Lunchtime #humour - Ordering a Pizza in 2021!
CALLER: Is this Pizza Delight?

GOOGLE: No sir, it's Google Pizza.

CALLER: I must have dialed the wrong number. Sorry.

GOOGLE: No sir, Google bought Pizza Delight last month.

CALLER: OK, I would like to order a pizza.

Contd... 1/9

Twas the night before Christmas and all over the ‘net,
Not a creature was stirring except a few cyber threats
The firewalls were configured at the egress with care,
But that wouldn’t stop us from being hit by ransomware. 1/

The children were nestled all snug in their beds,
While attackers hit the web server and established a beachhead
Mama with her EDR and I with my IDS
Were ready to tackle this hot infosec mess. /2

Down in the SOC there arose such a clatter,
I logged into my dashboard to see what was the matter.
This thing had better work, it cost so much cash.
How in 2020 can this thing STILL require Flash?! 3/

The alerts lit up the dashboard, it produced such a glow
But it’s because the threshold for alerting was configured so low.
When what to my wondering eyes did appear,
But 300 false positive alarms. I thought “I’ll be here all year.” 4/

Then a little old DLL, signed by SolarWinds,
Was really the Russians, masters of supply chain break-ins.
We need some new vendors said the CISO, and the salespeople came. 5/

Let's be clear about what @Google, @Apple & @amazon instantly destroying #Parler's ability to be the free-speech version of Twitter is. At its core, this is about the ability to freely communicate one’s thoughts to others — once considered a hallmark of "The American Way." 1/13

What Big Tech is doing is as if Ma Bell in 1973 (ask grandpa) decided what you're saying to a friend on a landline phone was “dangerous." So they cut off the line in the middle of the talk and then sent someone over to physically cut your phone line. That's Stasi shit. 2/13

I realize that a personal conversation is not the same as a public tweet, but the principle is the same. The company that "owns the wires" for your speec — and has special govt rules protecting them due to their position — cuts off your call due to what you are saying. 3/13

If this analogy doesn't work for you, imagine someone tackling and punching you for speaking on a soap box in a public park and then setting your soap box on fire. And the govt being OK with that in a country that has a First Amendment. Moving on ... 4/13

Big Tech shadow banned people for years to just do it "halfway." That way, they could continue to paint an increasingly thin patina on their original govt-protected pledge to promote the free exchange of thought. #freespeech 5/13

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.

I did more research into the Parler dump. What probably happened was not so much a "hack", but this: When Twilio/Okta shut them down, they just disabled email/phone verification to create an account. This means anyone could directly create huge amounts of accounts via their API.

Someone also found out that fetching Parler posts could be done by enumerating IDs (e.g 1, 2, 3) instead of random IDs that can't be guessed. Unclear if this was via the ordinary API endpoint, or that they found a separate one by monitoring app network traffic.

So you combine these two things and you can create a script to scrape all the posts on the entire platform, using a lot of different accounts to avoid suspicion. Anyone could download and run this script to spread it out over many IP addresses as well.

What I'm still not sure about is whether deleted (meaning flagged as deleted, it's common that services never actually delete data) posts could be fetched without any special handling.

The verdict: The people who wrote Parler are fucking amateurs.