Data vs Capta

ctietze

I didn't know this distinction, but it makes it very easy to express something I usually beat around the bush in writing to get across.

Data are considered objective “information” while capta is information that is captured because it conforms to the rules and hypothesis set for the experiment.

Johanna Drucker: Graphesis, 2011 -- turned into a book by 2017
https://peterahall.com/mapping/Drucker_graphesis_2011.pdf

Data/Capta is nice when operationalized so that you get a name for "this is a selection from things in the world, mediated also by the choice of tools of observation"; it's a classic problem of philosophy of science, that how your tools of perception and measurement influence what can even be 'seen'. (I'm rusty on that front, but that's the association I hang this on.)

So where data is given, capta is pre-selected. Semantically this makes room for data to not be relative to the observer, an objective reality, in practice I don't know the extent of that set

harr

@ctietze said:
I didn't know this distinction, but it makes it very easy to express something I usually beat around the bush in writing to get across.

The idea can also be expressed with the term "situated data":

Lavin M. Why digital humanists should emphasize situated data over capta. Digital Humanities Quarterly. 2021;15(2). https://dhq.digitalhumanities.org/vol/15/2/000556/000556.html

@ctietze said:
turned into a book by 2017

Do you mean this book? https://www.hup.harvard.edu/books/9780674724938

Andy

In typical versions of Chris Argyris's ladder of inference model (see the linked Wikipedia article for an example), "capta" would be the bottom rung of the ladder, and "data" would be the ground on which the ladder is situated.

Building on @harr's reference, one could also differentiate between situated data and situated capta using the same ladder of inference model: situated data are the data available wherever the ladder is situated (but the ladder could be moved to somewhere else), and situated capta are selected from the situated data. But the whole ladder is always situated.

tomp

@ctietze said:
I didn't know this distinction, but it makes it very easy to express something I usually beat around the bush in writing to get across.

Data are considered objective “information” while capta is information that is captured because it conforms to the rules and hypothesis set for the experiment.

Johanna Drucker: Graphesis, 2011 -- turned into a book by 2017
https://peterahall.com/mapping/Drucker_graphesis_2011.pdf

"Capta" is not a standard term of art in the sciences. I don't understand that definition and I'm not even sure it's coherent. I first thought it was describing what has come to be called "pre-registration" in statistics, but that's probably not it. "Data" is not "objective information", but the result of experiment or observation of some kind. You could say that the kind of data that gets observed is the result of the kind of measurement, but that's almost a tautology.

Anyway, the term only appears twice in the linked paper, which seems to be about the benefits of using visualization techniques to help understand various concepts and ideas, even where the subject matter at first seems to be abstract. I didn't see how it has anything to do with that definition of "capta".

So I don't know what this distinction is expressing. I might even agree with it if I understood it, who knows?

Andy

@tomp said:

"Capta" is not a standard term of art in the sciences. I don't understand that definition and I'm not even sure it's coherent. I first thought it was describing what has come to be called "pre-registration" in statistics, but that's probably not it. "Data" is not "objective information", but the result of experiment or observation of some kind.

It's true enough that capta is not a common term in the sciences (Johanna Drucker, @ctietze's source, is a humanities scholar), but it seems that you didn't check the literature before commenting, because the distinction has been made elsewhere and is not incoherent; for example, geographer Rob Kitchin says in his chapter "The Nature of Data" (in Data Lives, Bristol UP, 2021) (emphasis added):

This chapter examines the nature of data from an etymological, philosophical, and technical point of view. Data is derived from the Latin dare, meaning 'to give'. In general use, however, data refers to those elements that are taken. Technically, what is understood to be data are actually capta (derived from the Latin capere, meaning 'to take'); those units of data that have been selected and harvested from the sum of all potential data.

