Which Future Person Is Still You?

Diego’s worldview on personal identity

There is more than one Diego in this room. The ten-year-old has strong opinions about how the life should proceed. The twenty-five-year-old has several projects that are apparently binding on everyone forever. The present one would like to go dancing. Somewhere ahead, a retired Diego expects the entire committee to have saved for him.

Diego’s question is not merely which member should win. It is what makes them members of the same committee in the first place. How much continuity matters? Which kind? Does a more intelligent successor still speak for the original? Could a perfect copy be someone else? And why should a distant future version receive unlimited authority over the person who is here now?

He has explored these questions academically, experimentally, practically, and occasionally by encountering an old essay and discovering that its annoyingly brilliant author was himself.

The witty son of a bitch in the archive

In 2023, he described that pleasure:

There are few pleasures as great as forgetting a writing, reading it, almost as if it were someone else's, and finding it brilliant, only to realize it was penned by your own past self, that witty son of a bitch who you sometimes have personal identity with¹

The joke contains much of the philosophy. Past Diego is recognizably connected to present Diego, but he is not simply present Diego relocated to an earlier date. His memories, interests, dispositions, and priorities differ. Recognition can arrive as a rediscovery rather than a continuous feeling of ownership.

In On What Selves Are (2012), Diego begins by resisting the assumption that the word self names one simple thing. He describes it as:

Self is a mongrel concept in need of disambiguation.²

Different uses collect different functions: a body’s sense of its location, an autobiographical narrative, a social identity, an inner perspective, a person held responsible for promises. Even within one use, membership can have a cluster structure rather than a single defining essence.

The familiar claim that someone is not himself today illustrates the flexibility. We are not normally claiming that a new soul entered the building. We are adjusting the model by which we interpret his performance, actions, and character. The usual person and this unusual afternoon fit together imperfectly.

Diego also notices that our concepts perform social jobs:

When I say you owe me that money, it implicitly implies that you are the same person as the one to whom I lent that money.³

Personhood is not therefore a banking invention. The example reveals how much work we ask sameness to do. Keeping promises, assigning credit, remembering friendships, and holding someone accountable all rely on models of people through time. A theory of the self must understand those functions rather than simply admire the word’s grammatical singularity.

Replace the yes-or-no question with the relations

Derek Parfit is a major influence here. Diego’s older writing replaces a demand for a binary essence with questions about psychological connectedness and continuity. Memory can persist more or less. Character, dispositions, intentions, habits of inference, and other traits can overlap by degrees.

In Rationality versus Short Term Selves (2012), he summarizes the position:

This is our current state of knowledge on Personal Identity: There is Trait-Relatedness, and there is Self-Relatedness. After you know all about those two, there is no extra fact about personal identity.⁴

That early formulation is deliberately strong. The useful questions concern what connects the people at different times and how much those connections matter. Asking for an additional metaphysical identity-stamp does not, in that account, add anything to the facts already described.

The later Causal Dunes work is more differentiated about what psychological similarity leaves unsettled. It separates numerical identity, continuity of the organized process, psychological continuity, and phenomenal continuation. The enduring project is not devotion to one slogan. It is discovering what must actually be preserved when Diego says he wants to survive.

This is why the phrase what matters is so important throughout his work. It shifts attention from winning a vocabulary dispute to identifying the relation for which a person prudentially cares. A successor might fail strict numerical identity while preserving something precious. Conversely, calling a successor Diego cannot establish that the precious thing survived.

The castle contains many paintings of its owner. None is automatically an additional owner.

The future does not get every vote

This account has an unusual consequence in Diego’s writing about rationality. Long-term planning is not automatically the neutral standpoint from which all shorter desires should be judged. A person exists at multiple timescales, and the timescale selected partly determines which interests look rational.

He makes the distinction this way:

But there is also time-length, or time-span conflict. The conflict that arises from the fact that you are, at the same time, the entity that will last 200milliseconds, the entity that will last one second, and the entity that will last a year, or maybe, a thousand years.⁵

The marshmallow eater and the patient future planner can represent different temporal interests. Calling the future planner the real agent, then declaring the other one irrational for failing him, already assigns the crown before the discussion begins.

Diego is not denying that delayed gratification can achieve things. He is questioning whose achievement is being maximized and whether the nearer slices of a life are permitted to count.

His article concludes the example:

A decision to eat the marshmallow can, after all, be described as a rational decision, it all depends on how you frame the agent, the child.⁶

The same concern changes the relationship between prudence and altruism. If psychological connection to a much later version of oneself weakens, that version does not necessarily deserve an unlimited premium over other people alive now. In 2014 he described what reading Parfit had done for him:

made me realize that there is as much reason to care about you, reading this, as there is to care about retired me, and it’s just cheaper to help others, far.⁷

This was one route into his distinctive cheap-altruism reasoning. The point is not that future Diego becomes worthless. It is that the boundary between self-regarding and other-regarding concern is less absolute than everyday language suggests. Distant Diego has to make his case alongside other valuable people. He cannot simply wave the same passport forever.

Past Diego can become a minor tyrant too

The long future is not the only claimant. The past can install instructions that keep governing a life after the reasons and desires surrounding their creation have changed.

In Abandoning Cached Selves to Re-Write My Source Code Partially, I've Become Unstable (2012), Diego describes letting childhood and earlier commitments acquire a kind of legislative sanctity. He called himself:

a slave of "past me".⁸

Many commands were admirable: help others, address existential risk, preserve life. They had produced a life he valued. The issue was not that every old commitment was bad; it was whether its age gave it permanent jurisdiction over current experience.

