The new tactics to implement the #CancelCulture are quite ingenious. They are not going after what they want cancelled directly, they are going after how it gets to you! It’s attacking the free market, and it’s harder to spot and harder to fight before it’s done! Read on!

1/

~8yrs ago (Dec’12) I got a job @Google. Those were still early days of cloud. I joined GCP @<150M ARR & left @~4B (excld GSuite). Learned from some of the smartest ppl in tech. But we also got a LOT wrong that took yrs to fix. Much of it now public, but here’s my ring-side view👇

By 2008, Google had everything going for it w.r.t. Cloud and we should’ve been the market leaders, but we were either too early to market or too late. What did we do wrong? (1) bad timing (2) worse productization & (3) worst GTM.

We were 1st to “containers” (lxc) & container management (Borg) - since '03/04. But Docker took LXC, added cluster management, & launched 1st. Mesosphere launched DCOS. A lot of chairs were thrown around re: google losing this early battle, though K8 won the war, eventually 👏

We were 1st to “serverless” (AppEngine). GAE was our beachhead -- it was the biggest revenue source early on but the world wasn’t ready for serverless primitives. We also didn’t build auxiliary products fast enough. Clients that outgrew GAE wanted “building block” IaaS offerings.

1st to hadoop (map-reduce ‘04) but our hosted Hadoop launched in ‘15. AWS EMR was ~200M ARR by then. 1st to cloud storage (GFS ’03), but didn’t offer a filestore till ‘18! Customers were asking for it since 2014. Didn’t launch archival storage or direct interconnect till v. late.

Here's an overview of key adoption metrics for deep learning frameworks over 2020: downloads, developer surveys, job posts, scientific publications, Colab usage, Kaggle notebooks usage, GitHub data.

TensorFlow/Keras = #1 deep learning solution.

Note that we benchmark adoption vs Facebook's PyTorch because it is the only TF alternative that registers on the scale. Another option would have been sklearn, which has massive adoption, but it isn't really a TF alternative. In the future, I hope we can add JAX.

TensorFlow has seen 115M downloads in 2020, which nearly doubles its lifetime downloads. Note that this does *not* include downloads for all TF-adjacent packages, like tf-nightly, the old tensorflow-gpu, etc.

Also note that most of these downloads aren't from humans, but are automated downloads from CI systems (but none are from Google's systems, as Google doesn't use PyPI).

In a way, this metric reflects usage in production.

There were two worldwide developer surveys in 2020 that measured adoption of various frameworks: the one from StackOverflow, targeting all developers, and the one from Kaggle.

Question: HOW Many Accounts Does Twitter Follow???

Answer: Only 9

WHO Are They Following???

Linux code injection paint-by-numbers.

Can we launch a process that looks one way to (superficial) auditors but is, in fact, entirely different? (Think process hollowing and the like on Windows).

Firstly, how are processes created and what does related auditing look like?

The most common pattern is fork() → execve(). Where the fork() syscall create a duplicate of the running process context and execve() overlays a copy of the target program onto that context.

After calling fork(), we’ll have two processes, the original one and a new - duplicated - one (with a new pid).

Control will return from fork() to both process instances. In the child process, the return value will simply by 0, in the parent it will hold the pid of the child.

Thus we can determine whether we are running in the child context and call execv() accordingly, while allowing the parent to continue.

Now, let’s take a look at where the auditing hooks lie. From calling execve(), we’ll eventually land up in exec_binrpm().