I was looking something up the other day. It was the origin of “learning in the flow of work”, a phrase you’ll have seen in slide decks, vendor brochures, conference keynotes and job adverts. The source I found said the term was coined by Josh Bersin in 2018. Plenty of others say the same.

I remembered Jay Cross talking about this fifteen years earlier. That doesn’t establish who first used those exact words, but it does make me question how easily the history of an idea gets reduced to the person most recently associated with it.

Jay was writing about workflow learning in 2003. His Internet Time Group launched a site that year describing task-specific content and simulation built into the work itself, alongside real-time collaboration. In a 2004 paper, he argued that people learn far more about their jobs in the coffee room than in the classroom and that web services would bring workflow learning into practice. A year later, he and Tony O’Driscoll were writing about it in Training Magazine.

Jay pointed back further himself, to reading Gloria Gery in the 1990s. She argued that the context is the workflow and the content is whatever the user needs to do the work. That willingness to acknowledge what came before was part of how Jay worked. When people called him the father of informal learning, he explained that he hadn’t invented it.

I’m not suggesting Bersin took someone else’s work. He helped make the phrase familiar, and there is value in getting an idea heard and understood. But crediting someone with a phrase can easily become crediting them with the thinking behind it. Harold Jarche noticed this too, questioning accounts that presented learning in the flow of work, or workflow learning, as the latest advance in L&D.

Jay died in November 2015. Looking for his earlier work, I found that parts of his online presence are disappearing. Links that used to work no longer do. Some of the writing survives through other people’s posts, quotations and references, but following those back to what Jay actually wrote is becoming harder.

It made me wonder at what point someone’s work is forgotten. Is it when the last link breaks, or has something already happened before then?

Broken links are certainly part of the problem. Pew Research reported in 2024 that 38% of webpages that existed in 2013 were no longer accessible a decade later. A personal website can depend on someone continuing to renew a domain, pay for hosting and maintain it. Once the author has died, that responsibility has to pass to someone else if the work is to remain available.

But an earlier source doesn’t have to disappear to be overlooked. People cite the account they have read, and other people repeat that citation. Over time, the reference can move towards whoever is more visible or easier to find, while the earlier work remains available to anyone who knows to look for it.

Robert Merton described the Matthew effect, where recognition accumulates around those who already have it. He also used “obliteration by incorporation” to describe ideas becoming so absorbed into common knowledge that their sources are no longer acknowledged. These are different processes, but both help explain how an idea can remain familiar while its history becomes difficult to recover.

Much of L&D’s thinking is published through blogs, conference sessions and personal websites. Those are useful places to develop and share ideas, but the reference trail can be thin. We have seen how readily accounts of 70:20:10 circulate without much attention to what the original work did or didn’t establish.

Some of Jay’s writing is easier to trace because it was published elsewhere. His 2004 paper has a formal publication record and a DOI. Stephen Downes’s posts provide a contemporary record of work Jay was sharing. Neither guarantees permanent access, but they give us other routes back when an original link fails.

That is where the broken links become more than an inconvenience. They make it harder to check an attribution, recover a qualification or establish whether the original author meant what someone now says they meant. Those of us who remember an earlier source can question the account in front of us. Someone encountering the idea for the first time may have no reason to do so.

If we use learning in the flow of work, it is worth acknowledging Jay’s contribution and the earlier work he drew on. More generally, naming an author, a title and a date gives the next reader something to follow if a hyperlink stops working. Keeping an archived copy gives them a better chance of being able to read it.

If someone’s work has shaped how you think, go and check what is still available, what has been archived and what needs saving.