On the morning of March 14, 2026, a retired pediatrician named Donna Reaves of Tallahassee filed her 8,412th checklist to eBird. She had spent forty-one minutes at Lake Munson, recorded twenty-three species, and noted in the comments that the limpkin pair were on the same snag as last week.
Her checklist joined roughly forty thousand others submitted that day. It was a Saturday in spring migration, which is to say a normal, unremarkable day for the platform.
The Cornell Lab of Ornithology launched eBird in November 2002. The first year produced about a million observations. By 2010 the platform was approaching ten million. As of the most recent state-of-the-birds release, eBird holds over 1.8 billion observations from more than nine hundred thousand contributors in 252 countries and territories.
Those numbers are the headline. The more interesting story is what was given up to get there.
When Steve Kelling and the early eBird team made the platform free to use and free to download, they bet against decades of ornithological habit. Field data had historically been guarded. State records committees verified rarities slowly. County lists were currency. The idea that an amateur in Tallahassee would casually contribute to a global dataset that anyone, including a competing PhD student, could pull at any hour was unfamiliar and, for some, unwelcome.
What changed it was friction. eBird made entry easier than the alternative. A checklist could be filed from a phone in the field. Hotspots populated automatically. Personal lists were a side effect, not the goal — the database was the goal, but the database paid the user back in life lists, year lists, county lists, and the small pleasures of comparing your spring count to your neighbour's.
Twenty-three years in, the platform's strangest accomplishment is that its scientific value is now inseparable from its social one. People file checklists because their friends do. They check their bar charts against the global model because it is interesting to be wrong.
The science has compounded slowly. The Status and Trends models, published periodically out of the lab, now estimate weekly abundance and migration timing for over a thousand species across the Western Hemisphere at three-kilometre resolution. Those maps would be impossible from professional surveys alone. The North American Breeding Bird Survey, gold-standard for sixty years, covers about four thousand routes annually. eBird, in a single May weekend, produces several million observations across the same continent.
Critics of the platform point out, correctly, that the data are messy. Effort is uneven. Observers vary in skill. Some hotspots are oversampled and most counties are undersampled. Rural Mississippi has fewer eBirders per square kilometre than Manhattan, which produces a strange distortion in any naive map.
The Cornell team has spent two decades developing statistical corrections for this. The current models account for observer effort, checklist duration, distance travelled, time of day, and a long list of other covariates. They are not perfect. They are, however, an improvement on what existed before, which in many regions was nothing.
Pell Murphy, who edits this section, filed his first eBird checklist on a wet morning in Pisgah National Forest in April 2014. He remembers the experience as awkward — he was unsure of the protocol codes, uncertain whether to count a heard-only bird, and convinced his identification of a worm-eating warbler would be challenged. It was not. He has now filed over four thousand checklists. The first one is still in the database.
That permanence matters. A checklist filed in 2007 about an unusual sparrow at a backyard feeder in Rapid City, South Dakota, is still queryable today. The observer is dead. The feeder is gone. The data are not.
There is a generational question worth taking seriously. The median age of eBird's most active contributors is now fifty-eight. The platform skews older, whiter, and more male than the general birding population, which itself skews older, whiter, and more male than the general public. Cornell has been candid about this. A platform that depends on volunteer labour is also depending on who has the leisure to volunteer.
Several efforts have tried to broaden the contributor base. The Group Me Birding program partners with urban birding clubs in Atlanta, Detroit, and Houston. The eBird Caribbean portal, launched in 2010, has trained over four hundred local observers across fifteen island nations. Translation into Spanish, Portuguese, French, Mandarin, and most recently Tagalog has expanded the platform beyond its English-speaking origins.
Whether these efforts will shift the demographic centre of the dataset is unclear. Habits set in a discipline's first generation tend to persist.
The platform has also forced ornithology to grapple with a question it had previously been able to avoid. If amateurs can produce data at this scale and quality, what exactly is the professional ornithologist for? The answer, increasingly, is what it has always been — to ask better questions of the data and to design the surveys that fill in eBird's gaps. The Breeding Bird Survey continues. The Christmas Bird Count is in its 127th year. The Atlas projects in individual states are if anything more rigorous than before.
eBird has not replaced these. It has made the questions they answer larger.
There is one measurement worth pausing on. In 2013, a paper led by Daniel Fink at Cornell used eBird data to map the spring migration of the wood thrush across eastern North America at weekly resolution. It was the first time a song-bird's continental movement had been mapped at that granularity, and the resulting figure — a slow green wave moving north through April and May, then a slower wave south through August and September — circulated widely outside ornithology. People who had never thought about birds saw it and understood, perhaps for the first time, that what happens in their backyard in May is connected to what happens in Honduras in February.
The platform's quietest achievement may be this: it has made migration legible to people who never knew they were watching it.
Donna Reaves does not think of her Saturday checklists as data. She thinks of them as a record of what she saw. The fact that those records, summed across nine hundred thousand other observers, have become one of the largest biological datasets in human history is not, to her, the point.
Which is, perhaps, why it has worked.
Filed under
