Chapter 1. What the Number May Not Become

The fifteen Dionyas chart questions settle one thing in the design of their asking: a chart number may be asked about only together with a statement of what it may not be turned into. Each question carries a limit of its own, and each limit names a particular conversion, from rank to favour, from store to nationality, from absence on a list to absence of listeners, from a company to a cause. The questions about response go further and keep holding back open beside acting. The chapter that follows shows how, and adds to the two debates of the Introduction: the study of commensuration and quantification, which asks what a common metric joins and what it must not, and the study of public measures, which asks what a number does once people act on it.

1.1Data and method

The data are Dionyas's public list of inquiries, fetched with plain GET requests and kept as snapshots in the study's folder. The list holds 1,539 inquiries across 32 domains, of five kinds: 759 cases, 500 historical questions, 256 general designs, 18 chart questions and 6 signal questions. It describes its own type as editorial question design and gives its count of generated judgments as 0. A record has six fields that this study reads: a title, a question, a scope that says what the inquiry covers, a list of the material it needs, a boundary that says what it may not claim, and a state.

The selection rule is one sentence: every chart question in the list that is not linked to the foresight domain. All 18 chart questions sit in the evidence lane, and each has a leading domain and one other. Six leading domains hold 3 each. The rule leaves out the 3 whose second domain is foresight, which ask about momentum and timing, and keeps 15: three each whose leading domain is countries and languages, charts and measurement, reputation and trust, catalogue and longevity, and labels and organisation. The rule was chosen so that the paper would read how the studio frames a chart observation at a single reading, not how it frames change over time, and the 3 were left out by the rule, not by reading them first. None of the fifteen names an artist, a work or a year; the scope speaks only of "the chosen artist" and the markets observed.

The appendix gives every record in full. The records are in Korean; every quoted sentence below is from one English translation made for this study, sentence by sentence, with nothing added. Dionyas does not say who drafted these fifteen, and no fact about charts, stores, platforms or markets is drawn from them. Where a record says what a store chart is or is not, that is the record's framing, which this study has not checked.

Two kinds of property have to be set aside before reading. The first is the studio's form: every inquiry has a scope, needs and a boundary, and every one of the fifteen is phrased as a question. None of this is a finding. The second is frame, wording copied from record to record. Frame is measured here by five-word runs, counting a run as frame when it recurs in at least 5% of a site's records and in at least 3. Across the whole list the titles and the questions hold no frame at all, while the scopes are 46% frame, the needs 30% and the boundaries 41%. The reason is the domain table. In 256 of the 1,539 inquiries the boundary is the leading domain's caution line word for word, and 884 open with it; the whole list holds only 222 distinct boundaries.

The fifteen chart questions reverse that pattern. Their scope and their state are the same in all fifteen, and in the whole list they appear only in the 18 chart questions:

Inside the selection the scope is therefore 100% frame. The boundaries, by contrast, are all different: none is its domain's caution line, none appears in any other inquiry of the list, and at the site's threshold they hold 0% frame, as do the needs. In each record the question opens with the title, word for word, and then repeats the need as the evidence to review. So this study reads the scope and the state as frame and does not treat them as evidence, and reads as evidence what is each record's own: its question, the material it names and its limit. The fifteen name 49 items of material between them, from 2 to 5 in a record.

The questions take two shapes. Eight can be answered yes or no, or by choosing between two readings; seven ask what, which, how, to whom or why. Of the limits, 11 are flat negations, statements that something cannot be concluded, does not follow or is not the case, and 3 state only that something may differ or be missed. Six name favour, regard, opinion, loyalty or liking as what the number does not show. Every number in this paper comes from a script that reads only the snapshots, by simple word tests on the Korean original; the counts describe the records and nothing else.

1.2What a position may not be made to mean

The first finding is that the questions on markets, measurement and reputation each pair a chart reading with a conversion they refuse. The pair is the unit of design: the question asks what a position shows, the sentence beside it says what the position may not become. The plainest instance is the question about where an artist's presence gathers.

