Ideas of Geniuses (Idées de génies) by Etienne Klein and Gautier Depambour

From time to time, I blog about science and mathematics. Here is a new example. I just discovered a little wonder of popular science, at the same time simple, luminous and demanding. Ideas of geniuses, (Idées de génies) subtitled “33 texts which have shaken up physics”, by Etienne Klein and Gautier Depambour.

Etienne Klein is also the producer on France Culture of the excellent Scientific Conversation. I had already referred to it in connection with a post about Alexandre Grothendieck and another with Gérard Berry.

Through short texts, the authors make us discover ideas of genius like for example that of Galileo who explains and proves why one or even two kilograms of lead will not fall faster than a kilogram of feathers.

“In free and natural fall, the smaller stone does not weigh on the larger.”
When you place a large stone on a scale, not only will it weigh more if you lay another stone on top of it, but adding a wick of tow will increase its weight by the 6 or 10 ounces that the stone will support; but if you freely leave the stone and the wick attached together from a certain height, do you believe that in the movement the wick will weigh on the stone, so that it should accelerate its movement, or do you believe that the wick will slow down the stone, supporting it in part? We feel a weight weighing on our shoulders when we want to oppose its movement; but if we were falling at the rate that that weight would naturally drop, how do you expect it to lean and weigh on us? Can’t you see that that would be the same as wanting to injure someone with a spear who is running in front of you at a speed equal to or greater than the speed you are chasing? Conclude, therefore, that in free and natural fall the smaller stone does not weigh on the larger, and therefore does not increase its weight as it does at rest.

Galilée, Discorsi e Dimostrazioni matematiche intorno a due scienze attenenti alla mecanica ed i movimenti locali, 1638.

Bright, isn’t it? It also reminds me of Einstein’s inspiration for his theory of relativity although I have yet to read the sections relating to this other genius. All the chapters I have read are of the same style … A must read!

Food delivery startups – a quick analysis

Yesterday I discovered an Indian startup filed to go public on its national stock exchange. I did not know it, shame on me. Zomato is the latest filing in a small number but extremely visible services in the food delivery sector. Who does not know Deliveroo, Just Eat, Uber Eats and others.

I am not sure that twenty years ago I would have pu the sector in technology innovation, but I have to admit innovation has many faces. On Crunchbase, the sector is said to have 616 organizations with $12.5B in funding (see Crunchbase Food Delivery Startups). Traxn gives the largest players here. So thanks to my database of 777 cap. tables, I could have a look at some statistics, with the exception of FoodPanda (acquired by Delivery Hero, funded with $318M including Rocket Internet), iFood (Brazil, owned by Movile, $587M in funding) and Swiggy (India, $2.5B in funding including Accel). Here are the data at time of the IPO filings.

Start-up JustEat GrubHub Zomato Delivery Hero Deliveroo DoorDash Median *
Geography United Kingdom Illinois India Germany United Kingdom Silicon Valley
Founded Aug-01 Feb-04 Jan-10 May-11 Aug-12 May-13 Sep-02
IPO / exit Apr-14 Apr-14 Apr-21 Jun-17 Mar-21 Nov-20 Mar-13
Years to IPO 12,7 10,2 11,3 6,1 8,6 7,5 7,4
VC Amount 88 86 1529 1711 1856 2578 84
1st round 8 1 1 4 4 2 4,4
Sales ($M) 150 137 338 297 1 680 885 23
Income ($M) 10 15 -310 -202 -312 -668 -14
Market Cap. 1 465 1 988 6 782 4 700 10 220 17 384 560
PS 10 15 20 16 6 20 17
PE 147 133 102
Nb of Emp. 886 680 3 469 12 098 2 561 3 279 189

*: the median value is based on 777 startups compiled over years. PS and PE are the ratios of market caps to sales and earnings (profit)

Just Eat has no data on founders as the initial Danish company with 5 founders has been bought and moved/launched in the UK.

