Recently I've been attending grief counseling on Thursdays, a process I like to call Lossy Compression. It definitely makes for a more disjointed newsletter production schedule. I'm working through it.

But hey! My new course More than (key)Words is officially available whenever you want it. If you're keen to understand how to research and prepare your content for our AI future, I highly recommend it. It's relatively short but densely packed with both theory and practice. And although the course content is all the way there, the whole thing is discounted until I get the learning ecosystem fully configured.

And now, additional thoughts on the Content Production phase, in which we go from idea to execution.

–DC

Creative production in the age of the efficient mechanical approximation of workflow reproduction: A lament of the Oregon Trail generation

For David Patch, who was a wonderful teacher

I. The failure of mimickry

In the first acting class of my freshman year of high school, my teacher Mr. Patch instructed, "Acting is actually doing something." What he meant was, if you are in character playing an accountant on stage, you should actually be writing numbers in a ledger. Despite the discipline being named "acting," you can't just act like you're going through the motions. You have to actually do it.

It was Method-lite for 14-year-olds, delivered a decade before Malcolm Gladwell argued that mastery takes 10,000 hours of practice: if you want to be convincing in a part, you can't just fake your way to mimicking the actions of a person. You have to understand the emotional, physical, and mental steps the character would be going through, and your performance will represent that empathy.

Truly brilliant advice takes years to sink in. I have trouble getting outside myself, so "acting is actually doing something" didn't mean much to me until long after I graduated high school. 

Now, in self-production mode, I think about it all the time. If I want the content to be good, I have to actually research and write the thing. I have to let the ideas marinate. And yes, it takes time. It is not automatic. Acting is actually doing something.

I'm always amazed at the business world's lack of respect for craft and the time it takes to properly create. I remain astounded at the number of content farms that only give writers 24 hours to turn around an essay comprising hundreds of words. No wonder the internet is filled with grey slush. The same companies are fine throwing dollars at software to fix the problem halfway, but not as comfortable paying people the same amount to get the job done right the first time. 

Yes, we all read the books on creativity and assert that everyone's an artist, but generally we don't acknowledge that good creators take the time and build the knowledge and craft to get it right.

If we want to have a creative creator economy, we have to acknowledge the reality and time of truly creative processes. We have to give creators the time to think through their work so we're not always fixing the errors in post.

II. The government doesn't value decoded creativity

My payroll management software indicated that I might be eligible for a research and development tax credit. The software served up a questionnaire and asked me to send over research documentation, workflows, and plans. Because I spent a good chunk of 2023 excising my business from my head into an automated Airtable to store ideas and production workflows, literally producing prototypes and pilots, I was excited that I could potentially owe the ol' U.S. government less than I had originally anticipated.

Upon visiting my tax accountant last month, I was told that, because the R&D was not destined to become a coded software product, those research-based efforts at systemization, planning, creating, and measuring were likely not eligible. I was not surprised; the past ten years have taught me that anything considered STEM is labeled profitable and innovative, while creative ideation and human-generated artistic or intellectual production is sidelined. 

It remains disappointing: constructing a scalable methodology to distribute original content via digital channels is not as valuable to the government as developing plans for yet another hastily constructed MVP of a marketing automation platform, or whatever VCs are up to these days.

Instead, creative professionals are relegated to winning the grant or competing for the few good jobs and, generally, winning the contest. Even though there is clearly a demand for new creative work, as represented by the success of YouTube and TikTok. Any innovation or new idea in the established system of creating and marketing content is not even on the table unless it's embedded in software.

III. Knowing how to code is not knowing how to create

Last year an acquaintance mentioned that, in addition to their full-time job, they were planning on launching both a newsletter and video podcast, both on separate topics. Coming from the tech world, they looked at the creator economy and thought, "I can do that." 

Yes, this person had interesting things to say about many subjects. But to my knowledge, they'd never held a camera beyond their phone or taken a media production course. They'd never worked on any production at all, except for in a management position. They knew how to code and did some writing here and there, but they had never faced the challenge of producing content on a consistent basis. I asked what their plan was, and they couldn't fill in any blanks.

The thing about great ideas is that they also need to be accompanied by the skills to execute to completion. Whether manifesting a business plan or a content strategy, the ability to get the job done consistency is far more important than the idea itself.

A year later, I have yet to see a pilot. I am rooting for this person, but I also know that producing content is far harder than it looks to outsiders. Part of the magic is making content look effortless, but that look of ease means the effort is out of sight.

IV. It's clear I know nothing about the craft of accounting

Automation enables my solo business. In 2024, I can self-produce a video with higher visual and audio clarity than most 1970s films. I can automate my contracting and invoicing system, construct sales and marketing workflows, feed my content to an AI adapter so it can produce derivative marketing assets to distribute automatically across the web. Technically it should all work together perfectly, and I should be reaping the benefits. 