The question is ordinary market reading: where does the artist show up most. The limit names the slide that would follow, from a pattern across national charts to a pattern of public taste, and gives two reasons it fails. The users of a platform are not the public, and a high position in a crowded chart is not the same feat as one in a thin chart. Espeland and Stevens, Commensuration as a Social Process (1998), define commensuration as the comparison of different entities according to a common metric, and they discuss the stakes of calling something incommensurable. This study reads the limit as an incommensurability claim of a precise kind. It does not say that charts across countries cannot be compared. It says that the comparison, once made, is a comparison of chart positions and not of affection, and it names what differs underneath.

The same move recurs in the question on home and abroad, and here the refused conversion is about who people are.

A store chart is filed under a country, and the filing invites a reading in which the country is a people. The record separates three things the filing joins: the country a store serves, the country an artist comes from and the country a listener lives in. The question it sits beside asks how the structure of observation differs at home and abroad, and the material it names includes the countries of the artist's activity and local context. On this study's reading, the question asks about observation and the limit refuses to let observation stand for identity.

The question on the unseen market asks the most direct version of the problem, and asks it as a choice between two readings: is a market missing from the charts one not yet entered, or a blind spot of observation? Its limit reads:

A chart that lists a fixed number of places sees nothing below its cut. The record puts the cut in plain view and refuses the conversion of not listed into not there. Its question leaves both readings open: the market may indeed not have been entered, or the chart may simply not see it. The material it names, other platforms, local charts and the range of the supplier, is material for telling the two apart. boyd and Crawford, Critical Questions for Big Data (2012), argue that the assumptions and biases of large data sets must be interrogated critically. This study reads the unseen market as one such interrogation, set down before any data are in hand: the question asks first what the instrument can see.

Three questions from the measurement domain turn that interrogation on the chart itself. The first asks it directly, wanting to know what can and cannot be said with the material now, and sets beside it this limit:

This is the only limit of the fifteen in two sentences. The first refuses to treat a store chart as a poll. The second does not refuse anything; it states what stays with the data whatever is done to them. The material the question names is the original, the time of collection, the range of countries and the method of identification, the information a reader would need to know what kind of number it is. The word now in the question matters. It ties what can be said to the material in hand, and leaves room for more to be said later.

The second measurement question asks whether the range observed distorts the comparison, and its limit is the one sentence of the fifteen that names a summary figure.

The record does not say the average is wrong. It says the average has no fixed meaning apart from the set of countries that happened to be observed, so that two averages from different moments may not be the same kind of thing. The material it asks for, availability by market, denominators and update times, is what a reader would need to know which set was behind a given figure. Mennicken and Espeland, What's New with Numbers? (2019), stress tracking quantification across sites to appreciate its essential ambiguity. This study reads the observed range as locating one such ambiguity inside a single figure: the meaning of the average moves with the coverage, while the label stays the same.

The third measurement question is the most arithmetic, and its limit the most exact.

The question asks whether the parts of a ratio are present before the ratio is taken. The material names them: people reached, counted once each, over the people there are, or over the people who use the platform. The limit then refuses a shortcut that a reader in a hurry might take, dividing a rank, which is a place in a list and not a count, by a population. It names two things the result would falsely claim to be, a rate of penetration and a measure of national favour. Porter, Trust in Numbers (1995), asks how to account for the prestige and power of quantitative methods. This study reads the limit as aware of that prestige: a number divided by a population looks like a rate, and the record writes down, in advance, that it is not one.

The questions from the reputation domain carry the same design into the space between attention and regard. The first asks whether chart performance and the public's regard point the same way, and its limit says what the data lack.

This limit differs from the others. It does not refuse a conversion by argument; it states an absence. The record says that the chart data hold neither the reasons for liking nor the content of criticism. The material the question names, news, unprompted responses and multilingual context, is material that carries reasons. The second reputation question makes the refused conversion explicit and turns the question on the reader.

The question does not ask whether attention and favour have risen. It asks whether someone is telling them apart, which is a question about the analysis, not about the artist. Four of the fifteen questions take this form. They ask whether a comparison is being distorted, whether two things are being told apart, or to whom a response is being attributed, and so make the practice of reading the chart the thing under examination.

