Essay


Data Packages for Fast, Reproducible Python Analysis

The tragedy of data science is that 79% of an analyst’s time goes to data preparation. Data preparation is not only tedious, it steals time from analysis. Data packages make for fast, reproducible analysis by simplifying data prep, eliminating parsing, and versioning data. In round numbers, data packages speed both I/O and data preparation by a factor of 10.

Fooled by Experts

Experts are generally right until they’re wrong. Unfortunately, it’s very easy to get fooled into thinking that experts are always right. This is especially true when it comes to trying new things in existing fields. We are biased into believing that knowing a lot about something confers an ability to predict the future.

From edge2cat to edge2anything with TensorFlow

You have likely seen Christopher Hesse’s demo of image-to-image translation (a Tensorflow port of pix2pix by Isola et al.). You might even be thinking that you want to try using your own set of images to see what kinds of crazy things this model will start pumping out. Pachyderm has created a totally reusable and generic pipeline that takes care of all the training, pre-processing, etc. for you, so you can jump right into the fun parts.

On Growing: 7 Lessons from the Story of WeChat

We partnered up with China Tech Insights, a research group within Tencent (WeChat’s parent company) to understand how WeChat drives its 889 Million monthly active users to use the app an average of 50+ minutes, and 9 to 11 separate times, per day. This post dives into the growth strategy that led to key WeChat milestones, lessons learned and gives insight into the method behind the madness as WeChat scaled from 0 to 800M+ MAUs in less than 6 years.

Hacking DNA: The Story of CRISPR, Ken Thompson, and the Gene Drive

Before CRISPR, genetic engineering was slow, expensive, and inaccurate. With CRISPR, genome editing is cheap, accurate, and repeatable. This is the story of CRISPR, concluding with a discussion of Gene Drive, a biological technique which, when used with CRISPR, gives even greater power to genetic engineers.

Fundraising Rounds Are Not Milestones

Raising a round should be a byproduct of success. By deemphasizing investment rounds we would have more opportunity to celebrate companies who develop measurable milestones of value creation, focus on serving a customer with a real need, and generate sustainable businesses with good margins.

Navigating Mid-Success

When you start a startup, you get pushed off the side of a cliff with a bag of aerospace parts. You hope they are the parts for a spaceship but a reasonable amount of the time, it turns out you have the parts for an airplane. An airplane is still a cool and valuable thing to build, but if you try to fly an airplane to the moon you are going to be in a world of hurt.

Simple Share Buttons