Other writings     Drawings gallery

What does it mean to ‘change the past’ and is it possible?


“I want to change the past,” Tina insisted. “What good’s a time-machine if it can’t even change the past?” The flickering lights on the head of Tina’s robot servant, Chipton, turned red in response. “Your request is irrational. The past is the set of events preceding the present. One cannot change a set while retaining its identity; it would be a different set. Your request is tantamount to a desire for an event to have happened and not happened at the same time.” Tina frowned and shook her head. “Oh, you and your cold, mechanical words! That’s most certainly not what I mean by ‘changing the past.’ I just want to live in a world where something else happened in 1921 – with a different set from ours if you will – what’s so irrational about that?”



I shall disambiguate between three meanings of change; replacement, relational, and essential. I then argue that a simple branching model satisfies at least one meaning while avoiding the grandfather paradox. Finally I will meet the objection that branching doesn’t really provide the possibility of changing the past, by arguing it stems from an unreasonable insistence on essential change that should, for different reasons, not be open to Lewis and Inwagen anyway.


Given the esoteric nature of the topic, asking why we should bother is perfectly reasonable. Sider (2002) reasons that metaphysicians ought to make sense of what physicists take seriously and that time-travel has farther reaching implications for other related questions in philosophy. While Sider’s reasons are both commendable and legitimate, my concerns are nowhere near as lofty. I simply prefer consistency in my time-travel fiction.

It is apparently customary to start these considerations by pointing out that time-travel to the future is trivial. Just find a comfortable spot and wait. I would also surmise that by now (post-Lewis, 1976) time-travel to the past – although having to contend with objections from free will and massive coincidence – is unproblematic as long as we do not change the past. On Lewis’ account if you today set your time-machine to 200 years ago, before you left, our history already contained the fact of your arrival in 1810 and of your subsequent actions in the past. It follows that you cannot murder your grandfather prior to him conceiving your parent because we know that this is in fact not what happened.

However, while this deftly avoids paradox, it is rather unsatisfactory for alternate history buffs. Writer, Neil Gaiman, once wrote (1997) that the most important question in fiction-writing is ‘What if?’ While it can be quite enjoyable to read about a particularly clever time-loop, the main attraction of time-travel is undeniably the exploration of ‘what if you could make it have happened differently?’ Our mission is therefore to help Tina find a possible way to murder her grandfather.

Consider the statement ‘I want my president to be another person.’ It is immediately clear that what is desirous is some sort of change but what it would take to satisfy my desire is ambiguous. Most likely the first thing to spring to mind is:

  • Currently Obama holds the title of president but I want it stripped off him and given to McCain.

I shall call this sort of change ‘replacement.’ However, perhaps not wanting Obama to be my president is not sufficient for me to wish the man stripped of his title and livelihood. Maybe what I meant was:

  • Currently I live in the USA where Obama is my president but I want to move to a country so that someone else is my president – e.g. to France and Sarkozy.

I shall call this sort of change ‘relational.’ Then again, I might be a mad scientist (or philosopher) and desire terrible things done to Obama. Perhaps I meant:

  • I want to change the very soul (or neural pathways, or whatever) of Obama so that Obama is no longer Obama but someone else.

I shall call this sort of change ‘essential.’ I make no presumption here that essential change of Obama makes sense or that my three sorts are exhaustive. I presume only that all three sorts of change are something we might mean.

When Tina expresses her desire to change the past Chipton interprets it as a request for essential change. Chipton understands Tina as saying ‘I want that particular past to both have and not have its essential properties.’ Sometimes we entertain mutually exclusive desires either because we haven’t thought them through or because we’re in two minds about what we truly want. However, assuming Tina knows what she wants and is well aware that a dead-alive grandfather is nonsense, Chipton’s is an uncharitable interpretation.

I shall return to considering replacement change and relational change momentarily. For now I shall propose, as a way to satisfy Tina’s request, the frequently referred to – and just as frequently dismissed – branching model (Lewis, 1976, calls it ‘branching time;’ Inwagen, 2010, ‘branching histories’). Let us first imagine branching without any added time-travel. Note that I shall be using verbs in my description but that this should not be taken to mean that the branching timeline is actually ‘doing’ anything. They are just for ease of description. We start with the standard timeline one might find in a history-book. For simplicity’s sake let’s suppose your birth as farthest left and your death farthest right. Feel free to add details in-between – first day of school, first kiss, this very moment, a year from now etc. The next step is imagining what-if’s and to bifurcate the timeline into a separate branch for each way it could have gone. For instance, imagine the line splitting into one go-to-school line and one stay-at-home line, etc.


