iSpot

» see ispotnature.org as it lived, on the wayback machine

~ the goodbye ~

iSpot opened on 27 June 2009, built and hosted by the Open University with money from the Open Air Laboratories network. The premise was narrow and it never widened. You photographed something alive and posted it with a location, and other people proposed a name for it. If you were a beginner you got an answer. If you were an entomologist you got something to answer.

What made it more than a forum was the scoring underneath. Proposing a name that other people agreed with earned reputation, and the reputation was not a single number: it was held per taxonomic group, so being trusted on spiders bought you nothing on grasses. When several names competed for one photograph, the site added up the reputation standing behind each one and promoted the winner to Likely ID once it passed a threshold. A 2015 study of the system found that where two or more determinations competed, the weighting improved accuracy in 57 percent of cases and in most of the rest either sharpened the name or exposed a precision nobody could actually support.

It ran like that for seventeen years, less sixteen days. The final counts were 77,000 nature observers and more than 800,000 observations carrying 1.6 million images between them, with over 30,000 species identified. On Thursday 11 June 2026 the Open University decommissioned it, having decided it could no longer provide the support the site required. ispotnature.org still answers today with a single line: this website has now closed. The link beside it goes to a research archive page where the team says a data archiving plan is underway and that access through the NBN and GBIF repositories is being explored.

~ the chronicle ~

~ nine dimensions of reputation ~

The reputation model had nine dimensions, one per broad group of organisms, and that design choice is the whole reason the archive is worth anything. A general karma score lets a popular user outvote a specialist. Splitting it by taxon meant the person who had put twenty years into hoverflies outranked everyone else on hoverflies and nobody else anywhere, which is roughly how expertise actually distributes in natural history.

It also meant the site could tell you how confident to be without a human adjudicating every record. Likely ID was not a verdict from a curator. It was a sum that had crossed a line.

~ what the data was for ~

The observations were never only for the people who posted them. iSpot existed inside a biological recording culture that feeds county recorders and the research literature, and the whole point of a photograph carrying both a date and a grid reference is that an agreed species name on it survives the person who took it. Papers were written off the corpus, and the Open University used it in teaching. Mid-2014 it had passed 42,000 registered participants, and by May 2015 half a million observations carried photographic evidence.

~ labels before the models ~

I work on computer vision now, the kind that has to run on a flying machine and decide what it is looking at without asking anybody. iSpot was that question answered from the opposite direction. The picture went up and a human who knew the group typed the name underneath, and the agreement of other knowledgeable humans is what made the name stick. Eight hundred thousand images with expert-agreed labels on them is, in the vocabulary of my day job, a training set that took seventeen years and no budget to assemble.

I was not one of the 77,000. The expertise that held the place up was British and Irish, pointed at British and Irish fauna, and the moths outside my window in Ankara were never in that reference frame. What I recognize is the shape of the ending: the institution that paid the hosting bill looked at a line item, and the line item did not know what it was attached to.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »