Monthly Archives: February 2018

termsheet吐termsheets

谷歌创业孵化器的旧金山office里的房名是我命名的,概念是用创业相关的词汇:Stealth Mode, Burn Rate, MVP, Growth Hacking, Series A, Post-Money, etc.

一个有趣的事是,我们真正签字的termsheets是在一个叫termsheet的打印机打印出来的哈哈。

Group ​​​​

Do things that don’t scale V.S. scale

Apple is known for its doing things that don’t scale. For example, Apple hires humans to manually write witty and funny scripts for Siri before it gets smart enough to answer any question. That’s one of reasons why Apple products can be more resonate and human-touchy.

Google tries to avoid human interventions as much as possible. Google Translate doesn’t hire linguists to work on rules on grammars or syntax. It’s all automatic by ML algorithms. The strength is obvious – if you are going to translate 1000 different languages, it’s very costly to hire experts for each language.

Recently I came across another interesting example regarding maps.

This is a small town – a rural neighborhood of only 100 residents, where you can see so many details are left behind in Apple Maps compared to Google’s: landscape structures, small roads and paths, exact size/shape/coordinate of buildings. What might be surprising is that all those buildings in Google Maps have their own 3D models (what?!).

1-12+-+Ladd

Google Maps obviously neither spend all the resources of hiring hundreds of cartographers to draw each single block on the earth, nor have its Street View cars to drive into every corner of the globe. Instead, “Google is creating data out of data”. Google Maps created its own algorithm to learn from aerial view images taken by satellite, then converted those images into cartography and 3-D models of each buildings. From its own word:

“These building footprints, complete with height detail, are algorithmically created by taking aerial imagery and using computer vision techniques to render the building shapes.”

map

Conversely, Apple doesn’t seem to extract buildings out of its images. Apple Maps do things in a more traditional way – Apple collected cities one by one from satellites (now there’re 300 cities worldwide), and once Apple is ready with the data quality, a new city on the map is ready to release.


While the infrastructure of Google is meant to build for scale quickly, Apple could fall short at scale at the very beginning of the game, but can make sure quality of everything it releases is up to par. And overtime, Apple can catch up with the scale.

Who said there’s only one way to get things done?

 

Reference: GOOGLE MAPS’S MOAT