I interned at ChatGPT this summer. As of July 2026, OpenAI is one of the most spotlighted institutions by the media, with love and hate. I open up my phone and see traditional media and social media glazing or hating OpenAI for whatever reason, and then I go to the office, see what code is in GitHub and what slides are presented in meetings, and I couldn't help but realize, gosh, whatever the media is spotlighting, it's so shallow.
"Company X is IPO'ing for xyz amount of money"
"Company Y is developing the most capable AI model ever"
"Company Z is considering laying off 10% of its workforce"
"Company A is headed for its greatest revenue year yet"
"Company B has a new CEO, promising change"
"Company C is fighting a supreme court lawsuit, arguing for xyz reason"
If you zoom out to a 10-year time horizon, or even a 40-year time horizon, media headlines always revolve around these clichés. Sure, it's not misinformation, but it's so shallow, and it's missing so much other important information. To make things worse, media rarely come back to the same topic to show its trend over time, because by that time, the trend would have faded.
Seems like, no one cares about the whole truth over time anymore. After all, why would people care? Our lives are already busy enough, and media's purpose, unfortunately, has degenerated into ragebait entertainment instead of genuine deep facts and analysis.
It's not just the media, but information on the internet as a whole.
Sure, I love all the technical blogs teaching me how to get hands-on with xyz,
but there's so little information on anything beyond hello world.
Even in a lot of official documentation, the examples for more advanced features are very much hello worlds.
The problem is even worse for system design or anything involving several moving pieces.
Sure, bytebytego and github.com/system-design-primer are cool,
but there's so much between "we can scale a database by sharding" and actually using a distributed database,
or actually scaling a database.
And technical knowledge is just one tiny example. Think about job offers, company missions, team cultures -- this information simply does not accurately exist on the internet. And let's zoom out even more, not just tech or jobs. Any experience or knowledge that is somewhat more niche than hello world is not on the internet.
One needs to know, repeatedly practice, and reflect on an insight before understanding it well enough to share with others. But if one knows 50% of something, they might only be able to accurately write down and share 10% of it. The deeper the knowledge, the fewer people know it well enough to share it, and those people who know enough to share are simply too busy to be a LinkedIn warrior to scream, and it just takes a huge amount of time to share deep knowledge in the first place. As a result, most of the information online is vast in quantity and shallow in depth. Much of the good juice is latent in people's minds or exchanged only with good friends on good days. Whatever is visible is just the tip of an iceberg.
So, having friends is helpful. Because very quickly, Google and ChatGPT stop working accurately, and word of mouth from trusted sources really starts to kick in. If the whole point of having friends is to gather intelligence, then that's just an absolute shame, but you have to admit that, 何为好友呢？
子曰，“友直、友谅、友多闻。”
It roughly translates to, "good friends know ball, type shit."
LOL.
Further reading: The Submarine by Paul Graham, where he lamented about the media too.