Persistence and Success in the Attention Economy

Fang Wu, Bernardo A. Huberman

Abstract

A hallmark of the attention economy is the intense competition for visibility. Content creators often persist in uploading material, hoping to reach a wide audience. However, an analysis of nearly 10 million YouTube videos reveals a "persistence paradox": the more frequently an individual uploads content, the less likely any single video is to reach a success threshold. While technical quality tends to increase with experience, the probability of achieving a "hit" is statistically lower than winning a lottery, suggesting that novelty often outweighs mastery in the digital marketplace.

Introduction

In the world of content creation, the mantra "persistence is key" is ubiquitous. The assumption is that if you keep uploading, the algorithm will eventually find you. However, research by Fang Wu and Bernardo Huberman of HP Laboratories suggests that the reality of the attention economy is far more counterintuitive. In a zero-sum game for finite human attention, the dynamics of social media platforms may actually work against prolific creators over time.

Methodology

The study utilized a massive dataset crawled from YouTube, consisting of 9,896,784 videos from 579,470 unique producers. To handle the fact that older videos naturally accumulate more views, the researchers defined "success" relatively: a video is successful if its view count is in the top 1% of all videos uploaded during that same week.

To analyze these dynamics, the authors used a hazard function—a statistical tool from survival analysis—to calculate the conditional success probability. Essentially, they measured the probability that a producer's $k$-th video would succeed, given that their previous $k-1$ attempts had failed.

The Persistence Paradox

Contrary to the expectation that practice leads to success, the study found that the probability of success per video actually decreases with each subsequent upload. As shown in the data, both average view counts and the threshold for top-tier success decline as producers remain active on the platform.

Figure 1: Trends in attention. Average view counts (a) and the 99% quantile of view counts (b) both show a significant downward trend over the course of 100 weeks.

This decline suggests that success is heavily front-loaded. Most creators who eventually reach the top 1% do so very early in their careers. Approximately 34.1% of successful producers hit the threshold in their very first week of uploading, and more than half achieve their first hit before their fifth video.

Quantifying the Odds

The hazard function reveals a sharp decay in success probability as the upload index increases. On YouTube, your 100th video statistically has significantly worse odds of succeeding than your first.

Figure 2: The Persistence Paradox. The conditional probability of success (hazard function) declines sharply as the number of uploads (a) and the number of weeks active (b) increase.

When compared to a fair lottery—where every ticket has a constant 1% chance of winning—YouTube is far more punishing. A producer with a "persistence level" of 100 (willing to upload 100 times) has only a 22.6% cumulative chance of succeeding at least once. In a standard 1% lottery, that same level of persistence would yield a 63.4% chance of success.

Quality vs. Attention

The most striking finding is the disconnect between quality and attention. By using user ratings as a proxy for quality, the researchers found that the average rating of a producer's videos tends to increase with more uploads. Producers are learning and improving their craft, yet their success ratio continues to drop.

The authors suggest a "novelty decay" hypothesis: as a producer submits more content, their novelty wears off for the general audience. Even as technical skills improve, the broad appeal required to hit the top 1% diminishes, making it harder to capture the same level of attention as a fresh face.

Conclusion

While persistence is essential for skill development, it does not guarantee success in the attention economy. The platform's dynamics favor early hits and novelty over long-term technical improvement. This research suggests that many creators may be overestimating their true odds of winning, continuing to persist despite diminishing returns, much like gamblers chasing a longshot.

Summary updated

Abstract

A hallmark of the attention economy is the competition for the attention of others. Thus people persistently upload content to social media sites, hoping for the highly unlikely outcome of topping the charts and reaching a wide audience. And yet, an analysis of the production histories and success dynamics of 10 million videos from \texttt{YouTube} revealed that the more frequently an individual uploads content the less likely it is that it will reach a success threshold. This paradoxical result is further compounded by the fact that the average quality of submissions does increase with the number of uploads, with the likelihood of success less than that of playing a lottery.
BibTeX
@misc{wu2009persistence,
  author = {Fang Wu and Bernardo A. Huberman},
  title = {{Persistence and Success in the Attention Economy}},
  year = {2009},
  eprint = {0904.0489},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/0904.0489},
}
Figure