From Ace of Spades blog site’s overnight thread for December 3, there were two links describing the dissatisfying impressions of higher frame rates for motion pictures.
One article describes the unpleasing effects of 48FPS viewing of The Hobbit compared with an earlier viewing in the standard 24FPS. In contrast, the other article describes similarly unpleasant effects of frame interpolation in TV sets to remove blurring of fast movements. Both describe the same complaint even though the first involves actual images captured at the higher rate while the second involves computer generated images to insert between actual images. In both cases, the images are presented faster to better match the responsiveness to the human eye and this makes the scenes appear more real. The problem is that the exposed reality is of the staged sets, lighting that only occurs on stages, and heavy makeup on actors. The second article describes this as a “soap opera affect” in the sense that the faster frame rate make the scenes look like cheap productions. Another analogy to a soap opera may be of an amateur video of a live play. The acting and set design may be great, but there will be no mistaking the reality that this is a staged performance.
Apparently the slower frame rate provides a cue to the mind that the motion picture is an illusion. The brain recognizes the the images are not real and this gives it permission to be drawn into the fantasy being presented. The slower frame rate provides a confirmation that this is a story, or as the first article complains about the faster rate:
But that is not what the filmish people want. They like the less sensitive, blurry style of film better. One critic even suggested that directors should use soft-focus filters to debase the clarity of the new digital recordings and restore the “painterly” aspect of classic films.
These articles add something to last post about modern non-fiction is similar to what we once called fiction. In that article, I described older forms of fiction literature as dramatization of characterizations and action within a setting that was expected to be accurate or realistic. From the beginning, the reader is alert to the idea that the book is fiction and this gives permission to be drawn into the imagined reality. Despite the intent to tell a fictional story, the audience demanded accuracy in details so that the work required extensive research to get the particulars of the setting right.
Today, when a book involves research, the author will more often rearrange the material to sell the work as non-fiction. Instead of starting a book by introducing the characters (thus making it a work of fiction), we start the book by introducing the factual evidence. Eventually we’ll introduce the fake illustration but we have set up the reader’s mind to think of the book as a work of non-fiction instead of fiction. This introduction of facts primes the brain to look at the work as factual material.
This choice of arranging content (facts first or fake first) has a profound affect on the reader’s approach to the material even if the basic content of the material (a mix of facts and fakes) is the same.
I think the two above articles about motion pictures illustrate this perception difference. In those examples, the viewer recognizes the slower frame rate as an introduction to a work of fiction or at least conducive to interpreting the scenes as fiction. The slower frame rate allows viewers to allow themselves to be drawn into the story because they understand it is a story. When the frame rate increases, the mind begins to take the images more seriously as actual facts. The brain interprets the same work as non-fiction. This causes a change in perception about the same material. The more realistic frame rate encourages a more realistic interpretation that exposes the fact that this motion picture is a documentary of staged production. The faster frame rate draws the mind to imagine being on the filming set and among the filming crew instead of drawing us into the imaginary world the crew is trying to create. Or as stated in the first article:
Imagine you had the lucky privilege to be invited by Peter Jackson onto the set of the Hobbit. You were standing right off to the side while they filmed Bilbo Baggins in his cute hobbit home. Standing there on the set you would notice the incredibly harsh lighting pouring down on Bilbo’s figure. It would be obviously fake.
A point worth repeating is that the greater realism makes the scene more obviously fake.
In my last post I described a made-up example of a non-fiction book on web-site design. We know we are reading a book providing instructions and advice on building our own web-sites. When the book introduces an illustrative example, we immediately recognize the example as obviously fake. At the very least, the example has to be over-simplified to describe the key features as briefly as possible.
As I described in that post, that instruction book can be turned into a work of fiction by rearranging the text to first introduce the fake company with its need to produce a web-site. This would draw us into the story of that fictional company even though the technical details remain accurate and represent best practices. I feel this is what earlier generations of readers expected from their fiction. They expected their fiction to entertain them, but they also expected the work to either educate them or at least not insult their prior education. The introduction as a work of fiction allowed them to smooth over the accurate details to accept the fictional tale of characters and action.