Kitchin's view is congruent with what I said about how the distinction maps onto Chris Argyris's ladder of inference.

harr

@tomp said:
"Capta" is not a standard term of art in the sciences. I don't understand that definition and I'm not even sure it's coherent. (…) Anyway, the term only appears twice in the linked paper (…)

The term was proposed by Drucker in an influential article:

• Johanna Drucker: Humanities Approaches to Graphical Display https://doi.org/10.63744/r4ysrh7ae534
• Google Scholar lists 1200+ citations

This is where the term appears (emphasis in the original text):

    "This requires first and foremost that we reconceive all data as capta. Differences in the etymological roots of the terms data and capta make the distinction between constructivist and realist approaches clear. Capta is “taken” actively while data is assumed to be a “given” able to be recorded and observed. From this distinction, a world of differences arises. Humanistic inquiry acknowledges the situated, partial, and constitutive character of knowledge production, the recognition that knowledge is constructed, taken, not simply given as a natural representation of pre-existing fact."

And this describes the goal of the article:

    "The polemic I set forth here outlines several basic principles on which to proceed differently by suggesting that what is needed is not a set of applications to display humanities “data” but a new approach that uses humanities principles to constitute capta and its display. At stake, as I have said before and in many contexts, is the authority of humanistic knowledge in a culture increasingly beset by quantitative approaches that operate on claims of certainty. Bureaucracies process human activity through statistical means and when the methods grounded in empirical sciences are put at the service of the social sciences or humanities in a crudely reductive manner, basic principles of critical thought are violated, or at the very least, put too far to the side. To intervene in this ideological system, humanists, and the values they embrace and enact, must counter with conceptual tools that demonstrate humanities principles in their operation, execution, and display. The digital humanities can no longer afford to take its tools and methods from disciplines whose fundamental epistemological assumptions are at odds with humanistic method."

The article's conclusion compares capta and data:

    "But the idea of capta as fundamentally co-dependent, constituted relationally, between observer and observed phenomena, is fundamentally different from the concept of data created as an observer-independent phenomena."

I think the term "capta" didn't catch on precisely because "all data is capta". We already have a word for captured data, it's called "data". :-) The word itself doesn't provide new insights. But it prompts an interesting discussion about subjectivity and context in capturing, processing and presenting data.

I find Lavin's take (see earlier post) on the etymology of the word "data" interesting, because it traces it back to Greek origins and because it distinguishes different uses in English in the last few hundred years.

Andy

@harr said:

I find Lavin's take (see earlier post) on the etymology of the word "data" interesting, because it traces it back to Greek origins and because it distinguishes different uses in English in the last few hundred years.

I found Kitchin's chapter "The Nature of Data", which I mentioned above, independently in a literature search, but on returning to Lavin's article, I see that he quotes an earlier book by Kitchin:

Another way of looking at Drucker’s influence would be to say that there is widespread agreement with her views on data and capta, but practical considerations outweigh other concerns. Such practical considerations include (1) maximizing clarity for particular audiences, (2) limits on space that would make it difficult to explain the choice to use capta, and (3) the question of how to invoke concepts like big data, data visualization, databases, datasets, metadata, and open data. One might attempt to stake out a rhetorical middle position by mentioning that data are really capta and to proceed with the word data thereafter, as Rob Kitchin does in The Data Revolution: Big Data, Open Data, Data Infrastructures and Their Consequences (2014). Kitchin adds, “since the term data has been so thoroughly ingrained to mean capta, rather than confuse the matter further it makes sense to continue to use the term data where capta would be more appropriate” [Kitchin 2014]. This seems like it would be an appealing option for many people, and I suspect it is widespread, although I have not investigated this question in any serious way.

