How to Win the AI Talent War

The talent war is justified and rational. But recruits are turning down 8- or 9-figure offers. So how do you win?

In Civ VI terms, the win condition is not a Domination Victory but a Religious Victory:

You win by converting others to your vision of the future, instilling faith in your ability to lead them to the promised land, and persuading them that you are destined to prevail.

Money is necessary but not sufficient.

(1/8) Over-reliance on comp reduces companies to ATMs and people to chattel.

It also messes up internal dynamics and external vibes, and if your main selling point is short-term liquidity then you won’t get true believers.

Beyond dollars and GPUs, this is what you need to get (and keep!) the best researchers and engineers:

1. Mandate of heaven
2. Clear mission
3. Kleos
4. Esprit de corps
5. Star factory
6. Network effects
7. Recruits as recruiters
8. Freedom to cook
9. Leadership
10. Momentum

It’s wild. The Blake Lively lawsuit against her costar is an absolute PR kamehameha

Lively’s team basically:

- gathered mountains of incriminating receipts, including emails and texts that might have been leaked by the crisis comms consultant’s former employer

- waited until he thought he was in the clear, as his gloating texts showed

- ambushed him with a Friday night filing and a leak to the New York Times (my bet is Lively’s lawyers briefed the NYT on deep background), resulting in a scathing investigative piece

(continued) Blake Lively really does have great lawyers, with strong comms instincts.

The complaint itself is written very strategically:

- is structured like a story, starting with one gripping scene and leading the reader on from there

- highlights themes in societal discourse designed to evoke disgust for the costar Justin Baldoni (abuser of power) and sympathy for Lively (working mother)

- uses memorable anecdotes and examples instead of generic descriptions, eg Baldoni intruding on Lively’s breastfeeding

- proactively undermines Baldoni’s potential response by showing evidence of his public statements contradicting his private actions, implying he’s a hypocrite and liar

- taps into the national mood by accusing Baldoni of weaponizing being a male feminist, so anti-wokes would hate his posturing while progressives would hate his bad allyship

- clear and measured language makes the claims seem factual, not emotional or dramatized (which Baldoni would likely accuse Lively of being)

Speaking of leaks, Character AI could be an interesting case study

I could be wrong, but my read is that Character has employees who are leaking, maybe as a negotiating tactic with Google

A thread of breadcrumbs: A month ago, Kalley Huang at The Information (this becomes relevant later) wrote the below story, citing:

- "a person close to Character"
- a statement from a Character spokesperson
- "two people who spoke to company leaders"
- "a person who spoke with the employees" of Character

Google, xAI, Meta, and Sequoia were all mentioned in the story as having talks with Character but all of them declined to comment on the record.

It was an exclusive, which means she was the first to have the story. That usually happens through direct conversations with sources.

Character (I don't know who, so I'm using the company as a synecdoche for the people) spoke on the record, on background, and most likely off the record.

It's impossible to prove this last point so I'm just going off my experience and what I've seen happen.

Comms before the storm:
You need a crisis comms plan *before* crisis hits.

That plan should include

1) War room roles
2) Criteria for breaking the glass
3) Principles and priorities
4) Hour 1 actions and fact-finding
5) Messaging guidelines
6) Scenario plans and tactics

i.e., 1) War room roles & responsibilities

Have a list of who needs to assemble in a crisis, with their cell numbers and locations.

Most likely the CEO, CFO, CTO, general counsel, and heads of comms, people, product, and investor relations.

Each will have specific responsibilities.

It’s tragic that Silicon Valley Bank could lose 80%+ of its value in a single day.

But what’s crazy is that the financial collapse was largely driven by a communication collapse.

Their storyline unraveled and their messaging went off the rails, in 4 big ways.

(continued below)

https://twitter.com/anothercohen/status/1633966144901066752

(These are incomplete observations, made with humility and empathy for the good people at SVB doing their best in the fog of war.)

So what seems to have happened:

SVB took a hit to their balance sheet, like many others have, and they decided to raise some money…

Today the CMA released its provisional findings on our deal.

They’re exactly as they sound - provisional. This means the CMA lays out its concerns in writing, and now both we and MSFT have a chance to respond.

(thread)

https://twitter.com/cmagovuk/status/1623293699466592256

We hope between now and April we’ll be able to help the CMA better understand our industry, to ensure they can achieve their stated mandate

“How do I grow my newsletter audience?”

Writing a blog or newsletter is the best way to scale your ideas.

