Hacker News Reader

Comment by hintymad | Hacker News Reader

100X developers are the groups who turn impossibles into possibles. Some Uber engineers used to believe that Graphite was the best thing in the world and Nagios was second bar none, and they went ahead design a complex yet failing architecture for monitoring even when Uber's traffic was still tiny before Uber's M3 came along. This is what Netflix did with four engineers years before Uber's saga: https://www.quora.com/Are-Netflix-employees-really-that-good...

To those who couldn't scale out Graphite and got crushed under gruesome oncall alarms, the Netflix folks were 100x engineers.

View on HN

Exactly this comment.

10x developers exist. 100x developers exist. These developers don't program 10x as fast or have 1/100th of the code. They bring 10x or 100x more value as compared to other developers.

People in HN seem to forget: Code doesn't live in a vacuum. Code has purpose. Code fixes problems. 100x developers are the ones that bring true value. That value may come from a 300 LOC data structure that improves efficiency by 50% or a 30-page code architecture that improves scalability. The means to produce value are irrelevant. What matters is how much value you bring to the table.

Yeah it's not so much cranking out 100X the code as it is good decisions, good teamwork, and management that clears the road rather than throws up obstacles. Some people could not do the same, even given the same advantages, but at the same time it's pretty much impossible to produce the same kind of results in an environment without those. (Source: was on this team, but not one of the original 4)

100x in this sense is akin to doctoral level research. Think SpaceX turning the impossible into the possible with reusable rockets. Their returns on investment are now 100x greater than their competitors.

Graphite has been obsolete for 5+ years. There are commercial products with unlimited scaling, great UI, alerting and a hundreds things out of the box (see datadog and competitors).

That's not the point here. OP said, that the Netflix team could at the time deliver scale, while other teams couldn't.

Yeah, and the key difference is simple: The 100x engineers in Netflix just understood that file-based lookup for key-value pairs won't scale for OLAP-like workload on time series, and of course they possessed the sheer capability of implementing the right technology from ground up.

I was on Roy's team at this time. I am not one of the four mentioned in the link, but I definitely know exactly who is being talked about. I wanted to comment that these four were considered outliers even by Netflix standards for being very high performing. So bear in mind the example is an extreme one.

The Graphite story happened more than 6 years ago, or so I heard as I joined after M3 was in production.

Reply on news.ycombinator.com