Showing posts with label san francisco. Show all posts
Showing posts with label san francisco. Show all posts

Sunday, March 1, 2020

Coronavirus in the bay area

Context : At the time of writing March 1(Sunday) , there were 3 cases in Bay Area and 5-7 in Washington.

Prediction by March 8(next Sunday), we will have
  1. 100 cases in california
  2. Clusters of infected areas evolve 
  3. Some town in the west coast gets quarantined
Outcome(Edited on March 7, Saturday)

This is what we can do to flatten the curve

Disclaimer : I am no bio expert. These are learnings I accumulated from reading blogs. Please use your own judgement. Not everything you read on the internet is true. I am just posting the summary of my learnings here and some predictions drawn from that. 


Why am I posting this ?

Lots of friends are asking me whether to work from home and whether they should get supplies stocked up or not. I see a bunch of people clueless about the situation and lot of panic. Also people are not sure whether they should use bart/caltrain/public transport. Wanted to share some of the research that I did over the weekend.

Summary of my learnings and some numbers

Disease -> Spread -> Metrics -> Virality -> Infrastructure hotspots
  1. Coronavirus may have spread in the seattle area for weeks as gene sequencing results suggest
  2. The virus has a 14 day incubation period
  3. The test in the hospital takes 2-3 days 
  4. This results in a delay in detection of around ~15 days
  5. Thus there are way more people around affected than we know
  6. Detection of peak will be delayed after the actual peak
  7. As we see from italy, the government will have no way other than quarantine a whole city by the time they detect a cluster where the virus is spreading
  8. The disease has a CFR(Case Fatality Rate) of ~2.3%, that means that 23 out of 1000 people affected will die
  9. The disease has a R0 of ~2.5, that means that 10 affected people will spread it to 25 people. R0 is a measure of how infectious a disease is and is heavily used for planning strategies to mitigate the spread(R0 calculation is an ongoing task here and it lies between 2 and 3.5). This is what makes it an exponential rate of increase
  10. A very good read to learn more about R0 and CFR and to put things into perspective, and how different diseases(measles, flu, rabies) compare against each other on these two dimensions
  11. Inspite of the low CFR, this disease is a strain on the medical infrastructure due to clustering around particular locations. 
  12. California has 1.8 hospital beds per 1000 people
  13. As you can see, that given the CFR, if the disease spreads, we will only be able to care for the critically ill patients only. This is also something that has been observed in the past in other countries. Non-critical patients have been asked to stay home
  14. Couple of factors that could test USA: lack of paid sick leave and some people not having healthcare. The high population density in the bay area may expedite things too
  15. This will effectively create a Denial of Service(DOS) attack on our healthcare system
  16. This could very well put the world into a global recession where rate cuts may not really help
  17. Summer is still 1 month away. 

What steps can we take as individuals to prevent/protect themselves

  1. Please watch the government video : https://www.youtube.com/watch?v=nnz95Ssgt3k

References

  1. At Mar 2, 2.30pm PST 16 people are affected across the state of california - 6 people die in Washington
  2. Mar 2, 5.20pm PST - 100 cases of coronavirus in the US
  3. Mar 3, 10.30am PST, 24 people are affected in california
  4. Mar 3, 11.45am, Seattle times has taken down their paywall https://www.seattletimes.com/seattle-news/health/coronavirus-kirkland-seattle-updates-tuesday/
  5. Mar 3, 11.45 am New cases in washington looks like doubling every 6 days
  6. Mar 3, 2.20pm - 45 cases in california
  7. March 4, 11.28am 58 cases in california, 1 person sadly dies
  8. Mar 5, 10.45pm update : Coronavirus - Black Swan of 2020
  9. Leaked presentation in hospitals show that they are prepared for millions of hospitalizations
  10. March 7, 3.09pm, 104 cases in California. The predictions came out true. 
  11. If you have read till here, this twitter thread breaks down what exponential growth looks like. 
  12. Please stay healthy and stay safe for yourself, your family and your community. 

  1. Must watch video from NYC ER doctor - author of superbugs, race to stop an epidemic




Thursday, February 27, 2020

Difference between PM and TPM roles

The goal of the PM(Product Manager) is to answer the "what". The PM is the customer advocate. The PM works with "Customer Research" to formulate customer needs/pain-points into product requirements which feeds into the product backlog. The PM owns the goal to measure progress against the product vision and strategy. The PM works with Engineering to create/define metrics to measure progress against those goals. The PM is responsible for prioritizing the backlog to make sure the most critical items are being executed on.

The Engineering Manager(EM) is responsible for the "how". This is about how to build the right engineering solutions to achieve the goals. The EM is responsible for determining cost estimates, identifying dependencies from other teams, risk mitigation, project sequencing and planning needed to hit the goals. This is effectively the roadmap which the EM and PM builds together.