The above articles about higher frame rates in videos is that many viewers dislike the effect it has on fictional stories. To appreciate the fiction the audience has to work harder to overcome the mind’s recognition this is all staged with actors on a set and a full crew just off the edge of the frame.
I think something similar is happening in modern non-fiction literature, perhaps in the opposite direction. Unlike older practice of non-fiction that attempted to strip out anything that was not backed up with evidence, the newer non-fiction includes much fake material to illustrate key concepts in order to make the work more entertaining, or more readable. Writers of modern non-fiction expect their readers will recognize that the illustrations are not part of the factual information in the work. This makes sense in my examples of an educational text book (for example on physics) where the illustrations are obviously simplified or unrealistic exercises for the student to solve. However, more creative works risk having readers take the illustrations seriously and consider them as part of the factual information in the book.
I am reminded of an physics example of balancing a broom stick vertically on a platform that can only move sideways. I spent time imagining where in the process of the manufacture or use of broom sticks would such a skill be needed. Initially I did imagine that this a real practical problem. I tried to imagine a practical application involving balancing a broomstick on a moving platform. I experimented with the idea of balancing a standing pencil on the palm of my hand but even then couldn’t figure out when this would be practical even as entertainment (I never mastered even that trick). The example was meant to be fictional to provide a simplified illustration of more complex problems such as maintaining stable flight of a long slender rocket where the only opportunity to apply force is at the nozzle at the bottom end. A rocket example would not make a good example of the text because rocketry involves many complications of aerodynamics, thrust, and burning of fuel (reducing the rocket’s mass). The broomstick is a fake illustration of just one of these practical problems that occur with rocketry, for example.
My point is that I took the illustration in a non-fiction text book as an accurate depiction of a real application. The result was contrary to what the author intended. The illustration was not meant to be one of the facts.
In modern non-fiction, we take more liberties with illustrations than we see in text books. In text books, an illustration may be separated in a box or in a section clearly marked as an example or an exercise. In popular non-fiction, the illustrations flow with the text. In my made-up example of the book on web-site design, the text describing a particular principle may reference the fictional website within a paragraph describing some principle. The mere mention of the fictional site is meant to alert us to the fact that it is an illustration that is itself not real. As I mentioned in my broom-stick example, some readers may be confused into thinking that the illustration is one of the facts. The mix of fake illustration in a non-fiction book for the web-site book can risk confusing readers with wrong lesson that the principles only apply to just this fake web-site or web-sites that are similar to the fake one. The reader may not learn the intended lessons at all.
Journalism is an example of non-fiction that ideally strives to present verified information but presents a readable story by adding opinions or fictions. Recently there was an investigative journalism article published in Rolling Stone that described one student’s experience concerning a brutal gang rape at a Fraternity at University of Virginia. Following the publication, there have been several articles (such as this one) that criticize the article on journalistic principles of failing to check other sources and in particular to attempt to interview the accused and people who knew the accused.
In response to the criticism, the magazine issued a statement (discussed here) that attempts to clarify that the intent of the story was to present one person’s account of an incident, and that this person was found to be credible. The story was about one person’s account and that account happened to accuse other people of wrongful acts. The description of those wrongful acts illustrate the one person’s account. This illustration (detailing of a brutal gang rape by a group belonging to specific fraternity on a specific campus) was not fact checked because the purpose of the non-fiction was only to document the victim’s account of the events. In context of my previous post, non-fiction (the victim’s account of some events) can include illustrations (providing extensive details) that do not need to be factually correct because we should recognize that they might be fake. The underlying story is like an auto-biographical account of the victim’s experience.
From my perspective as a data scientist, I have conflicted opinions about this story. The story presents a case that some very awful events occurred where the perpetrators were not punished. I am disappointed the the journalist did not make the effort to attempt to get information from the accused parties or to verify key details about the actual description. I share the disappointment of those who criticize the journalist for not doing a more thorough job checking the facts before publishing the article.