But a following doesn’t just happen - you have to build it.

When I was running comms at Substack, this is the advice I gave writers for promoting their work: 1) Lay the groundwork

Know your “brand” and your value proposition:

A) What unique thing you’re offering readers

B) What unique background/perspectives/voice you’re bringing to the topic

C) Who your target audience is

That goes on your Home and About page.

Happy birthday to the immortal Professor Tolkien — may the hair on his toes never fall out!

In annual tribute to one of the greatest geniuses of all time, here’s a collection of wisdom from his characters… “I was talking aloud to myself. A habit of the old: they choose the wisest person present to speak to. The long explanations needed by the young are wearying.”

- Gandalf

Antitrust law exists to protect consumers, not competitors.

The FTC is doing the opposite with their baseless, ideologically-driven suit to stop the Microsoft / Activision Blizzard deal.

So let’s talk about it… The facts show this deal benefits gamers, workers, and global competition.

But the FTC is ignoring facts, along with precedent and business reality. Their case rests on the unfounded fear that Microsoft will make Call of Duty exclusive to Xbox.

3 reasons that makes no sense:

SBF did all the wrong things with money, but all the right things with press.

An analysis of FTX's 7-part media playbook, and what other founders can learn from it... Tech reporters aren't pollyannish about startups, and the kind of coverage FTX enjoyed didn't just happen. It was a sustained campaign by SBF and his team.

This isn't about how media fell short (many others were fooled too!). It's about the smart tactics FTX used to pull it off:

1) Last week SBF gave an all-time accidental interview to Vox, when he thought he was off the record.

This actually happens to people all the time, even those who are really savvy with the media.

Here are the specific rules of when you can be quoted by a reporter... 2) ON THE RECORD is the default.

Assume everything you say or do is on the record and can be reported. That includes your posture, your stammer, your outfit, your lunch order, everything.

If you don't want it to be reported, confirm *in advance* that you're not on the record.

Hit pieces come for every mission-driven startup.

These attacks are not only unfair but can be deeply personal, insulting founders’ ability, integrity, gender, intelligence, politics, relationships, and even appearance.

But the good news is that hit pieces can backfire. This is the horseshoe theory of media coverage.

When prejudiced reporters target companies with deliberately misleading coverage that’s just intended to get clicks, plant innuendo, or cause damage, it can actually help the company.

lulu.substack.com/p/hit-pieces

The playbook for communications and PR is outdated.

Traditional comms might work for incumbent corporations in traditional industries, but it doesn't work for founders and startups trying to build something new, that's by definition unprecedented and poorly understood. Founders are insurgents, not incumbents, and their comms should reflect that. When you're an underdog, you have to run lean, take risks, and build your own communication channels. You have to carve a reputation out of rock with your hands.

There are no easy answers. But:

We keep seeing false claims about Substack that seem designed to mislead.

Here’s just one recent example, which the outlet refuses to correct.

We believe the best way to fight misinformation is through open debate, so here are the facts.

1/x The writer claims Substack “actively recruits and pays extremists.”

We do not and have not, and he can’t name any examples. Yet when approached about it, he said: “I'd be happy to remove it if you could provide evidence that it's inaccurate.”

That’s not how this works.

Today the NYTimes wrote an article about Substack. While it's a compliment to remain top of mind for the paper of record, the piece contains a lot of hearsay, cherrypicking, and personal opinion presented as fact.

Here, I offer an alternate framing and excavate the buried lede: First, it's important to understand that The Times is covering a company they consider a competitor.
The fact that they themselves are one of Substack's "rivals" is not mentioned until the very end of the article. With that dynamic made clear, the article's framing makes sense.

The Canadian government has proposed legislation (Bill C-36) where people can be taken to court and penalized if they’re suspected of being ABOUT TO post something hateful online.

An individual would be able to report another PREEMPTIVELY for something they haven’t said yet. Alarming aspects, in ascending order:

1) Fuzzy and circular definition of “hateful” speech (“involves detestation…stronger than dislike”)

2) Encouraging citizens to report on one another — creepy

3) The ability to punish people for something they haven’t actually done yet (!)

At Substack, we don’t make moderation decisions based on public pressure or PR considerations.

An important principle for us is defending free expression, even for stuff we personally dislike or disagree with. We understand principles come at a cost. 🧵 I’m proud of our decision to defend free expression, even when it’s hard, because:

1) We want a thriving ecosystem full of fresh and diverse ideas. That can’t happen without the freedom to experiment, or even to be wrong.