It’s that time of the year! I spent 2 hours today building a spreadsheet to maximize employer HSA contributions while minimizing employee out-of-pocket costs for a family of 4 with 2 earners. What did you do for entertainment on your Saturday?
The Five Best Things
This article covers the events surrounding the major hack that verified accounts -- such as Elon Musk, Joe Biden, Jeff Bezos, Kanye West -- suffered on July 15, where their accounts announced a fraudulent Bitcoin sale. It then covers the changes that Twitter is adopting to prevent these and other attacks leading up to the U.S. Election in November. Some notable tidbits - verified accounts are still handled manually, Twitter was without a Chief Security Officer at the time, and every employee at Twitter was locked out that day and had to change their password in the presence of their manager.
I’m glad Twitter has finally hired a CSO at the end of September, but she doesn’t have a lot of time to ramp up for what could be the largest mis- and dis-information event coming up in a few weeks. Twitter’s plan appears to be watching posts with rising levels of inorganic engagement like a hawk, flash-checking their veracity and taking them down ASAP, with a lot of manual intervention. Twitter has been dealing with some backlash of late with its algorithm for photo cropping accused of bias against people of color, and plans to reduce its reliance on ML tools. At the same time, there are very serious concerns of deepfakes spreading unchecked on Twitter, including this very creepy one where a victim of the Parkland school shooting was re-animated (with his parents permission) to implore folks to vote in this election.
This article dives into the reasons that make Wikipedia a trusted arbiter of truth on the Internet, even as it approaches 20 years of existence. How did this happen? The author offers 3 reasons -
Peer production over ad- and engagement maximization.
Robust editorial oversight.
No reliance on personalization or amplification algorithms.
AI is still used to spot and alert editors to digital vandalism. Another NLP algorithm, Quicksilver is used to identify notable people who have missing Wikipedia entries.
Wikipedia is the “last best place on the Internet”. So why does it still beg you to donate money every so often? Our society places monetary value on engagement and personalization which, in turn drives mis-information and filter bubbles. Even so, there are some serious inequity issues surrounding Wikipedia articles.Yep. Wikipedia is a genuine huge barrier to women in public—most people’s first impression of someone is a Wikipedia page. Women are edited harshly, not contributed to, overlooked or deleted. You can help out by editing, adding and pushing back as much as possible.
Julian 😷 Missig @jmissig@Pinboard A well-published tenured professor who taught me had her wikipedia page removed for not being notable. I contested it and my request was denied. Then Wikipedia held some sort of “make pages for notable women scientists!” day. I brought that page back up and was denied again.
Propublica: Inside the Fall of the CDC
This long article chronicles the American CDC’s excruciating fall from grace, leading up to and after the Covid-19 pandemic. Spoiler alert: it’s due to political interference, inflated hubris, slashed budgets and short-sighted decision making that has led to massive deaths and an irreparable loss of trust in one of America’s most vaunted institutions -- a model that has been copied around the world. Some of the guerilla tactics and lone ranger efforts of Dr. Schuchat and McGowan are mentioned.
Propublica is a non profit investigative news publisher, much in the vein of Wikipedia above. Make this the one thing you read this week. I had no idea the CDC was involved in the shameful experiments on the Tuskegee, in which federal doctors withheld medicine from poor Black men with syphilis, then tracked their descent into blindness, insanity and death.
Daniel Jeffries in Towards Data Science: Rise of the Canonical Stack in Machine Learning
Shared with me by my friend Drew, this post by an employee at MLOps startup Pachyderm comes up with two great, key terms for our industry - the Machine Learning Life Cycle and the Machine Learning Canonical Stack. The lifecycle has 4 stages - 1. Data Gathering and Transformation, 2. Experimenting, Training, Tuning and Testing, 3. Productionization, Deployment, and Inference and 4. Monitoring, Auditing, Management, and Retraining. The table below is the author’s analysis on the tools the industry appears to be coalescing towards, aka, the canonical stack.
The post was written by an employee at Pachyderm so they favor their tools; however, I still think this is a good representation of the state of the industry. I find it interesting that thought a couple of AWS and Google Cloud products are included, there is not a single Microsoft Azure product in here..
Stanford University on Twitter -The committee couldn't reach Paul Milgrom to share the news that he won, so his fellow winner and neighbor Robert Wilson knocked on his door in the middle of the night.
WSJ: A millionaire hacker’s lessons for Corporate America a profile of a 21 year old “ethical” hacker, who finds exploits in a company’s public codebase and helps them resolve it before it gets in the hands of the bad guys, in exchange for money. Last year he made over $1M. Some concerning stuff about healthcare and universities’ cyber security defenses here.
Results from the tech industry survey I shared last week are here. Nothing too ground-breaking.
Vogue: With “Darling Divined,” Textile Artist Diedrick Brackens Plumbs His Southern Roots. Beautiful woven art that is now on display at the Blanton museum of art in Austin! Go check it out, they appear to be taking distancing seriously.
WSJ: Day-Care Centers Are Very Low Risk for Covid-19 Transmission, Study Says Interesting hypothesis here I had not considered: that small children are closer to the ground and thus less likely to transmit particles to adults’ airways.
NYTimes: A portrait of a market in India run solely by women Beautiful pictures and story covering Manipal, a far-flung and oft-ignored state in Eastern India! Chronicles the history of how the market emerged as the dominant voice of resistance against oppression and injustice — and the women emerged as the sentinels of a more equitable Manipuri society.
4-H STEM Challenge for kids; seems pretty legit
Bloomberg: Almost Everyone at SoftBank Thinks Going Private Is a Bad Idea Nothing new to read here other than I think this is a leak by an employee of SoftBank talking about how leaks are used by employees to influence CEO Masayoshi Son. A meta-leak?
WSJ: JPMorgan Pledges to Push Clients to Align With Paris Climate Agreement, USAToday: JP Morgan Chase to spend $30 billion to close the racial wealth gap, Vista Equity Partners founder Robert Smith: How Structural Racism Shows in Economics Should I start an activist capitalism section? I kid, I think actions like this are great. But I believe they are occurring due to an utter abdication of leadership and responsibility by most of our elected representatives on these issues.
Disclaimer: The views and opinions expressed in this post are my own and do not represent my employer.