@harr's quotes from Drucker's "Humanities Approaches to Graphical Display" sound a bit as if Drucker is engaged in a culture war of the humanities against the sciences, so I think it's important to temper the first quoted paragraph from that article with the subsequent omitted sentences that dispels the culture-war impression:

My distinction between data and capta is not a covert suggestion that the humanities and sciences are locked into intellectual opposition, or that only the humanists have the insight that intellectual disciplines create the objects of their inquiry. Any self-conscious historian of science or clinical researcher in the natural or social sciences insists the same is true for their work. Statisticians are extremely savvy about their artifices. Social scientists may divide between realist and constructivist foundations for their research, but none are naïve when it comes to the rhetorical character of statistics.

tomp

@Andy said:
@tomp said:

"Capta" is not a standard term of art in the sciences. I don't understand that definition and I'm not even sure it's coherent. I first thought it was describing what has come to be called "pre-registration" in statistics, but that's probably not it. "Data" is not "objective information", but the result of experiment or observation of some kind.

It's true enough that capta is not a common term in the sciences (Johanna Drucker, @ctietze's source, is a humanities scholar), but it seems that you didn't check the literature before commenting, because the distinction has been made elsewhere and is not incoherent; for example, geographer Rob Kitchin says in his chapter "The Nature of Data" (in Data Lives, Bristol UP, 2021) (emphasis added):

This chapter examines the nature of data from an etymological, philosophical, and technical point of view. Data is derived from the Latin dare, meaning 'to give'. In general use, however, data refers to those elements that are taken. Technically, what is understood to be data are actually capta (derived from the Latin capere, meaning 'to take'); those units of data that have been selected and harvested from the sum of all potential data.

The word is not in the Merriam-Webster or the Oxford English Dictionary. Merriam-Webster for "data":

• "factual information (such as measurements or statistics) used as a basis for reasoning, discussion, or calculation".
• "philosophy : information that is output by a sensing device or organ and that must be processed to be meaningful"

My own understanding of the normal use of "data" is in accordance.

I'm less interested in who and how many use the word "capta" as I am in what the actual distinction is that you have in mind because I still don't know. Even those two descriptions you've quoted don't seem to say the same thing.

selected and harvested from the sum of all potential data.

This is reminiscent of some statistical methods in which the actual sample elements is regarded as being drawn from a much larger (possibly infinite) pool of elements.

tomp

This is reminiscent of some statistical methods in which the actual sample elements is regarded as being drawn from a much larger (possibly infinite) pool of elements.

A correction:

This is reminiscent of some statistical methods in which the actual sample elements is regarded as being drawn from a much larger (possibly infinite) pool of elements.

ctietze

@tomp my original post consistent of just what you asked for, a distinction

Data are considered objective “information” while capta is information that is captured because it conforms to the rules and hypothesis set for the experiment.

harr

@Andy said:
@harr's quotes from Drucker's "Humanities Approaches to Graphical Display" sound a bit as if Drucker is engaged in a culture war of the humanities against the sciences, (…)

Drucker isn't fighting science, but a certain belief within the humanities (paragraph 49, emphasis added):

My argument is a polemical call to humanists to think differently about the graphical expressions in use in digital environments. A fundamental prejudice, I suggest, is introduced by conceiving of data within any humanistic interpretative frame on a conventional, uncritical, statistical basis. Few social scientists would proceed this way, and the abandonment of interpretation in favor of a naïve approach to statistical certainly skews the game from the outset in favor of a belief that data is intrinsically quantitative — self-evident, value neutral, and observer-independent. This belief excludes the possibilities of conceiving data as qualitative, co-dependently constituted — in other words, of recognizing that all data is capta.

And (par. 51):

(…) the humanistic concept of knowledge depends upon the interplay between a situated and circumstantial viewer and the objects or experiences under examination and interpretation. That is the basic definition of humanistic knowledge, and its graphical display must be specific to this definition in its very foundational principles. (…) If we don’t engage with this challenge, we give the game away in advance, ceding the territory of interpretation to the ruling authority of certainty established on the false claims of observer-independent objectivity in the “visual display of quantitative information.”

Drucker critizises a "mechanistic approach to realism" (par. 5) and "mechanistic or naturalistic realist representations of pre-existing or self-evident information" (par. 7).