On the other hand, I may not be as outraged as the critics because I see the account, as flawed as it is, as a data point that we should welcome into our data sets. In an earlier post, I described a future of journalism that sells its work to data repositories (big data stores) instead of selling narratives directly to readers. In a separate post, I described a valuable contribution to collect extensive first-person accounts of actual experiences. In that post, I described how first-person account journalism of a wide variety of individuals in areas where Ebola is epidemic can provide us data that can help us understand how Ebola spreads and better plan to control the spread. In that case, i would like to have hundreds if not millions of first person accounts of their routine lives before, during, and after a outbreak in their community. To obtain that quantity of first person accounts to populate a data store, we would not demand that each first person account be fact checked in terms of referenced persons, events, and places. What we need is simply an accurate recording of the first person’s account of their lives.
The above Rolling Stones article seems to be a good example of a first person-account that I said we should welcome in the data store. Such first-person accounts will inevitably involve others as that person describes his experience, that involvement will typically involve some kind of accusation or unfavorable characterization. However, the larger project is to gain a better understanding of the breadth of experiences of sexual violence on campus. This one story would be one of tens of thousands (if the rate of sexual violence is accurate). It would be great if we could get that number of accounts with similar depth as obtained in this one article. The total of these accounts will allow us to understand the problem of sexual violence much better so that we can better manage this problem.
This concept of collecting many first person accounts does not require the external fact checking that critics of this article are demanding. My argument is that these first-person accounts would not be published immediately (as Rolling Stones did) but instead the accounts would populate a large data store that would eventually accumulate many other first person accounts. If we had a market for collecting extensive first person accounts of events, eventually the data store would include sufficient conflicting or collaborating accounts to provide the fact checking that was previously missing from the individual accounts. This approach would be superior to the single-article fact checking because each first person account would present a complete story on its own providing context of background and other influences on their perspective of the event.
Eventually, analytics of the entire data set will support a conclusion that can be made into a story that can be published, but it would describe the broad sociological problem of all types of sexual violence on campus. It is my impression that this is the intent of the Rolling Stone article, to illustrate this broad sociological problem with a single case study. My approach of extensive first-person stories will more accurately depict the nature of this broad problem.
In my earlier posts about employing journalists as miners of fresh data of new first person accounts, I mentioned we do not have an economy that supports this enterprise. We have no mechanism to compensate journalists to collect first-person accounts into data stores. In the interim, they must sell their accounts through published articles where the articles need to have a narrative to attract a large number of paying readers. Meanwhile, this publication of the first person account does eventually find itself in big-data stores through those data stores using natural-language processing algorithms that read the published stories and the subsequent social-media responses. The data stores are getting populated with these accounts but through a very inefficient mechanism of using the old means of compensating journalists for published single-topic articles constructed for general readership.
Even without fact checking of the other points of view, the above article offers something of value in terms of providing a detailed description of one person’s account of an experience of brutal rape. This account could be filed in data stores with similar first person accounts. In this sense, I accept Rolling Stone’s defense that “we found Jackie to be entirely credible and courageous and we are proud to have given her disturbing story the attention it deserves”. They are correct that this account deserves attention but in the sense of becoming part of the big data store that will include everyone else’s story.
The complaints about the article are that they there was insufficient due diligence to obtain the opinions of the accused parties or to note that their attempts to contact those parties were unsuccessful. I agree that the profession of journalism should follow standards for getting the other side of the story. I agree that Rolling Stone and the author should be held accountable to those standards. However, I agree because the article is sold as a final published account for general readership who will have access to only what is published. This is the old journalism model of publishing individual accounts for paying readers.
In past decades, a print article would be read only when a person finds the paper copy somewhere. And once found, it will be difficult to find related articles. In that context, the article needs to meet a high standard because the editor is aware that this may be the only account of this episode that the reader may encounter. As a result, they needed a standard to rigorously obtain the other party’s perspective. We still expect the same standards today but we have better technologies that allow us to quickly find related opinions.
As we move toward a data-centered economy where we solicit individual stories to populate data stores instead of published articles, these journalistic standards for publication may no longer be appropriate. For data stores, we are interested in many one-sided accounts of reasonably credible sources.
