Stack Overflow fell below 1,000 questions a month, for exactly one month

On 28 October a Hacker News submission titled "StackOverflow now gets fewer than 1000 questions a month" reached 816 points and more than 700 comments. It pointed at a Stack Exchange Data Explorer query whose graph counts question posts by month. The claim was correct, for exactly one month.

The milestone lasted a month

The Stack Exchange API's questions route, which counts the same posts that query charts, returns 969 question posts for September 2026. For October, counted through 30 October with a day still to run, it returns 1,028. August was 1,019; July was 1,175. That is not a cliff but a plateau with a dip, and the dip has closed.

A threshold makes a good headline because it hides that shape: four months of drift become a single date on which something ended, and the round number that ended it was chosen by the audience, not the data.

The floor is near, the slope has flattened

What is not in doubt is scale. March 2014, the peak month, took 207,169 questions; against roughly a thousand now that is a fall of about 99.5%. Stack Overflow's first full month, August 2008, produced 3,744, nearly four times October's count.

The Meta post that framed the debate puts the daily figure at 88 in February 2026 and 42 in May, and monthly active users down from over 200,000 in mid-2020 to 11,000. Its own forecast, under one new question an hour by the end of the year, implies under about 720 a month. October ran roughly 40% above that pace.

Noise, or a moderation change

A single month of a public API count is a noisy instrument. Weekday patterns, a moderation backlog clearing, a retagging sweep, a holiday week: each shifts a four-digit total by tens, and none is visible in the figure. September's dip could be a real step down or one of those. October's recovery is as consistent with noise as with revival, and the count distinguishes neither.

Why the thread split on blame

Attribution is where the comments divided. Some read the graph as a verdict on Stack Overflow's ownership and moderation culture; some noted the peak was 2013 and the slide began before anyone had a chatbot to blame; others argued the questions migrated into GitHub issues for project-specific problems and into assistants for the rest. The Meta post adds lay-offs, a fact about the operator, not about demand.

The defensible summary is narrower. The floor is near and the slope has flattened; 2008 is now a taller bar than the latest month. Whether that is a death, a plateau or a change of address cannot be read off a question count, and a one-month milestone cannot be read at all.