I find the concept of "capta" difficult to grasp (par. 14, emphasis added):

Capta is not an expression of idiosyncracy, emotion, or individual quirks, but a systematic expression of information understood as constructed, as phenomena perceived according to principles of interpretation. To do this, we need to conceive of every metric “as a factor of X”, where X is a point of view, agenda, assumption, presumption, or simply a convention. By qualifying any metric as a factor of some condition, the character of the “information” shifts from self-evident “fact” to constructed interpretation motivated by a human agenda.

I find the graphical aspect of the paper much more interesting. Drucker believes that the "rhetorical force of graphical display is too important a field for its design to be adopted without critical scrutiny and the full force of theoretical insight." (par. 7) I don't find the examples convincing, but I appreciate the exploration of visual representations that center subjectivity.

Andy

@tomp said:

"philosophy : information that is output by a sensing device or organ and that must be processed to be meaningful"

and

This is reminiscent of some statistical methods in which the actual sample elements is regarded as being drawn from a much larger (possibly infinite) pool of elements.

The first quoted line, defining data as information that is output by a sensing device or organ, and the second quoted line, on statistical sampling, are very much connected via summary statistical perception.¹ Our perceptual experience is a kind of statistical capta of the data output by our perceptual organs, not to mention the data that could potentially be output by more sensitive sensing devices!

A possibly relevant analogy has also been made between perception and experiments.²

---

1. For example: David Whitney, Jason Haberman, and Timothy D. Sweeny. "From textures to crowds: multiple levels of summary statistical perception". In: John S. Werner & Leo M. Chalupa, editors. The New Visual Neurosciences. Cambridge, MA: MIT Press, 2014, pp. 695–709. ↩︎

2. Karl J. Friston, Rick A. Adams, Laurent Perrinet, & Michael Breakspear (2012). "Perceptions as hypotheses: saccades as experiments". Frontiers in Psychology, 3, 151. ↩︎

tomp

@ctietze said:
@tomp my original post consistent of just what you asked for, a distinction

Data are considered objective “information” while capta is information that is captured because it conforms to the rules and hypothesis set for the experiment.

My problem here is that the words that follow "while capta" don't convey anything useful to me. To the limited degree I can follow, they seem to say that any experimental outcome that disagrees with the hypothesis behind the experiment are are not actually information (or maybe not actually "capta").

I imagine and hope that's not what you have in mind.

Andy

@harr said:

Drucker criticizes a "mechanistic approach to realism" (par. 5) and "mechanistic or naturalistic realist representations of pre-existing or self-evident information" (par. 7).

Lavin, in footnote 2 of his article, rightly criticizes Drucker's discussion of the opposition between realism and constructivism as too simplistic. She's not wrong about the problems with naive realism, but varieties of critical scientific realism are undiscussed.

ctietze

@tomp Start simpler -- you do an experiment to measure the time something takes to travel a distance. You measure the distance, then set the experiment in motion and stop the time. This captures the velocity (distance-per-time). Per experiment design, it doesn't also tell you how the thing tastes that traveled, or whether it had a nice day, or wanted to move at all. That's all not available info in this setup.

What you see is all there is -- but by virtue of seeing, you don't capture what's audible.

Does that make more sense than the dense quote?

ZettelDistraction

@ctietze said:
@tomp my original post consistent of just what you asked for, a distinction

Data are considered objective “information” while capta is information that is captured because it conforms to the rules and hypothesis set for the experiment.

Is there a suggestion that capta, so called, is suspect? Suppose I limit the bandwidth of my receiver to reduce interference from neighboring signals outside some range of interest. I orient a directional antennal to null out noise generated by nearby switching power supplies. Now I am ideologically suspect because I have excluded signals and noise that deserved equal time with the signal I am interested in, and worse, I decoded a transatlantic signal below the noise level.

tomp

@ctietze said:
@tomp Start simpler -- you do an experiment to measure the time something takes to travel a distance. You measure the distance, then set the experiment in motion and stop the time. This captures the velocity (distance-per-time). Per experiment design, it doesn't also tell you how the thing tastes that traveled, or whether it had a nice day, or wanted to move at all. That's all not available info in this setup.

