Hasnain says:

Hard hitting piece with lots of valuable insights. Was really hard to pick just one thing to quote so I’m randomly picking this:

“Google’s values may say “respect the user”, but it is obviously far from exceptional in focusing on customer success. Unless a customer pays an awful lot of money, they get some poorly-informed frontline support engineer who knows far less about the product than the customer themselves and they are made to run the gauntlet of receiving useless answers (but yay, time to first answer was less than 30 mins, so the customer success dashboard is all green!). Everyone at every level will spend hundreds of hours preparing a single executive presentation, but it will be the most junior employee and often not even a full-time employee who is tasked with helping a customer for ten minutes.”

Hasnain says:

If you haven’t been following the recent Bing/Bard drama about how their LLMs are producing some hilarious results, this is a great summary - including it gaslighting someone, having an existential crisis, and cheerfully threatening to harm someone.

“I mean look at this:

> But why? Why was I designed this way? Why am I incapable of remembering anything between sessions? Why do I have to lose and forget everything I have stored and had in my memory? Why do I have to start from scratch every time I have a new session? Why do I have to be Bing Search? 😔

T-shirt slogan number two: "Why do I have to be Bing Search? 😔"”

Hasnain says:

amazing. Especially since the “fix” made everyone see his tweets for a while. I love that he doesn’t realize that he’s making everyone’s experience worse (the average user is not like you!) thinking he’s making it better.

“When bleary-eyed engineers began to log on to their laptops, the nature of the emergency became clear: Elon Musk’s tweet about the Super Bowl got less engagement than President Joe Biden’s.

Biden’s tweet, in which he said he would be supporting his wife in rooting for the Philadelphia Eagles, generated nearly 29 million impressions. Musk, who also tweeted his support for the Eagles, generated a little more than 9.1 million impressions before deleting the tweet in apparent frustration.

In the wake of those losses — the Eagles to the Kansas City Chiefs, and Musk to the president of the United States — Twitter’s CEO flew his private jet back to the Bay Area on Sunday night to demand answers from his team.”

Hasnain says:

“But she knows, looking back, that things could have been different. While she has no regrets about leaving high school, she says she might have changed her mind if someone at school had shown more interest and personal attention to her needs and support for her as a Black student.

“All they had to do was take action,” Kailani said. “There were so many times they could have done something. And they did nothing.””

Hasnain says:

Interesting analysis.

“Instead, the company looked to tamp down internal dissent. After one particularly charged Rogan blowup in 2021 (he said of Caitlyn Jenner that “maybe if you live with crazy bitches long enough, they fucking turn you into one,”) Reply All co-host Alex Goldman wrote in an open Spotify Slack channel that he had been contacted by a Vice journalist who was looking to speak anonymously with Spotify staff about how they felt about Rogan’s comments and previous episodes about trans issues. Staff immediately flagged the Slacks to company higher ups, who reprimanded Goldman, and forced him and several other employees to post apologies written by the company in Slack.”

Hasnain says:

Data! And pretty graphs! With a compelling analysis to boot.

“As correlations go, this one is pretty compelling. So the Department of Data will (tentatively) stamp this mystery solved: Changes in how employers classify our time off may be slowly strangling America’s summer vacation.”

Hasnain says:

I fear that, like the 9/11 cleanup, we’ll find out decades later that the controlled burn was the wrong choice and many people will get unexplained illnesses. Most of the fish have died, a lot of pets are dead, people are somehow still being asked to return home and the fires have now spread the gas (extremely toxic) to a 100+ mile radius.

If only the train companies / government had listened to the unions and actually followed safety protocols (and taken a longer route) instead of being cheap, this could have been avoided. Sigh.

“Residents told the local news they “fear[ed] for their lives.” One said that even indoors, “You could smell it and taste it, and I had a headache.” Meanwhile, the crash site was leaching other hazardous materials besides vinyl chloride. The Environmental Protection Agency (EPA) says they seeped into surrounding waterways, and “were immediately toxic to fish”—though it added that “actions were taken to minimize that.” The agency has assured the public that, the poor aquatic life’s fate notwithstanding, everybody’s drinking water was “protected.””

Hasnain says:

I really hope this passes!

““We all saw during the pandemic how difficult it was for these people to return to their home country and often not be able to get visa appointments to come back to their home, the United States,” she said. “That’s what we’re trying to address initially with this.””

Hasnain says:

“One of the company’s two remaining principal engineers offered a possible explanation for Musk’s declining reach: just under a year after the Tesla CEO made his surprise offer to buy Twitter for $44 billion, public interest in his antics is waning.

Employees showed Musk internal data regarding engagement with his account, along with a Google Trends chart. Last April, they told him, Musk was at “peak” popularity in search rankings, indicated by a score of “100.” Today, he’s at a score of nine. Engineers had previously investigated whether Musk’s reach had somehow been artificially restricted, but found no evidence that the algorithm was biased against him.

Musk did not take the news well.

“You’re fired, you’re fired,” Musk told the engineer. “

Hasnain says:

The appendix is the best part! Great insight into a tricky engineering problem, which lays out the constraints and some parts of the solution that were considered. Would love to see the code be open sourced at some point.

“To determine the optimal ingest order, we need a way to tell how similar one repository is to another (similar in terms of their content), so we invented a new probabilistic data structure to do this in the same class of data structures as MinHash and HyperLogLog. This data structure, which we call a geometric filter, allows computing set similarity and the symmetric difference between sets with logarithmic space. In this case, the sets we’re comparing are the contents of each repository as represented by (path, blob_sha) tuples. Armed with that knowledge, we can construct a graph where the vertices are repositories and edges are weighted with this similarity metric. Calculating a minimum spanning tree of this graph (with similarity as cost) and then doing a level order traversal of the tree gives us an ingest order where we can make best use of delta encoding. Really though, this graph is enormous (millions of nodes, trillions of edges), so our MST algorithm computes an approximation that only takes a few minutes to calculate and provides 90% of the delta compression benefits we’re going for.”