Here are the data to date (GrubHub has been acquired by Just Eat in 2020):

Market Cap. $B Sales – $B Loss – $M PS Employees
Just Eat 15,4 2,8 -151 5,4 9 000
Delivery Hero 39,4 2,0 -939 19,5 35 528
DoorDash 46,0 2,9 -458 15,9 3 886
Deliveroo 6,3 1,6 -225 3,9 2 060
Zomato 6,8 0,3 -310 20,1 3 469

What is interesting is the difference in dynamics between companies launched before 2010. Something not really new if you follow this blog, in terms of growth dynammics. Here are some more data about shareholders

Start-up JustEat GrubHub Zomato Delivery Hero Deliveroo DoorDash Median *
Found. 5,7% 6,2% 4,5% 7,1% 12,0% 9%
Emp. 9,4% 25,1% 7,0% 17,2% 18,9% 18,5% 21%
Emp. shares 9,3% 12,1% 3,1% 11,7% 9,4% 4,8% 8%
ESOP-granted 0,1% 9,7% 0,6% 5,5% 9,5% 11,8% 8%
ESOP-reserved 3,3% 3,3% 1,9% 5%
Dir. 0,2% 0,06% 0%
CEO 1,0% 3%
VP 0,2% 1,1% 0,2% 0,2% 0,8% 1%
CFO 0,2% 0,3% 0,2% 1%
Investors 66,0% 59,3% 70,2% 59,9% 74,0% 68,7% 51%
IPO 24,6% 9,7% 16,6% 18,4% 0,6% 16%
Total 100% 100% 100% 100% 100% 100% 100%
Nb of Dir. 2 2 2
Dir % 0,1% 0,03% 0,2%
Nb of found. 2 4 3 2 3 2
Found. % 2,9% 1,6% 1,5% 3,6% 4,0% 4,5%
Found. age 27 31 31 33 23 37,6
F1 28 27 31 33 29
F2 26 29 33 21
F3 31 20
F4 38

The founders are young, own little. How teh sector will develop, I do not know. There is already concentration. Barriers to entry look low. Some experts have doubt about long term profitability… tough to say. Deliveroo shows no IPO shares as the initial filing did not include any new shares. It may have changed at the recent IPO which was not a success; and here are the individual cap. tables.

Charles Geschke, co-founder of Adobe, dies at 81

Charles Geschke may not be as famous as many Silicon Valley entrepreneurs, but he is really a legend of technology and software. With John Warnock, he cofounded Adobe in 1982 and he is an exception in the group of founders as he was in his early 40s when he left Xerox to create the company which developed Postcript, PDF, Adobe, Photoshop and so many more products. He died on April 16, 2021.

I had read about Geschke (and Warlock) in a good number of books and blogged about him here:
In the company of Giants in November 2008
A success story: Adobe Systems – John Warnock and Charles Geschke in March 2009

I found yesterday this very interesting short video that you should watch (or read the transcipt below).

Here is the transcript: “when John and I started Adobe we had a sort of simple thought in mind in terms of how we wanted to organize the business

we wanted to build a company that we would like to work at and we sort of used that principle in terms of how we defined the management structure, the operational organizational thoughts that we had about how to put the company together and part of that recognition was that we had constituencies that every business has that need to be balanced

we had our shareholders we had our employees we had our customers and of course the communities in which we operated and if you think about running a business all four of those constituencies are mildly in conflict what’s good for one may not be good for the other and to be successful as a business and to retain the kind of quality employees that you want to have it’s extremely important that you continually monitor how those four constituencies are being served and keep them in balance

a couple other principles people would often ask as they worked at Adobe

well what will it take for me to enhance my career and both John and I would tell them well the first thing you have to figure out is how to fire yourself by hiring someone to work for you who can do your job better than you do and there’s no alternative but to get promoted once you do that

and the second piece of philosophy which is frankly the most critical one building particularly a high technology business is to tell every engineer and every manager that your job is to hire people who are smarter than you are as it’s a much larger population from which to choose and those turned out to in fact have been extremely important in building a business

and again I want to comment that without my relationship with John and our partnership it’s hard to imagine that we could have achieved what we did over the past twenty seven years and we’re extremely pleased with this award and the opportunity to have served our industry

thank you very much

the really cool thing about this award is it’s from engineers we’ve gotten entrepreneur awards before but they’re from business guys and getting award for entrepreneurship from engineers is very very cool

