Why is everyone reading Sapiens?

With my twitter feed talking about books this December, I was struck by how popular Sapiens: A Brief History of Humankind was. These weren’t people who exchange book recommendations with each other, so there had to be an external cause to this. And Sapiens is not a recent popular release like Michael Lewis’ The Undoing Project that I would expect to be universally popular at launch.

So I spent some time trying to figure out what happened. I’m left with some understanding of how book sales - unlike movies - are driven mostly by big names. My initial hypothesis was that the book probably featured in Bill Gates’ yearly book list, but that was easy to (in)validate. (Gates did review the book on his blog last year, but it doesn’t look like that contributed to the book’s success in December.) My best guess to why the book took off is: Obama.

Google trends for "sapiens harari" from January 1, 2014Google trends for "sapiens harari" from January 1, 2014

For my conclusions I used a few data sources: Google trends, which represents a general awareness/curiosity about the book, Goodreads reviews, which represent the book’s popularity in a more relevant audience, and a timeline of relevant events.

Goodreads reviews for Sapiens from January 1, 2014Goodreads reviews for Sapiens from January 1, 2014

Relevant events

The last two events coincide with the peak in the Google trends graph, and it would be safe to say that Obama on CNN would have more weightage than a new book (considering how insignificant book launches seem to be on these scales). The Goodreads reviews peak trails the Google trends peak by a few months, which I believe can be explained by the time it would take for people to getting to read it + the holiday season for relaxed reading.


Book popularity APIs are hard. I thought the Goodreads API doc would give me the popularity data I wanted, but there is no API for that objective. What I ended up doing was scraping reviews from their reviews widget, and Beautiful Soup made it easy.

def get_raw_dates_from_url(url):
    r = urllib.urlopen(url).read()
    soup = BeautifulSoup(r)
    raw_dates = soup.find_all('span', class_='gr_review_date')
    return raw_dates
January 15, 2017

High Output Management

Sometime in 2015 I realized how incomplete my understanding of management was. In my head management was a job to be done, but there was no such thing as good/bad management. The output of an organization was the sum of outputs of the workers, and changing the manager wouldn’t have any effect. This changed in my first year post college, when I worked with bad and good managers.

The difference was evident to me when keeping all other variables constant, my individual output increased under the good manager vs the bad one. This pushed me to understand how to be a good manager, and what could be quantified as managerial output. Read in that light, Andy Grove’s High Output Management ended up introducing me to managerial output, and I’m left with some takeaways to share.

New vocabulary

Management theory feels like stating the obvious, especially when you’ve worked professionally. But I think there is value to be created by stating the obvious, and converting experiences into words. Vocabulary makes it easier to pass concepts around, build complex concepts on top of existing ones. As complex entities become primitives and more usable, life progresses.

Even before I read HOM, I could describe each of the terms below through a vague combination of recollection and intuition. But by putting them into clear words, I can save the time I would have spent thinking about them, and apply them. (The list is by no means exhaustive; only those terms that left an impression on me.)

Questions of organization

Why do all tech companies have a sense of egalitarianism, with informal attire and open offices? Grove provides an interesting explanation. Since tech is a knowledge based industry, business decisions are based on the latest know-how, which is what workers are best at. For know-how insights to climb up the hierarchy and be a part of decision making, the environment needs to be non-hierarchical.

I can think of other such questions on organization that find answers in business fundamentals. For example, why is sales typically known for high ratio of variable pay? The answer probably lies in the ease of estimating impact of work. As firm theory suggests, the most rational wage for a worker is the marginal revenue product of labour. Hence, a business would like to compensate a worker only for the incremental value created, no more no less.

In sales, the impact of the salesperson’s work is easy to measure, as converted deals add up instantly to the company’s revenue. As a counter example, an R&D scientist has a lower ratio of variable pay because the impact of work is harder to measure. Even when measured, there is a long time gap between when the work was done and when it actually accrues as revenue.

  1. Important word of caution on this from Grove: there is no judgement of value or worthiness when it comes to TRM and relevant management style. We are after what is most effective.“

December 3, 2016 · books · management

Publishing and an open format

A Slack discussion on whether we should shift our company blog to Medium made me miss the days when RSS/Atom were popular and Google Reader flourished. Reader shut down in 2013, and RSS has since been extinct. What struck me is how incentives in publishing have evolved since, and how RSS could fit right in.

[I’m using RSS as the representative open format for timely distribution of content, in a standard platform-agnostic format.]

The incentives of readers have changed little over time, and by design, RSS aligns with them. With a standard format, RSS reduces context switching for readers, thereby enabling more content consumption in less time.

Publisher incentives: circa 2013

The prevalent business model for publishing was advertising. The incentive for (most) publishers was to drive up ad views and hence, website page views. As RSS decoupled content consumption from website page views, it was at odds with how publishing worked. While most websites had an rss.xml link (and still do), adoption of RSS by mainstream media existed in name only. RSS was used as a teaser for main website, with incomplete content, which misses the point of reducing context switching.

Publisher incentives: circa 2016 and beyond?

45% of millennials use ad blockers on their phones. Advertising revenue has moved from publishers to Facebook and Google, who have more users, and are a lot better at targetting them. From NYT:

In the first quarter of 2016, 85 cents of every new dollar spent in online advertising will go to Google or Facebook, said Brian Nowak, a Morgan Stanley analyst.

Media organizations are forced to relook their business models—which can be broadly generalized as1:

  1. Subscriptions from a loyal reader base (eg, The Economist or The Information)
  2. Marketing for other businesses, through sponsored content (Buzzfeed) or affiliate links (The Wirecutter)
  3. Marketing for monetizable products of the same business, events (Shekhar Gupta’s The Print), podcasts (Bill Simmons’ The Ringer) or non-media products (Stripe’s blog that markets Stripe).

These business models value the content instead of how/where it is presented. The business model has been decoupled from page views. Increasing a reader base, irrespective of where the reader is, aligns with the new models.

These readers reside on Facebook and similar products, which have increasingly gained market power at the expense of media companies. Facebook alone controls 40 percent of web traffic to media websites, as stated in the same NYT article. The likes of Facebook Instant Articles, and Apple News exist in the industry, deriving their market power through their existing user bases. Powerful and closed platforms bring problems of coercion and control.

That is where RSS, and products like Google Reader that make RSS user friendly, possibly fit it. Full adoption of an open format by content publishers will open opportunities for new reading products, that do not require a Facebook account or an Apple device. Which in turn also create new business models, that make use of content bundling. While competing with a Facebook for reader eyeballs is difficult—with the right tools and infrastructure, we have seen magic happen before on the Internet.

  1. I’m skeptical about pay-per-read models (micropayments). There’s little to show they work.

April 14, 2016 · publishing

View the the archives