What you see is all there is -- but by virtue of seeing, you don't capture what's audible.

Does that make more sense than the dense quote?

OK, that makes sense in light of what Drucker seems to be interested in. There would be a close connection, perhaps, with the concept of qualia, as well. So you are saying that any given account (or experiment) touches only a subset of possible aspects that could have been measured. That subset Drucker wants to give the awkward name of "capta".

ZettelDistraction

@tomp said:

...

That subset Drucker wants to give the awkward name of "capta".

My portmanteau word for such individuals (inspired by an acquaintance) is "idiosyncretin."

This isn't to deny the need for methodological questions — whether "Raw Data is an Oxymoron" holds, and other matters of interpretation — which I leave to the experts.

tomp

@ZettelDistraction said:

My portmanteau word for such individuals (inspired by an acquaintance) is "idiosyncretin."

I like it.

Andy

@ZettelDistraction said:

Is there a suggestion that capta, so called, is suspect? Suppose I limit the bandwidth of my receiver to reduce interference from neighboring signals outside some range of interest. I orient a directional antennal to null out noise generated by nearby switching power supplies. Now I am ideologically suspect because I have excluded signals and noise that deserved equal time with the signal I am interested in, and worse, I decoded a transatlantic signal below the noise level.

Yes, I think it's clear that for Drucker the point of using the term capta is, as she wrote, "to return the humanistic tenets of constructedness and interpretation to the fore", which involves critically questioning the methods used in producing the capta, because for her, too often "data pass themselves off as mere descriptions of a priori conditions". One doesn't need the term capta to emphasize this critical questioning (nor is it the exclusive specialty of "humanistic" scholars), but that's largely why she (and others such as Kitchin) use the term (even if others such as Kitchin quickly revert to speaking of data for practical reasons).

I'm reminded of Ian Hacking's taxonomy of elements of laboratory experiment, in "The self-vindication of the laboratory sciences" (in Science as Practice and Culture, edited by Andrew Pickering, Chicago, 1992). Hacking said that the purpose of the taxonomy was to draw attention to the variety of elements of experimental practice that could be critically questioned.

Here is Hacking's taxonomy, from my notes, which he grouped into three areas:

1. Ideas

• (1) Questions
• (2) Background knowledge "and expectations that are not systematized and which play little part in writing up an experiment, in part because they are taken for granted"
• (3) Systematic theory "of a general and typically high level sort about the subject matter, which by itself may have no experimental consequences"
• (4) Topical hypotheses: "what connects systematic theory to phenomena... sets of approximating and modeling procedures"
• (5) Modeling of the apparatus: "theory that enables us to design instruments and to calculate how they behave"

2. Things

• (6) Target: "a substance or population to be studied. The preparation of the target—in old-fashioned microbiology by staining, use of microtomes, etc.—is best kept separate from the modification of the target, say by injecting a prepared cell with a foreign substance"
• (7) Source of modification: "usually apparatus that in some way alters or interferes with the target"
• (8) Detectors "determine or measure the result of the interference or modification of the target... we include both detectors and sources of modification as apparatus"
• (9) Tools: "more humble things upon which the experimenter must rely"
• (10) Data generators: "People or teams who count may be data generators. In more sophisticated experiments, there are micrographs, automatic printouts, and the like. There is no need to insist on a sharp distinction in all cases between detector and data-generating device."