Chuck and I when we started the company weren’t on a get-rich-quick scheme we were frustrated at Xerox and our major frustration is we knew we could invent great technology but no one would ever use it and it would never see the marketplace and I think our major motivation in starting Adobe and continuing with Adobe was to make stuff that people would use a lot of people would use and I think inside of every engineer is that basic need to have your stuff used so that was the primary motivation behind what we did

the other thing in the hardest thing about building a business and keeping a business sustained and going on is continuing to innovate and we’ve never figured out how you institutionalize innovation innovation is a very sort of remarkable thing that sometimes happens sometimes doesn’t but the best you can do is build an environment where people are happy they’re doing an adventure they’re trying to create and hopefully in that process great things will happen”

Knowledge, skills and personality of entrepreneurs

A friend (thanks Kevin!) just retweeted the following: What kind of Knowledge, skills and personality traits are common to successful entrepreneurs?

I tend to agree 100% but I may have an idealized view of my own experience! It also reminded me another quote from the same period (2011 vs. 2010) by Steve Blank: Over the last decade we assumed that once we found repeatable methodologies (Agile and Customer Development, [Lean Startup], Business Model Design) to build early stage ventures, entrepreneurship would become a “science”, and anyone could do it. I’m beginning to suspect this assumption may be wrong. It’s not that the tools are wrong. Where I think we have gone wrong is the belief that anyone can use these tools equally well. In the same way that word processing has never replaced a writer, a thoughtful innovation process will not guarantee success.” Blank added that “until we truly understand how to teach creativity, their numbers are limited. Not everyone is an artist, after all. The full interview can be found on

and also Komisar: “I think there’s stuff you can’t possibly learn in school and I’m not even sure you can learn that on the job. There’s an entrepreneurial character. Some people have it and some people don’t. Some people may not think they have it, and they may have it. A lot of people they think they have it, and many don’t.”

Coursera files to go public (#750)

After Deliveroo yesterday, here is Coursera’s cap. table. It’s #750 in my long list of startups (see here my most recent analysis – data about 700+ startups).

Coursera and Udacity are probably the most famous MOOCS companies and I could not be surprised if they contributed to create the edtech category. Coursera just filed to go public on Nasdaq so its numbers are available.

Revenues of $293M, with a loss of $66 million in 2020. A lot of venture capital since its foundation in 2011, $464M in total from Kleiner Perkins (and legendary partner John Doerr, and NEA.

Founded by two Stanford professors, specialists of artificial intelligence, Daphne Koller and Andrew Ng, age 43 and 34 at the time of foundation. They are not managing the company anymore. No information about Koller’s shareholding probably because she owns less than 5% of the company and has no executive role.

Coursera founders Andrew Ng and Daphne Koller are computer science professors at Stanford University

Source : NPR

Deliveroo plans to go public

There was a lot of buzz today about Deliveroo announcing its IPO soon. By the way Coursera, the edtech company just announced it too and I will post about it next. So I had to build its cap. table and thanks to the openness of the British register of companies, I could do it (at least partially) even before the company filed its IPO document.

Interesting data I think about the company growth, its funding and the founders stakeholding. £1.3B invested to cover £1.1B loss, 2’500 employees in 2019 and £1.2B revenues in 2020. Among the best European VCs (Index, Accel) plus Amazon, Fidelity, DST, T.Rowe Price as late stage investors. What else?

PS (March 19): a former colleague mentioned an article saying that early investors would have made “60’000 per cent return” on their investement. At the same time, I discovered about Coupang in South Korea which looked similar to Deliveroo. So I also checked the multiple return for seed investors in Coupang. Here is first its cap. table.

So for Coupang, the initial price per share was $0.02, and I assumed a $35 at IPO, which makes a 1750x multiple.

For Deliveroo, I assume a price per share of £900 with a series A price at £8.36. It is true there were also seed shares at £1,5. This would be a 600x (or 60’000 per cent, not an annuela return though)

So Coupang is even better than Deliveroo…

The Mom Test by Rob Fitzpatrick

The “Mom Test” is an intelligent book for any student of Steve Blank and his “Customer Development” model: validating hypotheses to launch a startup by exploring the existence of customers and of a market, of course. But how do you concretely approach this delicate phase when you are not a specialist?