The third reputation question goes furthest from the number. It asks why the same event is read differently in each market, names original texts in the local language, cultural context and comparative reading as the material, and sets this limit:

Here the refused conversion is not of a chart rank at all but of words into a score. The limit is one of the 3 that state a possibility rather than a negation: a score can miss, not must. On this study's reading, it still marks the edge of commensuration as sharply as the others, at the point where meaning would be reduced to one dimension in order to be compared across markets. Espeland and Stevens, A Sociology of Quantification (2008), call for an ethics of numbers. The nine questions read in this section do not set out an ethics, but each writes down, beside a measure, one use of it that the record will not accept.

1.3Whose response, and what to do about it

The second finding is that the questions on catalogue and labels move from what a number shows to whom it belongs and what should follow from it, and that they keep the measure from deciding either. The first three ask about the music itself. The question on a few songs or many asks whether a response depends on a few songs or several together, and limits what the second answer would prove.

A broad presence on a chart could be read as a sign of a lasting bond with listeners. The limit refuses that reading and offers a plainer one, a release that briefly lifts everything around it. The question on new release and catalogue asks whether the two can be told apart at all, and its limit names a small fact about the data that could undo the comparison.

The date in a data feed is the date the supplier gives, which may not be when the song first appeared. A comparison of new and old that trusted the feed's dates could put an old song among the new. The material the question names, release dates by song, reissues, versions and charts, is material for catching that error.

The question on credit for a collaboration makes the problem of attribution its subject.

The limit is about a data structure. The record speaks of a single lead-artist ID, so on this study's reading a song made by several people would reach the data under one person's name. The question asks, again in the form that turns on the reader, to whom the response is being attributed. Lamont, Toward a Comparative Sociology of Valuation and Evaluation (2012), names categorisation among the subprocesses of valuation. This study reads the collaboration question as pointing to a categorisation made before any valuation begins: the record format decides whose name the response is filed under, and the limit says that filing cannot hold everyone's contribution.

The question on label and artist carries attribution from the song to the company.

Belonging to the same company could tempt a reader to explain a response by it, and the limit refuses to let it serve as an explanation. The material the question names, contracts, official credits, periods of validity and affiliations, is material for saying exactly what a company's role was and when. The limit does not say a label has no part in a response. It says that membership alone settles nothing about cause or contribution.

The last two questions are the only ones of the fifteen that ask what to do. The first asks which market response to consider first, and names as the material to review the observed change, the cost of a response and the goals of the person responsible. Its limit reads:

The question accepts that a chart may call for a response. The material it names puts the chart beside two other things, the cost of responding and the goals of whoever is responsible, so that the observed change is one input among three. The limit then names two responses, carrying out promotion and changing creative direction, and refuses to let the observed amount alone decide either. Espeland and Sauder, Rankings and Reactivity (2007), describe how public measures change expectations and permeate institutions, and they name self-fulfilling prophecy and commensuration as mechanisms that induce reactivity. This study reads the limit as written against that path: it names the point at which a chart would begin to shape the work it measures, and says the chart alone is not enough to take that step.

The second response question widens the choice.

Three options are named, and two of them are not actions in the market: further inquiry and holding back. The question does not ask which to choose. It asks which to keep open, which treats the decision as something that can stay unmade while more is learned. The material it names includes reversibility and evidence to the contrary, so that a response is weighed by whether it could be undone and by what would count against it. The limit is the broadest of the fifteen. It refuses a conversion from rank to two things a chart is often taken to forecast, the direction of the music and the artist's future. Strathern, 'Improving ratings': audit in the British University system (1997), holds that audit has a life of its own that jeopardises the life it audits. This study reads the options question as an attempt, in the design of a question, to keep a measure from acquiring that life: the chart is allowed to prompt a question, not to set a course.

This study does not say which response, if any, a reader of a chart should choose. Nothing in the fifteen records says so either. What the reading can say is that the questions about response were designed to keep the choice in human hands, with holding back named as a choice, and that they set the chart beside cost, goals, accountability and reversibility rather than above them.