Among the millions of images available on Wikimedia Commons is a stunning photo of a purple daisy with a yellow pistil. Taken in The Hague in 2004, it appears to be a variety of aster, a genus of flowering plants that mostly grow in Eurasia. It’s a nice picture, but its beauty doesn’t explain why the photo has received around 78 million hits each day — mostly from India — since roughly June 29 of last year. 

Chris Albon, the director of machine learning at the Wikimedia Foundation (the nonprofit that runs Wikipedia and Wikimedia Commons), said in a tweet Monday that the flower was accounting for 20% of data requests for media to one of the organization’s servers. Albon called the spike in traffic “an ongoing mystery,” and linked to a ticket on the Wikimedia forum where people were discussing what might be causing it. The flower photo soon began attracting attention on other parts of the internet, including Hacker News.

The original ticket on the Wikimedia forum indicated that the traffic was likely coming from a mobile app, a clue that helped guide early theories. One person identified a health conglomerate in India that has “Aster” in its name and speculated that its app or website might be using the photo and unintentionally sending traffic back to Wikimedia. Others remembered the generic image of the flower referenced as an example in their coding or design classes, leading them to wonder whether a particularly popular online course could be the culprit. 

Photos uploaded to Wikimedia Commons can be used by anyone for free and are often included as part of Wikipedia articles. Photographers from around the world have contributed to the project, including a Dutch man named Teun Spaans, who appears to be responsible for the picture of the purple flower. Spaans could not immediately be reached for comment.

Many online commenters noted that the purple flower had been used in sample code on sites like Stack Overflow, a question-and-answer forum for programmers. They theorized that an app developer may have accidentally copied and pasted one of the samples and forgot to remove a link to the photograph. Using code snippets from Stack Overflow is a common strategy among developers around the world.

“It appears to be a classic case of hot linking,” said Michael Donohoe, head of product at Rest of World. Hot linking is when a website displays an image by linking to the original site where it’s hosted, instead of uploading it separately. The practice allows the site borrowing the picture to avoid using their own server space — that’s why Wikimedia is forced to handle the requests to the flower photo. “As I’m sure you’ve gathered, people often do it by mistake [copy/paste code without knowing what it does]. When done by a developer, it’s usually sloppy coding and doesn’t look good,” Donohoe explained.

As speculators began trying to identify which app might be behind the traffic, another clue emerged: June 29, the day the flower photo first began attracting a significant amount of traffic, is the same day the Indian government blocked a group of apps made by Chinese developers, including the social media platform TikTok. Millions of Indians began flocking to the app’s domestic competitors, including MX TakaTak, Moj, and Josh. The evidence suggested that one of the TikTok replacements had inadvertently hot-linked to the flower image in their code. Then, after TikTok was banned and the copycat app suddenly became popular, traffic to the picture began spiking.

Public data collected by the Wikimedia Foundation indicates that the image has been receiving roughly the same amount of traffic since June 29, about 78 million hits per day. That suggests an app with tens of millions of daily active users is likely responsible. Traffic for the purple flower first began spiking around June 15, when it received roughly 5 million hits. That number quickly jumped after TikTok was banned in India and peaked at almost 120 million hits on New Year’s Eve. (Before June, the picture was getting only a few hundred hits a day.) 

For now, the culprit remains a mystery — at least to the public. Sukhbir Singh, an engineer at the Wikimedia Foundation, said on the nonprofit’s forum that, after extensive testing, it had identified the app in question and confirmed it was from India. He said the organization planned to release the name of it soon. “We have initiated contact with the app developers and are waiting to hear back from them. In the meantime, given the volume of requests, we have decided to ban those specific requests until the issue is resolved,” Singh wrote. The Wikimedia Foundation did not immediately return a request for comment.

Singh added that Wikimedia couldn’t find the image anywhere in the app itself, “confirming our theory that it fetches the image but does not display it.” In other words, every day, millions of phones are sending traffic to a beautiful photograph of a flower, but no one is able to actually witness its beauty. It’s like a weird version of the butterfly effect — except for throwaway code.