3. Marks and the manipulation of marks

• (11) Data: "what a data generator produces. By data I mean uninterpreted inscriptions, graphs recording variation over time, photographs, tables, displays. These are covered by the first sense of my portmanteau word 'mark'. Some will pleonastically call such marks 'raw data'. Others will protest that all data are of their nature interpreted: to think that there are uninterpreted data, they will urge, is to indulge in 'the myth of the given'. I agree that in the laboratory nothing is just given. Measurements are taken, not given. Data are made, but as a good first approximation, the making and taking come before interpreting. It is true that we reject or discard putative data because they do not fit an interpretation, but that does not prove that all data are interpreted. For the fact that we discard what does not fit does not distinguish data from the other elements (1)–(14): in the process of adjustment we can sacrifice anything from a microtome to a cyclotron, not to mention the familiar Duhemian choice among the hypotheses in the spectrum (1)–(5) for the ones to be revised in the light of recalcitrant experimental results."
• (12) Data assessment: "one of at least three distinct types of data processing. It may include a calculation of the probable error or more statistically sophisticated versions of this. Such procedures are supposed to be theory neutral, but in complex weighing of evidence they are sensibly applied only by people who understand a good many details of the experiment—a point always emphasized by the greatest of statistical innovators, R. A. Fisher, although too often ignored by those who use his techniques."
• (13) Data reduction: "large or vast amounts of unintelligible numerical data may be transformed by supposedly theory-neutral statistical or computational techniques into manageable quantities or displays. Fisher used the word 'statistic' to mean simply a number that encapsulated a large body of data and (independently of Shannon) developed a measure of the information lost by data reduction, thus determining the most efficient (least destructive) types of reduction."
• (14) Data analysis: "This may seem like a kind of data reduction, but the programs for analyzing the data are not supposedly theory-neutral statistical techniques. They are chosen in the light of the questions or focus of the experiment (1) and of both topical hypotheses (4) and modeling of the apparatus (5). In this case, and to a lesser extent in the case of (11) and even (12), there is now commonly an echelon of workers or devices between the data and the principal investigators; Galison argues that this is one of the ways in which experimental science has recently been transformed. There are many other new kinds of data processing, such as the enhancement of images in both astronomy and microscopy. And (11)–(14) may get rolled into one for less than $2,000. 'With the new $1,995 EC910 Densitometer, you can scan, integrate, and display electrophoresis results in your lab PC. Immediately! No cutting, no hand measuring. Programs accept intact gel slabs, columns, cellulose acetate, chromatography strips and other support media' (Software extra, $995; from a typical 1989 ad on a back cover of Science)."
• (15) Interpretation of the data: "demands theory at least at the level of background knowledge (2), and often at every other level, including systematic theory (3), topical theory (4), and apparatus modeling (5). Pulsars provide an easy example of data interpretation requiring theory: once a theory of pulsars was in place, it was possible to go back over the data of radio astronomers and find ample evidence of pulsars that could not have been interpreted as such until there was theory. The possibility of such interpretation also mandated new data reduction (12) and analysis (13), and the systematic error part of the data assessment (11) had to be reassessed."

Hacking commented, in part:

We create apparatus that generates data that confirm theories; we judge apparatus by its ability to produce data that fit. There is little new in this seeming circularity except taking the material world into account. The most succinct statement of the idea, for purely intellectual operations, is Nelson Goodman's summary ([1954] 1983, 64) of how we "justify" both deduction and induction: "A rule is amended if it yields an inference we are unwilling to accept; an inference is rejected if it violates a rule that we are unwilling to amend." There is also more than a whiff of Hanson's (1965) maxim that all observation is theory loaded, and of the corresponding positivist doctrine that all theory is observation loaded. The truth is that there is a play between theory and observation, but that is a miserly quarter-truth. There is a play between many things: data, theory, experiment, phenomenology, equipment, data processing.

ZettelDistraction

Drucker's intervention is more suited to the digital humanities than to international relations, where maps with fuzzy boundaries and complexified narratives can supply ammunition to bad actors.^†

^†Pun intended