E103: Tech layoffs surge, big tech freezes hiring, optimizing for profits, election preview & more

Sacks said he and Jason Calacanis were only part-time helpers at Twitter, while Musk remained the sole decision-maker. Their access was real enough that they met trust-and-safety chief Yoel Roth during the racist-spam incident, when Sacks said Musk ordered Roth to shut it down.

Sacks attributed the post-takeover spike in racist tweets to a coordinated 4chan bot attack rather than a moderation-policy change. He said hundreds of thousands of spam posts were quickly suppressed, after which reports of a 500% spike reached news outlets and advertisers and contributed to the boycott.

The fake “Ligma Johnson” Twitter layoff stunt became the episode’s clearest example of reporting failure. Several outlets repeated the prank before verification, while the panel split on why: Sacks emphasized confirmation bias, and Calacanis emphasized gutted fact-checking, live-news pressure, and the race for social clicks.

Sacks said his portfolio deteriorated from roughly two-thirds of companies hitting plan after Q1 to two-thirds missing after Q2, then almost the entire portfolio reforecasting after Q3. Even its strongest companies were seeing major headwinds, giving him unusually direct evidence of an economy-wide private-company slowdown.

One panelist described a private software unicorn whose estimated value had fallen below the total capital it had raised, making an employee’s options effectively worthless. He warned this could become systemic among scaled unicorns needing another round after their valuations fell beneath the preferred capital already invested.

Sacks argued founders were still mentally valuing cash at 2021 prices: $50 million remaining from a $1 billion financing looked like 5% dilution, but his current-value example treated it as 20%. With 4% T-bills and 6–8% corporate bonds available, money-losing growth companies suddenly faced much stronger alternatives for capital.

Chamath said he gave $1 million to a Senate PAC because he wanted Democrats to retain the Senate while Republicans took the House, producing divided government. He described the donation as an attempt to create political stasis through 2024 while he expected economic conditions to deteriorate.

Meta’s AI team predicted structures for 617 million proteins from metagenomic data, compared with AlphaFold’s roughly 200 million proteins from known genomes. Friedberg cautioned that Meta used a different, faster technique and said only about one-third were high quality, but the approach could expand discovery of medicines, antibiotics, fertilizers, and other engineered products.