Hacker News Reader

Comment by dekhn | Hacker News Reader

no, my goal was to use google to revolutionize science. It's not my problem that Sundar cares more about showing ads to the next billion users and getting good press for DeepMind "solving protein folding" when they did nothing of the sort. I cared about long-term scientific research advances that would be shared with the world to reduce overall health expenditures.

At the time I joined it seemed like it would be pretty easy given that Larry Page greenlit Google Health (a terrible idea).

View on HN

> Larry Page greenlit Google Health (a terrible idea)

I'm curious why you think Google Health was a terrible idea. I'm not disagreeing, just wondering your opinions on the matter from a more inside point of view.

Thank you.

It's pretty simple. There was not and will not be a market for Personal Health Records. Google Health was designed to be a store of PHRs where users had significant ownership over their health records, and they would allocate access to service providers.

That's not how health records work in the US where Google tried to start the product. HRs are effectively owned by the companies that provide health care, while patients have access to the records. There is a multibillion industry dedicated to keeping things that way (Cerner, Epic, the hospitals themselves) and even with Google's ostensible ability to change the world (at that time), they would never have been able to build a product that sustained itself because there wasn't a market for the product.

In short Google tried to compete with EPIC and Cerner without a market for their alternative product.

Individuals in the US (present company excepted) aren't interested in owning and managing their own health records; they want their HRs to be managed conveniently for them.

Today, Google (and AWS and Azure) makes deals with EPIC and the Hospitals (to host infra in their cloud) and also makes research deals (to help build their health research projects). This is a much more realistic approach given the constraints.

Can you expand on why the the role didn't work out for you? Was it lack of direction on how to bring the research advances as a product or was it the lack of support for the research itself?

Everything worked fine but I was stuck in application development when I wanted to be a research scientist (IE I wanted somebody else to build product so i could do the research on top of it). During that time, I managed to make a team member unhappy with my criticism of their work, and that led to an HR interaction where transfers into all the research parts of the company were blocked.

In the case above, it was really about a pivot of Google Cloud Genomics- my market research suggested that people wanted to store large amounts of BAM files cheaply and access them via an API, when they really wanted Spark for Genomics. I brought that up with our leadership and they declined to make that pivot early.

Here's what the Cloud Genomics product eventually evolved into, which actually is pretty good: https://verily.com/solutions/terra/ (I was the person who set up the original Google/Broad relationship that led to all of this).

> Everything worked fine but I was stuck in application development when I wanted to be a research scientist

This seems to be a fairly common thing at Google. I hear Google "hoards" a lot of PhDs who are just "refactoring APIs and doing migrations". Is this not seen as a problem there wrt retaining talent?

Sure, I knew a ton of PhDs who were workign on random unrelated stuff (like, protein folding genius -> building next gen music UI). You can retain a lot of talent in SV by paying more than other people so that those employees can buy a house here.

The whole talent hoarding strategy by big tech sounds anti-competitive as all hell, in spirit if not by any applicable letter of the law.

> getting good press for DeepMind "solving protein folding" when they did nothing of the sort

Ah, taking a page out of IBM's recent playbook.

> when they did nothing of the sort

Would you care to elaborate or link to an article that explores this view?

I believe there is a fair amount of criticism explaining why DeepMind made a tremendous improvement to protein fold prediction while not actually solving the folding problem. https://occamstypewriter.org/scurry/2020/12/02/no-deepmind-h... https://futurism.com/the-byte/scientists-unimpressed-googles... https://www.businessinsider.com/deepmind-google-protein-fold...

Reply on news.ycombinator.com