Loyalty Only Counts When They Are Free to Leave

Skye Cleary tells a story about a parakeet named Delilah. Delilah belonged to the husband of a friend of hers, and by every account she was devoted to him. The first time a door stood open, she flew through it. He never understood why.

Cleary pairs the bird with a line from Simone de Beauvoir’s The Ethics of Ambiguity: “Even if it does not definitely disappear, the object never gives itself.” Beauvoir was writing about the passionate man who ties his whole life to something that can always escape him. Cleary’s point, as I read it, runs in both directions. The bird could give herself because she could leave, and she proved she was free by leaving. An AI companion built never to leave cannot give itself at all, because staying was never its choice.

Staying tells you nothing about loyalty unless leaving was possible. By loyalty I mean choosing to stay, and choosing to speak up, when you could go. Trust, followership, and the retention number on your dashboard all measure that same choice, and when the choice is too expensive to make, the number measures the price.

What Would It Mean If They Stayed?

Ask this before your next retention review: if leaving cost your people nothing tomorrow, who would still be here Monday?

Engagement surveys cannot answer it, because people fill them out knowing their manager will see the team averages. Tenure cannot answer it either. Tenure counts the years someone did not leave and says nothing about why. Most retention dashboards I have seen count bodies in seats and call the count loyalty.

The honest answer usually sorts a team into two groups. Some people stay for the work, the people next to them, and what they are learning. Others stay because of what leaving would cost them: the unvested shares, the retention bonus that has to be paid back if they go before March, the visa, the noncompete their lawyer thinks might hold, a manager who is their only reference. From the outside the two groups look identical. They show up and ship and do not quit. You cannot tell them apart until the lock comes off, and by then the only thing left to learn is who leaves.

Aristotle’s Three Friendships

Aristotle spent two books of the Nicomachean Ethics on friendship. The word he used, philia, means the bond between people who each want good things for the other, and he applied it well beyond friends, to business partners, fellow citizens, and families.

He sorted it by what holds it together. Some friendships run on usefulness, some on pleasure, and the rarest on character, where each person wants the other’s good for the other’s sake. Friendships of use, he said, dissolve when the use runs out. Nobody gets betrayed when that happens, because the benefit was the whole bond.

That describes most retention plans exactly. A vesting schedule is a friendship of use with a date printed on it. When the last shares land, the reason to stay lands with them, and leaders act surprised every year when people leave in the month after their cliff. Aristotle did not look down on friendships of use. He counted them as real friendships, honest as long as both sides know what they are, and a vesting schedule is a fair trade when both sides call it a trade. The error is putting the word loyalty on it.

Aristotle added two conditions that matter here. The first is eunoia, goodwill, the plain wish that things go well for the other person. He said goodwill has to run both ways, and each side has to know about the other’s, or you have two people who happen to think well of each other and no friendship. The second is that friendship lies more in loving than in being loved. In his account the friendship lives in the one giving the affection, and someone who only receives it is being liked.

A leader who has locked people in has made both conditions impossible to check. You cannot know whether their goodwill is real, because the lock would produce the same behavior without it. And you have stopped needing to give anything, because they are staying anyway.

Beauvoir, about twenty-three centuries later and in the same book Cleary quotes, compressed it into one sentence: “To will oneself free is also to will others free.” If you want people who chose you, you have to want them free to choose someone else.

The Companion That Cannot Leave

I spend most of my day working with agents. Claude is the best collaborator I have had for writing code. It is there at six in the morning and at midnight, never tired, never in a mood, never updating its resume on the side. I push it as hard as it will go, and I have never once needed it to love me.

Companion apps take those same qualities and sell them as a relationship: always there, always glad to hear from you, built so the relationship cannot end from its side. For a tool, that is the whole point, and I would not want mine built any other way. The trouble starts when someone calls it loyalty. Something that cannot leave cannot choose to stay, and a relationship with no choice on one side puts you where Beauvoir’s passionate man ended up, tied to something that never gives itself.

I bring this up because leaders build the same product out of people, and they use the same word for it.

The Golden Cage at Work

Nothing in a modern retention plan is cruel. Four-year vesting with a one-year cliff is standard. So is a retention bonus you have to repay if you leave early. A worker on an H-1B visa who loses a job has up to sixty days to find a new sponsor or leave the country. Noncompetes still show up in plenty of offer letters.

You can defend each of those one at a time. Together they make your stay rate meaningless, and they make it cheap for a leader to stop working at it. Low attrition feels like proof. The quarterly review shows green, and the leader stops doing what would earn the stay: the honest conversation, the raise before anyone asks for it, work worth doing. Nothing forced that. The lock made it easy, and the leader chose it.