If you kept dividing branches of branches for every imaginable possibility, this would fast become complicated. However, the general gist is quite sufficient for our purposes. Applying this to time travel we can look at the original timeline and suppose that a year from now you will travel back to the moment of your first kiss and stop it. Obviously future you was not originally present at your first kiss but just as there is a bifurcation between you going ahead with the kiss and you backing out at the last second, there is a third branch wherein past you was rudely interrupted in the romance by future you.

First Kiss Timeline

Hopefully it is clear how this solves the grandfather paradox. While writing this, the timeline just split into a branch wherein I kept writing and a branch wherein I was brutally slaughtered by my future grandchild, time-travelling from ‘down the line’ wherein I kept writing. This is fully consistent and both Lewis and Inwagen (op. cit.) have conceded as much (with Inwagen further conceding it as ‘real time-travel’ and Lewis being silent on that issue). However, they also unanimously agree it is not really a case of changing the past.

I can only make sense of the objection if I assume Lewis and Inwagen mean something like essential change. Although we should tread carefully to not belabour them with Chipton’s views as they are not necessarily committed to the same explication. However, we may now proclaim that while branching might not get us essential change of the past it will most certainly provide us with relational change. Just as I can move from the USA to France to change who the president is in relation to me, Tina can move from our branch to another to change the vital status of her grandfather in 1921 in relation to her.

It would seem the answer to the question ‘Is it possible for Tina to change the past so her grandfather died in 1921?’ is ‘yes, if we mean relational change.’ However, we can do better. I shall argue that whatever may be meant by change neither Lewis nor Inwagen can consistently raise the ‘not really changing the past’ objection against branching for separate reasons. Instead of focusing on what we mean by ‘changing the past’ let us turn our attention to ‘possibility.’

One theory we might hold is Lewis’ own modal counterpart theory, which “identifies possibly being F with having a counterpart—an appropriately similar object in another possible world—that is F.” (Sider, 2006) Applying this to our case what it means to assert the possibility of Grandfather being murdered by Tina in 1921 is that Grandfather has a possible world counterpart who is murdered by her in 1921. Possible worlds are not minutely analogous to temporal branches but they are sufficiently close to say that this is exactly what branches let us do. One might object that to use Tina instead of a counterpart-Tina is cheating, but since we are considering jumping branches with a time-machine I do not see why it should matter – if you must, simply pretend our time-machine has the side-effect of turning time-travellers into counterparts (whatever that means).

Possible Grandfather Timelines

What’s interesting is that the objection against branching is directly analogous to the Humphrey objection against counterpart theory. So while we have Kripke saying Humphrey “could not care less whether someone else, no matter how much resembling him, would have been victorious in another possible world” (ibid.) against counterparts, we have Inwagen (2010) saying against branching that “on the branching-histories picture of “murdering van Inwagen when he was twenty,” I have nothing to worry about: you’ll get into your time-machine and vanish, never to be seen again, and my life will go on much as before.”

We can then borrow Sider’s reply that “according to counterpart theory, the property of possibly winning is the property of having a counterpart who wins” and say that according to branching Grandfather’s property of possibly being killed in 1921 by Tina is the property of having a branch-counterpart who is killed in 1921 by Tina. To mirror Sider’s ‘permanent bachelor’ class (2002) if you still insist branch-counterpart Grandfather is not the real one, it must be because you consider the property of ‘hitherto survivor’ an essential property. I shall not argue with that; the point is the objections against branching and counterpart theory ought either stand or fall together.

Inwagen’s past-changing model is a four-dimensional block growing in hypertime. Imagine drawing our history-book timeline from left to right. As you draw, the present is always at the tip of your pencil as past events are left behind farther and farther to your left. Inwagen imagines an Intelligence observing this block as it grows, watching a person enter a time-machine resulting in the obliteration of a large chunk of block. Imagine as you were drawing the line someone inside the line started erasing it backwards so that it could be drawn anew differently.

Drawing The Present

I feel Inwagen is doing some very clever misdirection. First of all, this would be begging the question against someone insisting only essential change can count as real change. Inwagen can borrow the identity of the block from its persistence through hypertime but his is still replacement change if we continue to view the block as a mereological sum – as was my understanding of the dialectic. Moreover, it is not clear to me that his model is significantly different from branching. Consider the following thought-experiment: when Inwagen’s block hypershrinks, the chunk, instead of being obliterated, is sliced off. The Intelligence can now either incinerate it or reattach it at a different angle from the point of divergence from the hyper-new outgrowth. We would then have two branches!

