Fact check: How do different sources estimate crowd sizes at political protests?

1. Summary of the results

Different sources employ a variety of methods to estimate crowd sizes at political protests, ranging from traditional techniques to cutting-edge technology. Traditional methods include the Jacobs Method (also called the Jacobs Crowd Formula), aerial photography, and manual counting systems ^{[1] [2]}. These established approaches have been used by reporters and crowd management professionals for decades.

Modern technological approaches have significantly expanded estimation capabilities. Cell phone location data has proven highly effective - a 2017 study of the Women's March protests used location data from 10 million individuals to validate that news and social media data can generate accurate crowd size estimates ^[3]. Advanced digital systems now include Convolutional Neural Networks (CNNs), counting by detection, counting by regression, and image analysis software ^{[4] [2]}.

Recent developments show the emergence of crowdsourcing methodologies. The June 2025 "No Kings Day" protests demonstrated this approach, where data journalist G. Elliot Morris and outside analysts used collective crowdsourcing efforts to estimate attendance at 4-6 million people across multiple cities ^{[5] [6] [7]}. This method has been compared favorably to the 2017 Women's March turnout estimates.

Video analytics and AI-powered solutions are increasingly used for real-time crowd estimation in public areas, mass transit, and retail environments, with built-in software analytics providing automated counting capabilities ^[8]. Knowledge distillation techniques are being applied to create lightweight models for crowd density estimation that can operate efficiently at the edge ^[9].

2. Missing context/alternative viewpoints

The original question doesn't address several critical aspects of crowd size estimation that create significant political and financial incentives for different stakeholders:

• Media organizations benefit financially from dramatic crowd size claims, as larger numbers generate more clicks, views, and advertising revenue. News outlets have varying accuracy levels when reporting protest sizes, with some methods performing better than others ^[3].
• Political organizers and movements have strong incentives to inflate crowd numbers to demonstrate popular support and attract additional participants and funding. The recent "No Kings Day" protests specifically used crowd size estimates as a political weapon against the Trump administration ^{[6] [7]}.
• Government authorities may have competing interests - sometimes minimizing opposition protest sizes while maximizing supportive event attendance. The Trump administration claimed 250,000 supporters for their military parade while independent analysts estimated the opposing protests at 4-6 million ^[5].
• Technology companies developing crowd estimation software have commercial interests in promoting their solutions' accuracy and capabilities, particularly in the growing market for public safety and retail analytics ^[8].

The analyses reveal that crowd counting remains highly subjective despite technological advances. The controversy over a COVID 'freedom' protest in Melbourne, where estimates varied dramatically, illustrates how different methodologies can produce vastly different results ^[1].

3. Potential misinformation/bias in the original statement

The original question appears neutral but omits the inherently political nature of crowd size estimation at protests. By framing this as a purely technical question about "different sources," it fails to acknowledge that:

• Crowd size estimates are weapons in political warfare - the recent "No Kings Day" coverage explicitly frames crowd numbers as "warnings for the Trump administration" and "rebuking Trump" ^{[6] [7]}.
• There are no truly neutral parties in political protest crowd counting. Even academic researchers and data journalists like G. Elliot Morris operate within political contexts that may influence their methodologies and interpretations ^[5].
• The timing and selection of estimation methods can be strategically chosen to support particular narratives. The emphasis on crowdsourcing for anti-Trump protests while questioning official counts for pro-Trump events demonstrates this selective application.

The question's framing suggests that crowd size estimation is primarily a methodological challenge rather than recognizing it as a highly politicized practice where different stakeholders use varying techniques to support their preferred narratives about public support and opposition.