He spent two days alone thinking and writing about the relative importance of different timespans and circles of concern. Then he made an unusually deep change to his self-government. The result was not the smooth installation of a better spreadsheet:

Instead, I find myself being prone to an evolutionary fight between several distinct desires of internal selves. A mix of George Ainslie's piconomics and plain neural darwinism/multiple drafts.⁹

The account describes freedom and instability together. Removing old organizing commitments opened alternatives, but it also changed the system through which attention and action had been coordinated.

For a portrait of Diego, the episode is especially useful. His talk of subagents is not only a metaphor he applies to other people. He has used it to interpret lived changes in which forms of himself govern the others. Agency is an arrangement with a history; revising the arrangement can change what feels like the agent.

I want a life, not permanent obedience to my present shopping list

Does survival require keeping the same goals forever? Diego’s answer is no.

In a 2016 reflection after an AI-safety meeting, he contrasts models of the self and sides with one that allows the content of wants to change. He objects to treating a person as though there must be a stable master whose fundamental goals remain constant while only a subordinate system changes.

He writes:

People also usually don't care about the fact that their goal content integrity will not be preserved, seldom forcing future themselves to have the same goals as they do.¹⁰

The thought is easy to recognize outside philosophy. One can want one’s future to go well without demanding that future self keep every present hobby, ranking, or ambition. Growth changes the contents of a life. The desire for its continuation need not be a desire for permanent repetition.

He puts the positive concern in language much warmer than a theory of goal preservation:

We usually have goals we want to preserve, but mostly we want to rage against the dying of the light, and we want the experience of the world we have to be awesome through and through.¹¹

This does not mean every transformation is welcome. It means constancy of a present list of goals and continuation of what one cares about are different questions.

That distinction also explains his early concern about improving a person’s intelligence before asking the improved person what the future should contain. In his critique of coherent extrapolated volition, he asks whether increasing intelligence can gradually remove the features that made the original person the relevant beneficiary:

Maybe smarter versions of ourselves are not us at all¹²

A brilliant successor can have a perspective worth hearing without automatically acquiring the right to vote in the original’s name. An upgrade is not, by definition, faithful political representation.

The upload must preserve more than its testimonial

The practical danger becomes sharp when the proposed transformation destroys the original.

Diego has worried about this since his earliest public writing on posthumanity. By 2023 his personal preference was explicit:

When I am cryopreserved, I would prefer to be brought back in my biological original form, just defrosted, and not destructively uploaded. I am somewhat confident that destructive uploading causes death¹³

The point is not that he wants to remain composed of the same atoms forever. Ordinary life changes its material continually. Nor does he declare that no artificial substrate could ever support the relevant kind of being. The question is whether the particular procedure preserves the continuation he cares about.

The Causal Dunes collaboration makes several achievements distinct. A successor can match behavior, retain memories, implement a computation, reproduce an organization, or continue a particular experiencing being. The first achievement is not a receipt for all the others.

Consider nondestructive copying. Original Diego remains alive while a second system inherits his memories and says it is Diego. There are now two causally situated successors with a shared past and diverging futures. Destroying the original afterward removes one participant; it does not visibly add a continuity relation the copying previously lacked.

In the collaborative volume:

Even perfect instantiation of a type does not guarantee token identity. Two simultaneous, molecule-for-molecule organisms would still be two organisms.¹⁴

Gradual replacement has a different structure. If each new component enters the continuing organism’s control and maintenance before the next change, the ongoing process may remain connected in a way that scan-and-reconstruction does not. Diego wants the differences examined, not covered over by a single promotional word such as uploading.

His position also allows a copy to matter morally. A new conscious person with Diego’s memories would not become disposable because he failed to be the original Diego. The question of who continues and the question of who deserves concern are separate. There is room in the castle for a descendant; that does not require murdering the host.

Continuity is the treasure, not a synonym for stasis

A simple personality test would make Diego difficult to classify here. He values self-preservation intensely, denies that ordinary identity is a simple unchanging essence, makes space for short-lived selves, allows desires to transform, and resists treating a perfect psychological copy as automatic survival.

The combination becomes intelligible once we ask what he is protecting. It is not a frozen set of opinions. It is not an eternal soul. It is the continuation of a living, changing, experiencing organization, with its attachments and possibilities—plus an inquiry into which of those relations carries what matters.

He has long treated identity as capable of having value in its own right. Explaining a former Facebook name in 2017, he wrote:

VICH meaning Values Identity, Complexity and Happiness as a utilitarian.¹⁵

The pleasures matter. The complexity matters. The identity relation can matter too. A life is not simply an interchangeable container that may be emptied once an equal quantity of happiness has been placed elsewhere.

So the future Diego is welcome, but he is being asked for something more specific than a name. What connects him to this person? What changed? What was retained? Who benefited? And, above all, did this transformation carry the person into another chapter of his life—or leave a superbly informed reader to announce that the author survived?

Further doors

• On What Selves Are (2012): the conceptual architecture of selfhood.
• Rationality versus Short Term Selves (2012): temporal scale and whose interests count as rational.
• Abandoning Cached Selves to Re-Write My Source Code Partially, I've Become Unstable (2012): the theory entering a life.
• Conversation about personal identity after the Ai Safety and Neural Networks meeting (2016): survival and changing wants.
• Causal Dunes V: The First Cut Mattered (2026), sections 9–10: the current separation of substrate, copying, and continuation.

Continue: The Sand Spirits Are Coming