axplusb

Red Hat net income is growing pretty consistently though [0].

With this price for debt, IBM could be cash-flow positive on this deal within a couple of years even without any big synergy, just by the simple effect of cumulated growth on (net) income.

[0] https://www.macrotrends.net/stocks/charts/RHT/red-hat/net-in...

Solaris by Stanislaw Lem

Permutation City by Greg Egan

Ubik by Philip K. Dick

Judea Pearl's The Book of Why.

Basically Pearl argues that classical statistics completely ignored the concept of causality so far and introduces a complete framework to bring causal inference into statistical/data analysis. The framework is based on graphs and asks for causal hypotheses (like econometricians would do with instrumental variables) and allows to compute/quantify causal effects.

Anyone working with data should probably read this book. The fact that Pearl brought in a professional math/science writer as co-author is a huge boost to the main ideas accessibility and make for a nice albeit deep summer read.

Yes, I was thinking about the law of small numbers too ! Pretty sure the college with the smallest rate of startup founders is also a small college.

Before jumping to conclusions: https://en.wikipedia.org/wiki/Law_of_small_numbers

Edit: Actual Wikipedia entry = https://en.wikipedia.org/wiki/Insensitivity_to_sample_size

In a closed system, hence not the Earth.

I'm surprised to find no mention of Solaris by Stanislaw Lem among the fiction references. In this novel, a whole planet is somehow a living organism, truly alien to human conception of life.

My take from this book is that:

- scientific knowledge is embedded in some intellectual ether made of underlying hypotheses often not explicitly stated called paradigms

- paradigms follow a Darwinian evolutionary process, i.e. better paradigms evolve out of not-so-good previous paradigms

- paradigm replacements start with an epistemological crisis , i.e. facts that the current paradigms don't explain well enough.

Hope it helps !

Guns, Germs and Steel by Jared Diamond offers an interesting and academically quite solid perspective.

On that topic Harari's Homo Deus is a pretty interesting read. He argues that Humanism is the de facto "new" (2-3 centuries old) religion. Soon to be replace by the celebration of something even more global -- data.

Superintelligence, The Idea That Eats Smart People: http://idlewords.com/talks/superintelligence.htm

When you sign a legal contract, you are also likely to spend a non-negligible fraction of the contract to lawyers so that they review it thoroughly. Think of the legal costs of fund raising or the time spent negotiating and reviewing a contract for a large deal.

Some things that are at stake on the lawyers side are their competence (can they actually make sure the contract is "secure" ?), their reputation (track record of competence established over time) and some insurance mechanism (if things go wrong can you get something back from their insurance).

All of the above seem to be missing in the case of "smart-contracts".

My plan currently allows 5GB of roaming data per month, which under my current consumption pattern (I don't travel much more than a couple of times a month) is plenty.

1. Oligoploly

2. Collusion

It's also why phone services are so much more expensive in the US than say France.

Regarding practicals: is there anything more than the overview slides ?

Gradient boosting also often does wonders.

I am wondering how much of this gets to be real-time. Are they computing the difficulty of finding a spot based on Maps/Waze users' live data or using daily/weekly patterns on past data?

There are actually two sides of what is referred to as causal inference. Either (a) inferring a causal graph from the data, or (b) given a graph and data, measuring the causal effect of variables among each other.

The broad idea in (a) is to start with a fully connected graph, and eliminate edges between nodes that can be tested as independent, or independent conditionally on other nodes. This gives you a non-directed graph which can be oriented by several methods (identifying V-structures, looking at residuals of regressions of X on Y vs Y on X).

The theory in (b) actually generalizes instrumental variables and lays out graphical configurations where you can measure the causal effect of a variable onto another variable, and how to compute that effect.

A great reference: https://www.amazon.com/Causality-Reasoning-Inference-Judea-P...

A nice introduction: https://www.youtube.com/watch?v=RPgvfSeQB8A

I would recommend the book Technological Revolutions and Financial Capital: The Dynamics of Bubbles and Golden Ages [0] by scholar Carlotta Perez. She studies longer cycles though (~50 years) but the mechanism she describes seems pretty sound for the matter at hand.

It goes more or less like this:

- capital in search for long term returns goes to early moves of a big technological shift

- as successes from the new technology get more and more apparent, it attracts a much larger slice of capital available, and eventually gets over-funded (the real opportunity of this technology is limited)

- a bubble forms, most capital is in for a quick speculative return

- back to square one with a new technology (and former bubble bursts)

[0] https://www.amazon.com/Technological-Revolutions-Financial-C...

How about Mom, Dad, and Junior share rides with other Moms, Dads, and Juniors who are headed in the same direction?

As an econ PhD dropout, I never quite understood why such an emphasis was made on physics-lookalike models. I was always more encouraged to come up with a bunch of equations that would mimic some phenomenon, than dig around and understand this phenomenon in more depth.

A few arguments that I can agree with though:

- Models are a compact formulation of complex theories. I remember a professor writing a dozen equations and claiming he just taught us Marx's Capital. In retrospect I think that's pretty funny, and quite telling of the superficiality that is rampant in the profession. But still that might have been a great executive summary (I still haven't read the book). Another example is Piketty's "r > g" that really is a lot of food for thought.

- Models can lead to actionable insight. I think there are people in Central Banks and Economics departments of governments who rely quite heavily on some models to forecast GDP, unemployment etc.

But this comes with a huge caveat: it's often easy to tweak a model to make it describe pretty much the opposite phenomenon of the one intended.

So the real benefits of that kind of academia were and still are pretty unclear to me. I believe some academics were hiding their lack of deep understanding of economic matters behind a fog of math. Even Blanchard (former chief economist at the IMF) admitted to something along those lines in the wake the 2008 crisis.

As awesome as this looks, I'm totally missing the point.

If they truly believe AI is dangerous, how does promoting / accelerating it is supposed to help?

Or is it a way to commoditize R&D in machine learning so that it will never be a bottleneck for startups?

I would be interested in A/B comparison of company responses according to immigration status (i.e. whether or not a candidate needs visa sponsorship). Any data available?