Two Paragraphs, Ten Versions
This experiment investigates Hito Steyerl’s concept of the “poor image” by transferring it from the digital realm into the human one. Instead of watching a file lose quality as it travels through networks, I observed how a text loses fidelity while travelling through memory, attention, and individual interpretation. Two separate paragraphs were selected from Steyerl’s essay In Defence of the Poor Image. The first was used for audio reproduction; the second for reading-based summarisation. Five participants interacted with both — creating a total of ten derivative versions.
For the audio stage, each person listened to the same paragraph read aloud three times in a row. After the third hearing, they had to retell the paragraph from memory. Despite identical input, each retelling diverged quickly. The opening sentence remained surprisingly stable across all participants — almost everyone repeated “A poor image is a copy in motion” with minimal distortion. But from the second sentence onward, the text began to erode. Complex vocabulary disappeared first; long noun chains collapsed; metaphors were flattened. The verbal versions became shorter, more fragmented, more repetitive. Participants remembered motion, deterioration, copying, ripping, remixes — the verbs survived, but the conceptual architecture dissolved.
The oral retellings behaved like Steyerl’s “poor image”: highly compressed, low-resolution, and shaped by the limits of the transmission.
For the reading stage, a different paragraph was given to the same five participants. They had exactly one minute to read it silently. After the minute ended, they wrote a summary based only on memory. These written reproductions showed a different type of degradation. The writing was calmer, more structured, but far more selective. Readers preserved abstract ideas (“quality becomes accessibility”, “the image becomes more abstract”, “uploaded, shared, reedited”) rather than the original phrasing. Instead of mishearings, there were omissions. Instead of linguistic drift, there was conceptual prioritisation. Each participant kept the parts that felt meaningful and discarded everything else.
What emerged from these ten versions is a two-part model of the “poor image”:
- audio → chaotic degradation, where the text breaks down through noise, forgetting, and improvisation;
- reading → strategic compression, where the text becomes a reduced conceptual skeleton.
Together, they reveal that the poor image does not depend on screens or file formats. Human cognition itself produces low-fidelity copies. Memory is lossy. Attention edits. Understanding transforms.
Meaning survives not through accuracy, but through circulation. In this experiment, the idea travelled — and in travelling, it became something new.
References
Steyerl, H. (2012) In Defence of the Poor Image. In: The Wretched of the Screen. Berlin: Sternberg Press, pp. 31–45.
Queneau, R. (1998) Exercises in Style. Translated by Barbara Wright. London: John Calder.