women awarded in the past 20 years (2001-2020: 28) as in the past century (1901-2000: 30) — Chemistry: 6, Physics: 4, Medicine: 13, Literature: 16, Peace: 17, Economics: 2. Surely many more to come!
Data Loves Covid
After a friend criticized my ‘back of the envelope’ calculation of Covid impact on annual deaths, I did a more detailed statistical research in this medium post to quantify how many excess deaths could be attributed to Covid (in France). In short: it’s noticeable but not huge (two standard deviations above a model of the past 15 years), and might balance out in future years due to a ‘harvest effect’. This article was picked up by a data science focused Medium publication.
Paris firefighters were able to track the progress of Covid by measuring its concentration in sewage water. It’s low cost, reliable and can act as an early detection system (as long as it doesn’t rain too much). The half-life of the virus there can be 20 days at 4C, and 1-2 days at summer temperatures. Luckily, the virus is too diluted in sewage to be contaminating.
The first info on this in France was available as a pre-print back in April. This article quoted similar research done in China months earlier, and even for SARS back in 2005! I find remarkable that ‘pre-prints’ might accelerate research (and increase scrutiny). This measurement in particular was quite reliable, with no apparent bias, conflict of interest, or misalignment (aside from seeking attention). Maybe another area the Internet might change.
Last, maybe some sewage samples could help identify the origin of Covid-19—even the WHO is not sure it started in Wuhan, and it could well be that, like the Spanish Flu, China was simply the first one to report it. Will they find — I quote — a ‘smoking bat’?
Are Policies Effective?
This Stanford research paper analyzed the effect of stay-at-home and lockdowns on early contaminations in the US. While those contributed somewhat to the reduction, the most effective predictor is (unsurprisingly?) population size and density, which induce higher social distancing. Note: of course all this goes hand in hand with broad testing.
Sometimes it feels like governments are a bit like goalkeepers who, despite having a statistically higher chance to stop the ball during a penalty kick by not diving, feel compelled to pick a side (as they would feel worse missing if they didn’t jump). This concept might apply to stock trading too (‘investments do better than investors’), but how many of us can sit on their hands in good and bad times?
Also, a study in Denmark with 6,000 participants, only half of which wore masks, showed no significant difference in the infection rate.
It seems lots of contaminations happen at home (only critical cases end up in true isolation or in hospitals). Does confinement make sense then?
I used to travel a whole lot and I feel I’m getting ‘cabin fever’.
Data seems to indicate people below 60yo are at very low risk (<5% of total deaths in France).
Figuring who died primarily of Covid is not easy: underlying conditions and incentives to over-report (e.g. Reuters fact check on UK numbers).
Some UK economics researchers did an early cross-country comparison to estimate the ‘price of life’ associated with an early or delayed lockdown. The results may surprise you.
Vaccines are there. The big mRNA experiment is ongoing in the West. Personally I would be more partial to a more traditional type using attenuated or inactivate forms of the virus like in some Chinese vaccines (e.g. Sinovac and Sinopharm), but geopolitics will probably prevent this choice.
Finally, if you have liquidities today, it might be difficult to find attractive Covid-era assets… Real estate? (maybe in ‘remote work cities’?) Stocks? (tech is very high, but some more traditional stocks might be unfairly judged — I also put some $YOLO into $GME and got out during an upswing — I can’t understand it well enough). For non-tech stocks I looked into banks, cruises (there is large pent-up demand) and some FMCG. Angel investments are also possible for the brave and patient — one of mine has been thriving during Covid but it took a global pandemic to get to this inflection point. Gold and Crypto are not quite for me yet but Elon likes some coins!
A bit more patience and we’ll be fine. Diamond hands, everyone!