Reading memo: 'Narrative Collapse' by L.M. Sacasas

Source citation

Sacasas, L. M. 2020. ‘Narrative Collapse’. Substack newsletter. The Convivial Society (blog). 14 June 2020.


Narratives are a technology, and should be seen as such. It is not merely that technologies shape narratives (though they do), but that the narratives themselves are also technologies. Not in the sense that they are material artefacts, like (say) a computer or a pencil. However, if an algorithm is a technology---a procedure for sorting data which generates understanding and a kind of sense---then so is a narrative, albeit one we tend to take for granted. They are

"deceptively simple technique[s] deployed by the human mind in order to make sense of the world"

If we find the claim difficult to credit, we may only need to think about the distinction between asking what I am and asking who I am. We might answer the first without a story, but it's hard to imagine answering the second, because that's a question of meaning, and moreover one that relates to "our fundamentally time-bound existence":

We experience life as a succession of moments yielding a discernible past, present, and future

However, narratives are also significant because they are selective, and they have purpose: they include or leave out events so as to present meaningful relationships: not just what happened but why and what was significant about it. We might reflect at this point about the importance of narrative in generating a sense of first-person ethics: we don't merely reflect that some moral principle or other is relevant to us just because it's right, but because it's right for us at some specific point in our own existence. The narratives we tell are therefore significant in that they influence the way we see the rest of the world:

Stories [act] as a filter on reality. We never merely perceive the world, we interpret it. In fact, our perception is already interpretation. And the work of interpretation depends to no small degree on the stories that we have internalised about the world. So when we hear about this, that, or the other thing happening, we tend to fit the event into our paradigmatic stories.

This seems to me like a pretty apt description of the kinds of thinking and ethical presentation that people are going through when they reach for explanations about how they came to do bad things, such as murders.

Sacasas makes some very interesting points specifically in relation to online culture. He is writing about online culture, an 'information ecosystem' organised more according to the logic of a database than according to that of a narrative. This point needs some elaboration.

His basic argument is that the move to digital technology makes narratives, as a technology, 'glitchy', in that they cease to operate quite as we might expect them to. Drawing on Marshall McLuhan, he argues that the scale, pace and pattern of human communication online have all changed. We're all used to hearing about the first two of these: that we encounter more information, at a greater pace, than former generations. The third is subtler: that the 'pattern' of information brings it to us in novel social contexts and in a form that more closely resembles a database than a story.

Illustrating this point, Sacasas draws on media scholar Katherine Hayles, who points out (Sacasas's paraphrase), that:

when you read a narrative [...] you are encountering the product of a series of choices that have already been made for you by the author out of a myriad of possibilities from the database of language. The countless other choices that were possible are present only to the imagination. You see the words the author chose, not the ones she could’ve chosen. You see the path marked out for you as a reader, not the multiple paths that were rejected.

On the other hand:

When you encounter a database [...] you see the opposite. You see the field of possibility and any number of paths through the database remain hypothetical and potential.

The data points in a database exist in their own right. Their significance and their meaning is not baked into the form in which we encounter them. We can navigate them with a good degree of independence, relative to following a narrative whose beginning, middle and end are laid out already.

Caveat: none of this means we don't encounter narratives in the online, digital world, but it does mean that we encounter a huge proliferation of them, and that we can also encounter many that are mutually contradictory, from which it is easy to make selections, and which are easily countered by other competing narratives, or by the inclusion of information which a narrative's authors have omitted. This kind of 'jamming' or contradiction of a narrative was simply less easily achieved when information was offline, and the reason is that easy access to information hugely changes the balance of power between the authors and consumers of a narrative.1

A final comment of Sacasas's in respect of narratives and databases is that both are 'mnemonic technologies': their function is to store information. Narratives achieve this by organising data points into stories. Crucially, however, they are also "instruments of forgetting": they leave things out, and indeed must do so in order to function, even in quite trivial instances: when someone asks us how our day went, we don't try and describe every breath, footstep or nose blow, and instead discard an enormous amount of what happened. In this sense, narratives differ drastically from databases, a form of "untamed memory" which "undermines the traditional functions of Narrative" precisely because it exists only to store information ready for retrieval and reuse. Once entered in databases, data points are not often processed further or changed.2 Sacasas goes further, quoting Manovich to sugges that databases and narratives are "natural enemies":

The Database3 is blind to traditional categories such as credibility or trustworthiness. The Database is indifferent to truth. All entries in the Database have the same value, although they can be differently organised. This is confirmed by the mostly futile efforts of the Narrative-minded managers of the Database to artificially impose these categories through “fact-checking” notices, warning labels, blue checks, etc.