But as creative producers know: most automation makes a product lossy. When I use automation to assist or a creative production task that I know how to accomplish "properly" without software, I find myself more often than not frustrated at the output. The software has taken shortcuts to spit something out, and it has failed to acknowledge the principles of the craft. 

Audiences unfamiliar with the nuances of the medium will not see the discrepancy between how the computer automates content and how a professional would construct a similar product. But I see the shorthand in the process, and I'm frustrated that I have to learn a new way of doing things and correct a machine's—or a CEO's—mistakes. The software company did not build with experienced users in mind.

I've spent most of this week customizing a "template" to build a website on a learning management platform so my original content can be hosted and monetized. In practice, because I cannot upload code or set sitewide styles with the website builder, the "template" becomes like a consultant who introduces unnecessary complexity, then doesn't understand what it takes to produce content for that complexity. I am left cleaning up a piecemeal, disconnected mess. 

Whenever I lean on a creative automation, the result is never exactly as I expected. I'm always reworking or straight-up recreating what the automator did, in the way I know will enhance the message. Perhaps I should just let it go and catch the fish who respond to an 8-email automated sales sequence written by ChatGPT, but I'm not yet ready to go there.

I know the areas where I'm less familiar with the real-life process may look messy to experienced professionals. Editing sound with an AI autofixer is not the same as working with an audio engineer. My tax accountant likely wants to advise me on how and when to use Quickbooks automations. The AI-enabled future of the creator economy is clunky, junky, and imprecise. 

V. Forever awed by real Hollywood movie magic

In the evenings I watch classic movies to clear my head. Recently I've watched The Revolt of Mamie Stover; My Dinner with André; Aguirre, the Wrath of God. All were crafted long before the days of automation. They filmed in studios and on location, in the jungle, in restaurants and subways. The cameras focus on real people's faces as they experience emotions and tell complex stories. Now that I know what AI-generated content looks like, I'm amazed at what a team of humans can accomplish with cameras, skill, and the filmmaker's belief that they are telling a cohesive story. 

Acting is actually doing something.

Werner Herzog's Aguirre actually depicts people dressing up as conquistadors and navigating turbulent waters. When the film shows an event that was faked because of general human safety and humanitarian reasons—no one was actually beheaded on set—the care taken to make that action look real on film is far more complex than an engineered prompt.

I hope that, in the near future, we'll distinguish culturally that the thought and care humans put into making something new is more desirable than the automated facsimile. That creation isn't only pattern recognition and anomaly detection; that true creativity means assessing the world around us and powering down autopilot, taking stake of what we see and considering, thoughtfully, "What is the best way to construct a solution, a workflow, an idea that connects with other people? And how would we actually build that, with our human brains and experience, before we bring software into the equation?"

Otherwise, like the argument in this essay, things get lossy very quickly.

Content tech links of the week

  • Models all the way down is essential to understanding how large datasets used to train generative AI are compiled and labeled. Why yes, AI does rely heavily on data labeling from an amateur rando from Wisconsin! This is my first introduction to KnowingMachines.org, and I'm keen to continue exploring the project.
  • Speaking of the origins of datasets, content engineer Michael Andrews describes a new-to-me term: paradata. I understand the concept–of course it matters how the data is gathered—but am glad it's getting renewed attention from the engineering crowd.
  • Is social media really destroying teenagers' brains? Nature takes a common but flawed argument to task. Considering Reviving Ophelia pathologized teenage depression long before the internet went mainstream, I'm on the side of "maybe it's the culture, not the conduit." But! It's most likely a mix of the two.
  • What's the logical fallacy behind content design? Valentine Watkins explores the "distinction without a difference."

The Content Technologist is a newsletter and consultancy based in Minneapolis, working with clients and collaborators around the world. The entire newsletter is written and edited by Deborah Carver, independent content strategy consultant, speaker, and educator.

Advertise with us | Manage your subscription

Affiliate referrals: Ghost publishing system | Bonsai contract/invoicing | The Sample newsletter exchange referral | Writer AI Writing Assistant

Cultural recommendations / personal social: Spotify | Instagram | Letterboxd

Did you read? is the assorted content at the very bottom of the email. Cultural recommendations, off-kilter thoughts, and quotes from foundational works of media theory we first read in college—all fair game for this section.

I appreciate the folks who build new things to counteract the media doomer tone of "everything on the internet is bad." That's why clunky, rudimentary PI.FYI is my new favorite social network, where you can find me effing around @fightwithknives. Here's an invite code; as of this writing there are 7 invites left.