Longtermism
The future is vast, and it matters.
Longtermism is the view that positively influencing the long-term future is among the key moral priorities of our time. It starts from a premise most people already accept: future people matter. It then takes the possible scale of that future seriously. Humanity's potential lifespan can stretch across millions of years and across the reachable universe, so the number of people who could exist dwarfs the number who already have, by factors like 10²⁴ and beyond. People alive today are a brief moment next to the total stretch of human time. The constructive claim is that this generation inherits a duty similar to earlier ones: create better conditions for descendants, as ancestors fought for freedom, political autonomy, voting rights, civil rights, abolition, and rebellions against slavery with more than their own lifetimes in view.
What is longtermism?
The gentle version is easy to state. A person who will suffer in a thousand years counts no less than a person suffering today. Distance in time, like distance in space, is a weak reason to ignore someone's welfare. Philosophers Hilary Greaves and William MacAskill developed this into what they call strong longtermism in their founding academic work: the far future holds so much potential value that shaping it becomes the thing people should focus on when they are trying to do the most good.
Longtermism grew inside effective altruism and sits on ideas already present in transhumanism and cosmism. Existential risk, especially from advanced AI, gets treated as a supreme priority because a permanent truncation of the future would erase not only the living but the vast possible populations that might follow. That is why critics of the view say the present eight billion start to look small next to cosmic-scale forecasts. The positive reading keeps the living in frame while still refusing to treat the unborn as morally weightless.
Astronomical stakes
The engine of the argument is scale. Nick Bostrom's 2003 essay "Astronomical Waste" argued that every century of delay in making use of cosmic resources forfeits an immense number of possible worthwhile lives. MacAskill and Greaves work with estimates on the order of 10²⁴ future humans. Bostrom's higher figures, including those later discussed around digital minds, reach toward 10⁵⁸ potential lives. Against numbers like these, even a tiny upward nudge in the chance of a vast, flourishing future can look enormous next to many near-term interventions when people use expected-value math.
Taken to the extreme, that structure invites a known objection. When the imagined payoff is huge enough, small probabilities can start to dominate ordinary judgment. Opponents compare this pattern to Pascal's Mugging: a tiny chance of an enormous reward can be used to pressure people into almost any present cost, because the spreadsheet keeps saying the bet is worth it. Serious longtermists still need proportion. Scale matters, and so do evidence, tractability, and the risk of letting abstraction run ahead of what anyone can actually know.
How the priority shifted
Among the TESCREAL letters, longtermism arrived last, yet it often holds the others together. Its roots are in effective altruism's early focus on measurable near-term help, mosquito nets, deworming, and cost-per-life math. In the late 2010s many people in that community began calling for a longtermist turn. Within a few years, especially by 2022, preventing AI-driven existential catastrophe had moved to the center of a lot of EA identity and funding, including a large portion of grants from Open Philanthropy.
Once that shift landed, the connection to neighboring ideas got clearer. Work on posthuman capability in transhumanism, and on leaving Earth in cosmism, can look like optional sci-fi until someone asks what is at stake if the human story is cut short. Longtermism answers with scale. If later populations could be vast, then keeping the future open and improving the path into it can rank with the highest-leverage projects a civilization attempts. In that framing, misaligned AGI is often treated as the threat with the largest total downside, which is why alignment research gets cast as one of the most consequential technical problems people have faced.
There is also an older human pattern underneath the new vocabulary. People who fought for freer governments, voting rights, civil rights, abolition, and escape from domination were rarely acting only for their own comfort. They were trying to leave a wider set of choices for whoever came next. A constructive longtermism stays in that lineage: build forward so descendants inherit more capability and safety, while people alive now remain part of the moral picture.
Choice over control
The misunderstanding
Émile P. Torres and other opponents argue that longtermism gives the wider TESCREAL set of ideas a blank moral warrant. Point to enough unborn people, they say, and almost any sacrifice asked of the living can be made to look rational. In that story, spreadsheets keep promoting tiny reductions in extinction risk over fighting disease, hunger, and injustice now, until today's population looks negligible beside a promised cosmic future. They point to Greaves and MacAskill on strong longtermism, Bostrom on astronomical waste, and the late move of effective-altruist money into AI risk as evidence that ordinary moral priorities were going to lose.
The positive reality
People who do not exist yet still have moral weight, and the possible size of the human future is large enough to shape priorities. Longtermism, used well, means taking responsibility across generations: lower the odds of an early ending, improve what later people inherit, and continue helping people who are already here. That matches how earlier fights for rights and freedom were aimed at descendants as well as at the living. Work on AI alignment, biosecurity, and sturdy institutions belongs with present health and justice efforts in one forward-looking practice, judged by evidence and sense of proportion when someone offers a huge future number as the only argument.