These insights can help us think about the kinds of narrative struggles long-term prisoners engage in with their captors. It is not that all of these points translate directly to prisons, which remain a resolutely offline world. Seeing the records a prison holds about a prisoner as somehow 'non-narrative' would be mistaken, since much of what can be found in the various information systems is highly selective, highly structured, seeks to be 'authoritative' by adopting various more or less consistent authorial procedures. In some cases (for example, a judge's sentencing remarks or an OASys offence summary), data about a prisoner are entirely 'narrativised'.

Scholars of punishment have long noted the modern tendency to classify, categorise and typify, and it is no coincidence that prisoners' resistance efforts have often focused directly on the various information technologies which have, at different points in time, facilitated these processes. Risk assessment does precisely this, and some of the following entries from Sacasas's "Provisional Inventory of Consequences" are highly suggestive with respect to risk assessment technologies and procedures, as I hope my comments will make clear:

  1. The Database tolerates, indeed encourages narratives, but it cannot sustain and actively discourages Narratives.

  2. All narratives generated from the Database are tenuous and subject to constant revision. They are but one possible path through the database. Everyone knows alternative paths are possible.

This resonates strongly with the words of prisoners who complain that the progress they have made in prison---the ethical changes they have undergone, the advancements in character they perceive in themselves---receive no credit. In short, this is because when it comes to interpreting their actions, "alternative paths are possible", and indeed any amount of discrediting information, starting with a serious index offence, is available to question their claims. Indeed, they are encouraged to develop 'penal avatars' [cite] and tell stories about themselves, but only a very few appear to sustain Narratives---in the sense of authorial and authoritative control over how their stories will be read and understood. This is in part because:

  1. Narratives seek closure (the story must end). The Database is open-ended (it assimilates new data indefinitely). The Database resists the Narrative impulse to control and stabilise meaning.

Relatedly, the imperative for closure rests on the time-bound nature of narratives: they are "unavoidably temporal", whereas databases are "indifferent to time". The fact that the prisoner may have changed is immaterial, from the Database perspective: they are still, by virtue of the fact that they have committed acts of violence in the past, at an elevated risk of doing so in the future, a rock-solid statistical fact which remains so even if other physiological, psychological or existential changes have rendered a particular individual at a particular stage of their life less potentially harmful than in the past.

Sacasas also reflects on what he calls 'narrative collapse'; little reflection is needed here beyond pointing out that this state aptly describes the 'ruptures' experienced by early-sentence stage prisoners:

  1. The inability to establish Narratives yields an experience of perpetual flux, unsettledness, instability. It amplifies a sense of disorder. Consequently, it can also yield the impulse to impose order by whatever means. Losing your Narrative is traumatic.

As Sacasas points out, the undermining of Narratives is a feature of the technological ecosystem, not a bug---and as such there is a kind of inevitability to losing Narratives. Those who try to establish authoritative authorial control over their narratives are generally doomed to fail:

  1. The worst possible position to be in is that of believing you can still weave a Narrative to tame reality and generate consensus.

This is because in a world driven by the logic of the Database, information may always be available to undermine a narrative. It is important here to point towards the power imbalances which persist in prisons. Sacasas makes the opposite point here, arguing that the very availability of information online makes it harder for those in positions of authority to establish firm, consensus Narratives:

  1. The Database dramatically expands access to information, challenging the authority of even the most venerable professional weavers of Narrative. Official narratives are just one more datapoint, one more entry in the Database, one way among many of generating a path through the entries.

Yet in prison, power disparities greatly reduce the scope for effectively resisting official, authoritative Narratives (however well-founded), at least if the aim is to counter them with something else that will usurp their authority. As point 5 above suggests, prisoners do not tend to find it easy to lose their control over their own narrative:

  1. [...] Because narrative is so critical to our sense of identity [...] we are tempted to zealously guard our narratives.

And yet because the Database cannot account for, and indeed is uninterested in, development over time, "redemption [...] can only make sense in narrative terms', because 'only narrative can temporally relativise the meaning of words and deeds'. All entries in a database, in principle, "carr[y] equal weight" and as such, this technology does not readily afford its users authoritative judgements that a person has been 'redeemed' or that they are 'rehabilitated' or that they have been 'punished enough'. Only human acts of interpretation can do that---and they will always be subject to contestation, because the narrative that they represent still exists in a wider context where other information is available to undermine it:

  1. [...] the only way to manage memory in the Database, and thereby attempt to establish a N/narrative, is to delete entries altogether. Otherwise, your N/narrative could be upended at any moment by the unaccounted for entry.