We are in a transitional period for journalism. The older period demanded an article that presented all view points with equally attention to fact checking. The newer period will demand collections of one-sided first-person accounts to include in data stores where the competing first person accounts for the same incident will be available for analytics to prepare future story of what really happened. In the new period of big data, we should welcome such one-sided first person accounts even though the account includes accusations of others. The big data concept of journalism welcomes independent contributions of the competing first person accounts as we are currently experiencing in this case where multiple articles are written in response to the Rolling Stone article. For example, the Rolling Stone article opens with a quote that they claim comes from a well-known University song but now we see many objections that the stanza occurs in no version of the song anyone can remember. From a journalist perspective, this is a controversy. But from a data perspective, this is just data. One first-person account recalls hearing those words in the song, and many other first person accounts do not recall those words. This is all good data in that it accurately reflects the individual first person accounts.
The goal of big data is that having all of this data available in a single data store (currently inefficiently scattered across the Internet) allows us to individually cross-check the stories and come to our own conclusions. Due to printing cost limitations, the older form of journalism would have done this cross checking and then reject any unverifiable accounts. This rejection is a net loss of the fact that the accounts were initially offered.
The newer form of journalism is to provide us a detailed first person account where the focus is on faithfully capturing that one person’s recollection of some event. The fact-checking will come later when we compare and combine the various competing accounts. The Rolling Stone’s article is a valued contribution in the context of populating a data store. If populating a data store were the project, we would never have erred into jumping to conclusions based on an assumption that this account represented reality. In this data store, this contribution is just one first-person account from a victim that was deemed to be credible. We need more accounts before we make a conclusion.
I started this post discussing how a new technology (faster frame rates) changes how we view a media (motion picture). The faster frame rates do a better job of capturing reality but that reality is that the motion picture is a staged performance involving harsh lighting, heavy make-up, and actors instead of the imagined reality that the film-makers intended to present. From a data science perspective, the faster frame rate demonstrates the value of more bright data in telling us what is really going on in the real world. What is really going on in the real world is that actors are on stage facing cameras.
I am stretching to make an analogy of the faster frame rate with a first-person account in journalism. Like the inserted frames of fresh observations, the first person account provides new information that is the bright data of a testimony from a credible victim. Like in the motion picture, this additional information jars us into looking at the media differently. Instead of seeing the motion picture as a fiction, we see it as a documentary of what happened on some stage. As watchers, we are confronted with the new burden of smoothing out this detail to recreate the fictional experience.
Likewise, in journalism of first-person accounts, we are seeing those first-person accounts for what they are and this places new burdens on the reader to seek out the other information to complete the investigation.
I argue that this is where we are heading as a society. We are heading toward a data-driven culture where decisions are based on the data. In a data driven culture, these decisions postpone interpretation of the data until the time a decision is needed. At that time, we will analyze the available data. Until that time, the raw data (such as more first person accounts) will continue to accumulate.
We are in a transitional period where we are using both old and new approaches and our population includes some who are thinking of the new approach and others who are thinking of the old approach. One possible explanation of the modern controversies (such as the one described above) is this transitional period involves two different approaches with completely different expectations. Like the movie lovers of old films objecting to the hyper-realism of faster frame rates, consumers used to old journalism are objecting to hyper-realistic recording of a first person account without obtaining a response of the accused parties. Eventually, we will become accustomed to the new data centered reality of hyper-realism. We will learn how to appreciate the superior value it will provide.
Also in the introduction, I described the alternative approach of computer-generating the intermediate frames. I noted that it is curious that the computer-generated extra information resulted in the same sense of hyper-realism. However, I objected that this hyper-realism misleads us into believing this is what the film-crew experienced on stage. The interpolated data is model-generated dark data instead of fresh observations of bright data that would have been available to the film crew as the time of the recording.
The analogy of dark-data hyper-realism in journalism is that injection of the journalist’s personal opinion into an otherwise non-fiction account of what happened. Opinions should be recognized as illustrations of the fact and those illustrations are fake because the journalist’s opinions are not facts. An opinion is a form of fiction, it is telling us what the author thinks instead of what the author knows. We accept fictional illustrations in works of non-fiction but we should be able to recognize them as fictitious illustrations. An opinion is such an illustration.
