If I delete and replace a character in a computer document, to what extent is it the same character?

Question

This is a question partly about philosophical terminology and categorisation as well as existence.

Suppose I type the random string "janwekxq" into a Word document. I then accidentally delete the letter "e" leaving "janwkxq". Finally I retype the "e" to restore normal spelling.

To what extent is this the same letter "e"?

In computer terms

There is no guarantee that the new "e" is stored in memory in the same location as the old one.

The document may be saved and reloaded. It may or may not be displayed on the same portion of the screen each time.

The information may be considered equivalent in terms of algorithms and operations on data even though stored differently.

Question

In philosophical terms, how can we categorise this letter "e" as an entity that has a lasting existence in a particular document and on a particular computer? It seems to me that we can throw away any meaning because we used a random sequence of letters.

Answer 1

What constitutes “a letter”?

One very useful rule of thumb is that entities ought not be multiplied beyond necessity. This sometimes treated as a variation on Occam’s Razor - that in the absence of any distinguishing evidence, the simpler explanation is to be preferred.

A corollary of this was proposed by Willard Van Orman Quine with particular reference to analytical theory-crafting. In his 1948 paper “On what there is”, Quine discusses the idea of what the ontological commitments of our statements are, coming down on the idea that we are committed to those aspects of our referred terms that feature as values in a scheme of quantified variables, in accordance with our best operative model of how the world is.

Being precise, then, we ought to say that your “letter e” doesn’t really exist. Exactly what change in the state of the machine in front of you has been made as a consequence of your pressing the buttons in front of you must depend on the implementation of the Word Processing software you’re using and how that application makes use of RAM chips on your device (presumably via an OS abstraction if some sort), but ultimately there is no simple “letter e” that is contained on the image on the screen - the display of a character is an emergent property of a complex underlying system. It looks like nothing had changed from your perspective, because the effects of the changes made has resulted in identical displays on your monitor - however it could well be the case that something fairly radically different has taken place. Consider e.g. cloud-based word processing like Office365.

If however you decide that there is some value in introducing a different kind of ontology, you might find it helpful to borrow another pithy Quineism, which is to insist on “no Entities without Identities”. What difference does the individuation of “letters” make in your model - when should we take two letters to be “the same”, and when should they be numerically different? If you can give clear and distinct “identity conditions”, this is at least a starting point for using an existential idiom.

If by “that letter” you mean something like “the character in position x in a linear sequence of alphanumeric characters in the working memory corresponding to this word processing file”, that’s fine. Programmers and data analysts do that a lot, and it comes with a neat answer to your question - it is an identical e because the same conditions are met - the same character in the same position in the same file.

But that might not be what you meant. It helps to be specific about it, and hopefully the rigor of philosophical language analysis will be useful in asking those kinds of questions of your casual turns of phrase.

Answer 2

Each piece of data has multiple addresses as means of maintaining identity over time and after changes.

This is similar to humans having a name, a postal address, an email address, a birth certificate, a social security number, a driver's license, fingerprints, DNA ... For each of those (except maybe DNA), they can get lost (in accidents) while the person lives on. Or they can be changed. Note even DNA can be problematic in the case of twins (and other clones).

In a computer, the physical memory address of the letter e of your example may change all the time, even without human action, and often the same screen letter will have an address on disk, in RAM, in different caches, and in CPU registers. There is also a logical memory address, which might also be changed without any human interaction. Software may additionally have an arbitrary number assigned to data objects like character strings, by which those may maintain identity even when memory location changes.

Whether any of those memory addresses constitute the same letter or not is a matter of definitions as regulated by the software.

From the users point of view, most modern writing software will consider document versions as part of the ultimate identity, so your document "janwekxq" might have a version 42, and after deleting and adding again the letter "e", the document would have version 44, thus a distinct identity (unless you used the undo function to restore the previous version). But not all software maintains such versions, showing there are multiple ways to interpret identity of software data.

Identity must sometimes also be maintained in networks, such as the letter "e" on your screen and on my screen. Consider as an example an electronic currency like Bitcoin, or digital transactions with banks or online shopping sites. So the issue is known to be larger than pointed out in the question.

Philosophically, such identity transfer is a freely definable contract between all participants agreeing on what shall be considered identical.

A somewhat similar concept outside software might be the identity of the Dalai Lama soul being considered reborn in other individuals. Or the James Bond movie character played by different actors.

Outside of maybe atomic physics, identity across timespans is always a matter of definitions and contracts, with physical continuity just being the default template for ordinary life, and self-consciousness of intelligent agents like humans being a constant source of philosophical conflict.

Answer 3

Identity is relational — nothing has identity except in relation to other things — and thus you cannot really talk about identity without talking about scope. With respect to that string of letters, the initial 'e' and the final 'e' are the same object, because they occupy the same role with respect to the string of letters. With respect to computer memory they are different objects; that different scope entails all the unseen technical differences between the two.

This question is no different from asking whether two copies of a novel are the same novel. They are obviously different books (on the physical scope), but they are obviously the same story (on the literary scope). It's also no different from asking whether you or I are the same person today as we each were yesterday. There are innumerable physiological and psychological changes that occurred in our bodies and minds overnight, but those kinds of changes are normally ignored when we establish our identity.

One needs to establish a frame of reference before we can begin to establish identity; we will not know how to respond to someone asking after some 'john' unless we can suss out whether he's looking for a person or a bathroom. Find the frame of reference, and (generally) identities will resolve themselves.