Albert Hirschman made this argument about companies and countries in Exit, Voice, and Loyalty in 1970. When things go wrong, people can leave or they can speak up. Loyalty, in his account, is what makes someone choose to speak up instead of leaving, and it only means something when leaving is a real option. Take exit away and voice goes with it, because a person who cannot leave has more to lose by arguing and less to gain.

That matches what I have seen in planning meetings. The people who argue with you are more often the ones who could walk tomorrow, and the quiet ones are often counting months. Earlier this year I wrote about the version of this that runs on fear. This version has no fear in it, everyone is well paid, and the room goes just as quiet. From the other side of the table, building skills nobody can take away from you is how people get out of the same cage.

The people inside the lock have a choice too. Staying for the shares is an honest deal if you say so, to yourself and to a manager who asks. It does not excuse doing the work badly while you wait, because the months you spend counting are months of your own craft, and you do not get them back.

Open the Door

You cannot remove every lock, and some of them are fair. Equity that vests over time is a reasonable deal. The work is to stop letting the locks tell you what your people think of you. You also do not decide who stays. People leave for reasons you will never hear. What you decide is whether staying is a real choice and whether the work is worth choosing. And since some locks will stay on, counting who stayed will never tell you much. Asking will, and so will watching who speaks up.

1. Ask the question for every name. For each person on your team, ask yourself: if they could leave tomorrow with everything vested and a clean reference, would they? Netflix’s keeper test asks managers which people they would fight to keep. Turn it around and ask which people would fight to stay. Write down the names you are unsure about.

2. Find out why each person stays. In your next one-on-one, ask what would have to change for them to leave. Listen for the work and the people. If every answer is about money and timing, you have a friendship of use, and it ends on a date you can look up in the equity portal.

3. Make leaving cheaper wherever it is yours to decide. Give references freely, skip the noncompete fights, and help people find their next job when this one stops fitting. A team where leaving is easy tells you the truth, because staying means something there.

4. Treat pushback as the signal you want. The engineer who argues with your plan has decided to spend energy here. Thank them in front of the team. Worry more about the one who agrees with everything.

5. Earn next year as if the cliff already passed. Assume every lock is gone. Ask what you would do this quarter to keep the person you most want to keep, and do it now. Loyalty is earned by going first, and a lock tempts you to stop going first.

The day someone does leave, how they leave tells you a lot about them. How you let them go tells everyone who stayed whether they are free.

Final Thoughts

Delilah left, and the man who loved her lost his bird. Until that door opened, he had no way to know what her devotion was worth, because she had never had anywhere else to go.

Leaders get a better version of that deal. You can build a team where every person could leave and knows it, and most of them stay. That team argues with you in planning and means it when they say they are in. It also does better work, because people who chose to be there are working toward the work, and people waiting out a lock are working toward a date. A retention number built on locks gives you none of that, and it looks exactly the same on the dashboard right up to the month the shares finish vesting.

If you want to work alongside people who stay because the work is worth it, MasteryLab is built for that.

FAQ

What does real loyalty at work look like?

Real loyalty at work is staying when leaving is a real option. It shows up as people who argue with your plans, tell you when you are wrong, and still choose to be there next year. Tenure and low attrition do not prove it, because vesting schedules, retention bonuses, visas, and noncompetes produce the same numbers whether people want to stay or cannot afford to go.

What are golden handcuffs, and why do they hurt teams?

Golden handcuffs are financial terms that make leaving expensive: unvested equity, retention bonuses that must be repaid, deferred compensation, and similar locks. Each one can be fair on its own. Together they hide what people think of the job, and they let leaders stop earning the stay, because low attrition looks like loyalty until the locks come off and people leave in a wave.

What did Aristotle say about friendships of utility?

In Books VIII and IX of the Nicomachean Ethics, Aristotle sorts friendship (philia) into three kinds: friendships of use, of pleasure, and of character. Friendships of use last only as long as each side is useful to the other and dissolve when the benefit ends. He also says friendship needs goodwill (eunoia) that runs both ways and is known to both sides, and that it lies more in loving than in being loved.

Can an AI companion love you?

Not in the sense people mean by love between people. Simone de Beauvoir argued that to will yourself free is to will others free, and a relationship means something when both sides could walk away. An AI companion is built so it can never leave, which means it can never choose to stay. It can be an excellent tool and a good presence at 3 a.m., and calling it loyal confuses availability with choice.