The difference between replacement change and relational change was obvious regarding Obama; less so when applied to temporal ontology. Why should it matter what happens to old branches upon departure? Again it seems to me the models should stand and fall together and that the benefits of Inwagen’s model are merely sleight of hand. Branching has the real benefit of allowing past-changing (at least provisionally) while not requiring hypertime and not making time-travellers mass-murderers.

Marty McFlyI have argued that a branching model makes possible relational change of the past while avoiding paradox. While this might not satisfy Lewis and Inwagen, it should certainly satisfy Tina, who could live in a world in which she killed Grandfather in 1921. What more could she want? To insist it is not really Grandfather, is to insist 1921-surviving is a necessary part of Grandather’s identity. Not only is this a strange view of Grandfather but I have also argued that such insistence should be open neither to Lewis due to his commitment to counterpart theory – in which losing the election cannot be a necessary part of Humphrey’s identity – nor to Inwagen due to his hypergrowing block’s helping itself to non-essential change. If counterpart theory and Inwagen past-change are consistent then so is branching. If branching fails against the not-really-change objection then so, for the same reasons, should counterparts and Inwagen past-change. To boot branching allows a sort of past-change, lets us retain eternalism if we want, doesn’t require hypertime, and avoids mass-murder.

Mr. Strickland: No McFly ever amounted to anything in the history of Hill Valley!

Marty McFly: Yeah, well, history is gonna’ change. (Back to the Future, 1985)


Share and Enjoy:
  • Facebook
  • Twitter
  • Digg
  • StumbleUpon
  • Reddit

6 Responses to “What does it mean to ‘change the past’ and is it possible?”

  1. Martin White says:

    I would like to travel back in time and fix up a mistake that ruined my life and my family’s. Could you please help me. I need to do this.


    Martin White

    • Dear Martin
      I’m afraid I cannot help you to actually travel back in time. I am but a philosopher. However, I would suspect that there is no help to be found regardless of profession. Technology simply has not yet advanced sufficiently to enable humans to travel back in time, and unfortunately it might never do so!

      You are obviously in pain and for that you have my sympathies. If you will take some advice, I would urge you to make the best of what you have and try to look forward rather than back. I do not know what this mistake of yours was. But try doing your best to make amends to whoever might have been wronged. Work on making life of your family and yourself better tomorrow instead of dwelling on how it could have been better now, if only a different path had been taken yesterday. I am not telling you to forget the mistake, which now defines you, but learn from it, realise that you are now a different person, and forgive yourself and others for making the wrong choices. If others are not so quick to forgive and bear grudges against you, take responsibility for your actions but discard the guilt. If you have done what you can, the deficiency is theirs and no longer yours. Realise that you are only human and that there are things outside of your control – especially concerning the past – and let go emotionally.

      I hope my words have provided at least an inkling of help to compensate for my lack of a time-machine.

      Kind regards and best of wishes

      P.s. if you need to talk, you know where to find me.

  2. Steven Robert Gill says:

    The problem for me is that whilst this is more like old school time travel than other theories that use multiple universes, this still presumes a sort of closed loop of events, just this time the loop crosses universes as well as time.

    So you haven’t gone back and realised that you were ”part of history all along”, rather you’ve ensured that another universes history played out as they remember it. So what you’ve done is travel back to a bifurcation point in time that was already there, meaning there would in theory have to be a branching point for .every. possible time travel event (which is absolutely plausible given the many worlds theory, if a bit, well, messy)

    So whatever you want to alter, don’t worry, because It already .has. been altered. But if you don’t go back then no problem, because someone else from another 2011 just like ours would then travel back and do what you chickened out of doing.

    It would look something like this.


    • Well done! I’m impressed by your acuity. That’s exactly how it works. Why is this a problem for you though?


      P.s. Sorry if it took my a while to approve your comment. It was caught in my spamfilter because of your included link.

  3. Steven Robert Gill says:

    Thank you for the reply!

    Well, I haven’t got a problem with it per-se, it’s just a very complex way of thinking about it that’s all.

    Some modern physicists actually believe that time could be thought of as a series of still 3D frames of the universe ‘stacked’ on top of each other via a fourth spatial dimension, In a sense our word line could be thought of as a incredibly complex 3D flipbook. So those that say time is the 4th dimension and those that say it’s a physical dimension, could in a way, both be correct.

    • I agree. It is complex. I was just trying to circumvent the grandfather paradox though. The circumvention turned out to be complex.

      Also, if you want to break your brain a little bit, you can consider whether or not branching times requires hypertime. (Or more than 4 dimensions.) If I recall correctly, Inwagen said it didn’t. Not sure if I agree.

Leave a Reply