Author Rob Fitzpatrick says he has faced this situation multiple times and gives excellent advice including how to conduct initial interviews and learn relevant information from them. This is, I believe, the main and rather rare quality of this book. An absolute must-read when you feel helpless on the subject and even more if you do not think you need advice!

This is a small 122-page book that I really recommend reading. Here are some extracts that I hope will convince you

Every question we ask carries the very real possibility of biasing the person we’re talking to and rendering the whole exercise pointless. (Page 3)

And I add a strong statement from Steve Blank: Talking to customers is hard.

The measure of usefulness of an early customer conversation is whether it gives us concrete facts about our customers’ lives and world views. (Page 12)

The Mom Test:
1. Talk about their life instead of your idea
2. Ask about specifics in the past instead of generics or opinions about the future
3. Talk less and listen more.
(Page 13)

Blank talks about “a day in the life of your customer”. You need to understand the actions and the interactions, who does, who decides, who pays.

Here is a list, according of the author, of good and bad questions:
“Do you think it’s a good idea?”
“Would you buy a product which did X?”
“How much would you pay for X?”
“What would your dream product do?”
“Why do you bother?”
“What are the implications of that?”
“Talk me through the last time that happened.”
“Talk me through your workflow.”
“What else have you tried?”
“Would you pay X for a product which did Y?”
“How are you dealing with it now?”
“Where does the money come from?”
“Who else should I talk to?”
“Is there anything else I should have asked?”

at page 15 and he lets you think about what is bad and good before giving his views.

What you should have in mind is given page 22: “They own the problem, you own the solution.” And this is so true as Henry Ford or Steve Jobs mentioned, customers do not know what they want!

So (page 49), when interviewing, “Start broad and don’t zoom in until you’ve found a strong signal, both with your whole business and with every conversation.

How to begin?

In his original book on Customer Development, 4 Steps to the Epiphany, Steve Blank solves this by recommending 3 separate meetings:
the first about the customer and their problem;
the second about your solution;
and the third to sell a product.
By splitting the meetings, you avoid the premature zoom and biasing them with your ideas. In practice, however, I’ve found it both difficult and inefficient to set them up. The time cost of a 1-hour meeting is more like 4 hours once you factor in the calendar dance, commuting, and reviewing.

If the solution isn’t a 3-meeting series, then what is it? You may have noticed a trend throughout the conversation examples we’ve seen so far: keeping it casual. (Page 56)

Rule of thumb: Learning about a customer and their problems works better as a quick and casual chat than a long, formal meeting.


Then you need to deliver (page 62): “When you fail to push for advancement, you end up with zombie leads: potential customers (or investors) who keep taking meetings with you and saying nice things, but who never seem to cut a check.

Rule of thumb: “Customers” who keep being friendly but aren’t ever going to buy are a particularly dangerous source of mixed signals.

Ideally you should find a champion as an early customer. Page 73: “Steve Blank calls them earlyvangelists (early evangelists). In the enterprise software world, they are the people who:
• Have the problem
• Know they have the problem
• Have the budget to solve the problem
• Have already cobbled together their own makeshift solution”

Of course to ask questions, you must organize conversations. This is what chapter 6 is about…

A short extract: “The framing format I like has 5 key elements.
1. You’re an entrepreneur trying to solve horrible problem X, usher in
wonderful vision Y, or fix stagnant industry Z. Don’t mention your idea.
2. Frame expectations by mentioning what stage you’re at and, if it’s true,
that you don’t have anything to sell.
3. Show weakness and give them a chance to help by mentioning your
specific problem that you’re looking for answers on. This will also
clarify that you’re not a time waster.
4. Put them on a pedestal by showing how much they, in particular, can
5. Ask for help.”

Rule of thumb: Keep having conversations until you stop hearing new stuff.

And then you will need to focus by doing customer segmentation and slicing. This is chapter 7.

Rule of thumb: Good customer segments are a who-where pair. If you don’t know where to go to find your customers, keep slicing your segment into smaller pieces until you do.


Avoid creating (or being) the bottleneck. To do that, the customer and learning has to be shared with the entire founding team, promptly and faithfully. That relies on good notes plus a bit of pre- and post-meeting work.

Everyone on the team who is making big decisions (including tech decisions) needs to go to at least some of the meetings.