The Rolling Stone article purports to present the first person accounts of credible sources. However to make the story readable, the author injects the author’s opinion about the University’s reasonable response to a complaint about sexual violence. The University responded to the initial complaint by outlining three reasonable options for the victim to pursue but left it to the victim to take positive action to pursue at least one of these options. The victim did not pursue either one. Those are fair first-person accounts of what happened. However the article continues with the author’s opinion quoted from Neo-Neocon’s article:
Like many schools, UVA has taken to emphasizing that in matters of sexual assault, it caters to victim choice. “If students feel that we are forcing them into a criminal or disciplinary process that they don’t want to be part of, frankly, we’d be concerned that we would get fewer reports,” says associate VP for student affairs Susan Davis. Which in theory makes sense: Being forced into an unwanted choice is a sensitive point for the victims. But in practice, that utter lack of guidance can be counterproductive to a 19-year-old so traumatized as Jackie was that she was contemplating suicide. Setting aside for a moment the absurdity of a school offering to handle the investigation and adjudication of a felony sex crime – something Title IX requires, but which no university on Earth is equipped to do – the sheer menu of choices, paired with the reassurance that any choice is the right one, often has the end result of coddling the victim into doing nothing.
The latter part is manufactured information by the author: the author’s opinion. It fills a role of an illustration in work of non-fiction in that it reinforces a certain lesson to learn from the facts, but the illustration itself is not a fact about the case. It is the author’s opinion that the best practices followed by most universities was at fault by failing to motivate the victim to select an effective course of action. The flaw is in expecting that victims of violence can not be permitted to make their own choices for how to proceed with an accusation. This piece of information is out of place in an article the claims to accurate present accounting of credible sources.
As in the computer interpolated images to simulate a faster frame rate, the reader sees this manufactured information as part of the same story. In movies the faster frame rate gives the impression of a cheaper production more frequently associated with daily soap operas. In journalism, the injection of author’s opinions leaves the reader with the impression of reading a cheap novel. In Neo-NeoCon words: “When I read Erdely’s piece, it seemed to me that its style resembled a romance novel gone bad”.
I think journalism is at a critical point right now where it needs to change how it sells the data it collects. The core information of the Rolling stone article is a detailed accounting of a credible witness and of the University’s response. This would be valuable data to include in a big data store that welcomes all of the other competing accounts. Unfortunately, such a data collection process has no means to compensate the investigative journalist. For now, the journalist needs to construct a readable story to present this valuable case of a single point view with the journalist’s opinions. This allows the journalist to earn money for the work and it also makes the data available for consideration in big-data projects (that current collect such published stories).
The single point-of-view account is valuable but only when we have access to similarly detailed competing accounts. We need this information, but the old method of commerce is not only inefficient but damaging to reputations. These valuable single point-of-view accounts by definition will not include the accounts of the others involved in the scenario. Those accounts will come later and through independent reporting. We need to find a different way to employ journalists to collect first person accounts without having to present them as final articles for publications. A new data-centered economy would pay journalists to collect the first person accounts into data stores that only later produces a marketable product of an analysis of all points of view.
Addendum: After writing the first draft of the above, I learn that there is more news coming out casting doubt about the original story such as a new statement by Rolling Stone and a report from Washington Post. This developing story demonstrates old journalism in action. This is reassuring, especially in context of making sure that appropriate justice is achieved with respect to guilty parties.
I do not think these new developments change the points I am making above. I still think the original Rolling Stone report provides a valuable data point of a first person account. This would be valid bright data to put in the data store because it reflects one person’s account of some experience. As we are currently witnessing across the Internet, the data store would continue to grow to include competing accounts and even later retractions or revisions. Having this data available allows us to perform larger scale sociological studies such as the nature and prevalence of sexual violence on campus. Even if the story turns out to be a complete fabrication, we will obtain bright data that can inform us that such misleading accounts can occur and these accounts can be convincing when first told. This data will bring us closer to understanding the broader problem.