The TPM(Technical Program Manager) is now responsible for driving day to day execution. In some companies, there is a specific TPM role for this. In some companies this responsibility is shared by the PM and EM. TPMs are more useful where there are multiple teams involved and it is a huge cross organization collaboration effort. Often times infrastructure, services and platform teams tend to have TPMs and consumer facing teams tend to have PMs. However, these are more of heuristics than rules. 

Friday, February 21, 2020

Build vs Buy Tradeoffs as Managers

Cultural Tradeoffs
A lot of the companies in the valley have a strong sense of purpose backed by a mindset of innovation and research. Fear of being outmaneuvered by the next startup or not being able to on top of the latest industry trends may be some of the underlying deep rooted reasons. This spurs innovation, new research keeps happening, engineers and scientists keep publishing papers and patents. Like any completion funnel, some of these research ends up getting productized leading to revenues and the virtuous cycle of investment on in-house research keeps going on. Some other advantages of building in house and open sourcing
  • Brand building
  • Leverage open source community
  • Engineering recruiting
  • Innovation and product development
  • Strategic value development

This is all great. But there are always two sides to the same coin. Taken too far, this leads to the Not Invented Here(NIH) syndrome. The deep rooted fears behind this may be ingrained in the organization : defensive mindset, lack of objectivity, tribal territorial fear, touch of arrogance, complacency. NIH has several side effects like : "re-inventing the wheel", huge recurring maintenance costs, loss of focus as an organization, weak prioritization and goal setting, increasing costs.

Corporate Examples
At the time of this writing(2019-early 2020), companies blinded by "Capital as a moat" strategy pursued by VCs(read softbank), are subject to these states of dysfunctions. The typical path of misfortune while building for the wrong problems looks like : founder raises ton of money -> head count -> hiring ramps up to show growth and progress to the VCs -> NIH syndrome -> build everything -> engineers love this as they have opportunity to grow -> more senior engineers needed -> team grows -> managers get bigger teams -> becomes senior manager. This kind of bloat doesn't end well. As Warren Buffet calls it, "Only when the tide goes out you know who is swimming naked".

From internal conversations with Googlers, I feel Google was also plagued by NIH from 2010-2014. During this phase, there was a culture of arrogance that Google has the best technology and scale and they didn't even treat AWS as meaningful threat. They practically missed the essence of AWS and software as a service. This is why AWS had a huge headstart even though this was playing out in the public markets and everyone was aware of what AWS was building.

What can manager's do
As managers it is super important to be aware of the state of the company/organization/team relative to both external and internal technology/products. We need to make sure that the pendulum doesn't swing too far in either direction. As managers, raising capital(VC money/head count) is important, but it is also super important to ruthlessly prioritize and build the right products and the right culture within the teams.

Exercise for the readers : Often times companies make big acquisitions. Only some of the times are these acquisitions successful and the acquired company is successfully integrated into the mothership. What are strategies behind these successes with examples. Please leave your thoughts in comments. 

Sunday, March 31, 2019

What is network effects and how does it apply to Uber

One of the fundamental investment mental models in silicon valley and for evaluating startup investments is network effects.

What is network effects ?
“When the value of a product to one user depends on how many other users there are, economists say that this product exhibits network externalities, or network effects.” Carl Shapiro and Hal Varian

Examples of network effects :
1. Two sided marketplaces like eBay, Amazon, Uber
2. Social networks like Facebook

How does network effects work for companies like Uber ? 
In order to build two sided marketplaces like eBay and Uber, there has to be supply side liquidity and demand side needs. Supply side liquidity means more drivers on the platform. More drivers will attract riders (demand side needs) due to lower pickup times. As more riders come on the platform, driver earning potential will increase. Driver utilization rates per hour will increase. The driver will get more rides during every hour they spend on the network which will reduce idle times and lead to better network utilization. This will drop prices, attracting more demand and more riders. This ends up becoming a positive cycle that reinforces itself and the marketplace grows. 

Creating two sided network effects is hard because of the creation of both the supply side and demand side dynamic. Hence, Uber had to spend some money on fixing the cold start problem. At the same time, it is equally hard to kill two sided marketplaces because of the same reasons. For the driver, it pays off to be in this network compared to other networks because of better utilization rates. Also it pays for the driver to be in the bigger network because the bigger network is available in less densely populated areas. Most of the competitors of Uber are present in high density areas as they are local challengers and dont have the network strength to challenge in the outskirts. For the rider, it pays of to use the Uber network due to cheaper prices, lower wait times and better utilization rates in both dense and not so dense areas.

In the future articles, I will
  • compare this with network effects in other companies like FB, Amazon
  • how uber is strengthening the supply side and building a defensible moat over it


References
1. All about network effects
2. How to measure network effects

Books I am reading