The tech guys don’t need to go to most of the meetings, but you’ll all learn a ton from hearing customer reactions first-hand occasionally. You’ll also be able to help each other catch and fix your conversation mistakes and biases. (Page 99)

What is the number of people that should face customers? 2 is ideal, 1 is not enough to take notes and avoid bias, more is messy.


I still ask dumb questions all the time. You will too. Don’t beat yourself up over it. In fact, just yesterday I screwed up a particularly important meeting by slipping into pitch mode (this was yesterday at the time of writing… hopefully not again at the time of reading). (Page 112)

with a nice final quote : “Having a process is valuable, but don’t get stuck in it. Sometimes you can just pick up the phone and hack through the knot.” (Page 113)

PS: I think the Mom Test is more convincing than Reis’ Lean Startup, you can read here about the reason of my skepticism.

PS2: thanks to Laurent and Monica for advising me to read this little gem!

Venture capital by Bill Janeway (part II)

In a remarkable new series of videos, Venture Capital in the 21st Century, Bill Janeway describes the value and challenges of technology innovation. I mentioned in my last post the perfomance of venture capital, his third video.

The first video, Investing at the Technological Frontier, describing the radical uncertainty of innovation and how it contributes to economic development.

In the second video, What Venture Capitalists Do, he further develops his thoughts that I summarize through a few screenshots below. (They are self explanatory and you should certainly listen to Janeway if you are curious or intrigued).

The 4th video, The Failure of Market Failure, opens the debate of state intervention and private speculation. This important topic has been largely debated by Mariana Mazzucato and you will find additional posts under tag #mazzucato.

Evaluating Venture Capital Performance by Bill Janeway

I must admit I did not know Bill Janeway. I should have, given his long expertise in venture capital. His recent contribution was mentioned by many including Nicolas Colin and on a personal note, friends from IMF. They just mentioned to me 8 videos which seem absolutely brilliant: Venture Capital in the 21st Century.

I just watched the third: Evaluating Venture Capital Performance | #3 | Innovation in the 21st Century. Here are the slides.

Janeway reviews the performance of Venture Capital firms and recent changes in the venture capital market. He starts by summarizing the stylized facts of venture capital returns (highly skewed, very persistent, and correlated with the stock market). VC capital increased rapidly in the late 1990s, peaking in 2000. VC returns have since settled down, with longer holdings and fewer IPOs. But with the climate of zero real interest rates since 2008, new unconventional investors (private equity, hedge funds, etc.) have waded into venture financing directly, hunting for the high returns of the next big tech giant. A “Unicorn Bubble” has developed as a result, where dubious firms have been financing their growth by selling illiquid securities at inflated prices to deep-pocketed investors with little expertise or control over the entrepreneur. This may have implications on the long-term link between venture financing and technological innovation.

I just copied a few screenshots:

Venture capital is highly skewed and follows a power law, just like startup success models.

Venture capital returns are highly correlated to those of Nasdaq as shown above and below, so… ?

Good VCs are good and bad VCs are bad.


“The message here to limited partners is very clear.
A blind allocation to venture capital, just allocating a fixed proportion to venture capital runs the major risk of what’s known as adverse selection.
The funds you want to invest in, the persistently successful ones, don’t need your money.
The ones who want your money are the ones you want to avoid.”

GAFAM do not suffer from the crisis (part II)

Yesterday I published data in Tesla, Google and Facebook do not suffer from the crisis. and after linking my post to the usual Twitter, LinkedIn and Facebook, one of my readers (thanks Manuel!) told me it would be fun to add Uber as a comparison. I said I would if/when I find the time and then thought why not AirBnB, Apple, Amazon, Microsoft?

I could only compile data about revenues of these firms and I think it is striking enough:

I wrote yerterday the growth rate was above 100% (doubling every year) in the early years declining to around 40% (doubling every other year) then to 15% (doubling evry five-year). Here are the growth rates of these old and new Titans. It begins again with 100+% for all of them. Too early to say about the future of Uber and AirBnB.
The three others of the GAFAM.
– Microsoft even had a 50% growth in its second decade, Amazon was closer to 30% and Apple struggled with 20%.
– In their 4th decade, Microsoft had an average grwoth of 10% and Apple 30%.

OK Manuel?