  1. The rhetoric of persuasion in the Database amounts to one principle: repetition.

Perhaps most intriguingly, Sacasas suggests that the tension between Narratives and the age of the Database is indicative of the mismatches of cultural and technological scale which decision-makers operating risk assessment technologies must operate within. Sacasas suggests that:

  1. Narratives may not be adequate for understanding the complex reality that confronts us, but they may nonetheless be necessary to get us to do act [...] responsibly in the face of that reality. In other words, we’re now operating at a scale for which our most basic cognitive tool may no longer be adequate.

This, I think, accounts well for the fundamental conflict between prisoners and their captors when it comes to risk assessment. It is more interesting to consider how this conflict takes place between lowercase narratives rather than between lowercase (first-person biographical) narratives and uppercase (authoritative) Narratives. I take it for granted that Sacasas's points about authoritativeness/authorship are sound, and if this is so then it is reasonable to assume that prisoners will generally be unrealistic if they assume they can remove the discrediting stain of the conviction or delete Database entries relating to them. Indeed, the fact that so many abandon efforts at appealing their conviction might be taken as empirical evidence on this point, since even those who are exonerated often fail to lift the burden of the conviction. So I will leave aside the question of authoritativeness, and think instead about about the compatibility of two parallel narratives.

If narratives are an essential ordering device enabling 'forward progression' by setting past actions in relation to present and future ones, then for prisoners to maintain a view of themselves as worthy beings with a worthwhile future, against the stain and stigma of a murder conviction, they must craft narratives about themselves which make sense of their offending and offer reassurance that they are able to manage the risk that it might occur again. These narratives are a necessity if they are to persuade risk assessors that some of their liberties may (even cautiously and provisionally) restored. This narrative belongs to their 'penal avatar', concerns the offence and their 'risk', and exists in relation to a wider Narrative defined by their conviction. If they are comfortable with what it communicates about them and their moral status, then it may be a more comfortable 'fit', and the representational choices it implies for them will be less burdensome.

However, they may also maintain parallel narratives which are broader in context, less tightly defined by the conviction, and more strongly associated with the ethical commitments and experienced and remembered facts of their biographies. These narratives belong to their 'first-person' ethics, concern their life circumstances, values and aspirations and exists in relation to myriad other N/narratives relating to their lives. Where this personal narrative is less compatible with the 'penal avatar' version of themselves---where, for example, they insist that it must include ethically important information about their status as persons which risk assessment deems to be irrelevant, or where they insist that relevant information about their offending must be taken into account---they will experience this as painful and dishonouring.

This is a subtler point than saying that "those who maintain innocence find risk assessment painful". Rather, it insists that the Database-driven rationality of risk assessment and risk management is aggregative, anti-temporal, labelling, and indifferent to truth. It seeks to manage the risk of harm, and in this sense it is calculating and normatively minimalist. And prisoners can understand its rationality. However, seen from the individual scale it takes on the characteristics of what Bernard Williams calls a 'morality system'. Its rationality is of a third-person kind, and it is rational at a scale which frustrates narrative and precludes redemption. For those who live in shame at their offences, or in terror of a deeply discrediting repeat, the lowered status and circumscribed agency are a price worth paying. But for those who do not, or who honestly find their offences and their risk to be somewhat exculpated by wider context, or superseded by subsequent ethical progress, the constraints that this 'morality system' places on their autonomy and their self-representation are experienced as deeply painful.

  1. Sacasas points out the etymological link between 'author' and 'authority' via the Latin root word, 'auctor'---the crisis of authority in a digital media ecosystem is in some sense a consequence of shifts in the balance of power between those in different social positions who claim the authority to be authors, and to tell a given story. 

  2. As anyone who has worked with a dataset for research purposes will know, one of the first tasks required is to understand how to 'clean' and re-curate data, so that they are usable for whatever analytical procedures are intended. 

  3. Towards the end of this piece, Sacasas uses upper- and lower-case lettering to distinguish between different senses of his uses of 'narrative' and 'database', as follows: "uppercase D, Database [refers] to [the] digital media environment taken as a rather messy whole. [U]ppercase N, Narrative, impl[ies] a traditional deference to and reliance on authoritative, comprehensive narratives." In other words, lower-case references to both terms denote singular instances of these technologies, rather than their ecological aggregates.