I heard the phrase for the first time back in the nineties, when I was an Intel representative in a multi-company exploration of the then-new practice of Knowledge Management: “If only Siemens knew what Siemens knows”, a manager from that company shared a saying that was often used there. And indeed, a major issue with knowledge in large companies is the difficulty of sharing it across the organization. Research shows that employees spend about a fifth of their time trying to track down information they need – that’s one day a week – and they often fail to find it at all. As a result a lot of time is wasted and numerous wheels are laboriously re-invented.

There are many reasons for this problem, some technical, some cultural. One major issue is that everybody uses email, and email creates multiple “black holes” – isolated, locked repositories that email disappears into, never to be seen again, forever outside the reach of people who need it.

Email as a black hole

Before we look at those repositories we must acknowledge that much email doesn’t get stored at all, but is deleted after it’s read. It is simply deemed ephemeral, not worth keeping; but it may still include useful knowledge for future situations, and its deletion erases that knowledge.

But let’s consider the email that does get stored. These messages typically go into a user’s local repository – a PST file or equivalent – where the knowledge they contain is locked forever. It remains accessible to the user, but to nobody else. There are a number of worrisome mechanisms that make this a bad thing:

• Personal repositories are backed up much less rigorously than organizational repositories – if they’re backed up at all. The likelihood of a PST file disappearing due to some corruption or other mishap is higher (unless the user knows how to seriously back up stuff).
• If the user leaves the company, their machine is formatted and the knowledge in their email is gone forever.
• Even if the user and the repository abide, after a while the user may no longer be able to retrieve a desired message (unless they have a great search tool).
• Even if the user can find the information, other users may not realize which coworker to ask for it (think 100,000 employee corporation, with people moving around a lot).

So you can see how the same tool that still remains one of the best channels for asynchronous one to one communication is a very problematic one for knowledge management.

Pros and cons of going social

An obvious answer to this problem seems to be “Let’s put everything in shared repositories!” In reality that is equivalent to replacing email with social media that make communication in a team take place in shared spaces, like Slack or IBM Watson Workspace. Then all the messages are kept on a server, where they are searchable, and where backup is taken care of professionally.

This makes sense up to a point, and certainly many organizations are doing it. However, it is a great tool for teamwork, and not so great for one to one messaging, where email – like its predecessor, the written letter – is often the preferred method. So going social can preserve a lot of what “Siemens knows”, but if Jill drops Jack an email where she mentions an idea that occurred to her about a product feature, and Jack files it in a local folder, and two years later – after she’d left the company and he’d moved to another department – Tom could benefit from Jill’s thought, there will be no way for it to come to him.

What can we do about this?

One could envision having the email servers keep a copy of every message they convey and allow searching them; but this is not going to fly, because by its very nature one to one email (and, often embarrassingly, one to many) contains personal, sensitive, or compromising details. People will not stand for having their emails open to public scrutiny. Indeed, I still feel for the employees of Enron; after their miscreant company crumbled, the US government made the entire content of its email servers available to researchers worldwide…

So what’s left is to encourage people to use social channels for team communications, and to provide a means for employees to retain useful knowledge that comes to them via email. This may mean creating a searchable repository for ideas and encouraging employees to feed it regularly. This is much more than a technical problem (in fact the technical side is trivial); the main difficulty is that to retain knowledge requires time and effort, which people are not eager to invest for the greater good. I’ve seen this problem at work in organizations. What you ultimately need is to educate, motivate, and incentivize people to put in this effort; I’ve seen that work too. This is part of instilling a Knowledge Management culture, of course.

Will computers help?

These days I’m giving much thought to the possibilities of Artificial Intelligence as a means of helping people cope with the problems the computers have created. We already see AI alleviating the problem of email overload. I’ve written about the potential of AI in resolving the privacy issues in mobile app installation. So how about AI coming to the rescue with this black hole issue?

My thinking is that because email carries a mix of important ideas, unimportant ideas, small talk, personal matters, and so on, we need the recipient to sort them out and extract what’s important for a company repository; that’s the time consuming part that I refer to above as a hurdle. But an AI program would be able to read the message and figure out that the idea for a new feature is worth keeping, whereas the diatribe about the boss’s management style, or the invitation to lunch, are not. Such an AI program could pluck out of the email stream the bits and pieces Siemens – or your own company – need to retain in accessible form, and move them to the company repository, with tags and commentary it would generate itself. Or it might first pass the part it’s intending to store by the recipient for confirmation.

You think no AI program can do that? Actually you’re probably wrong even today, but if not, give Moore’s law a few more years…