Re: Re: Re: Re: Re: Where is the paddle? We need the paddle!
June 25, 2009 8:25 AM   Subscribe

Email patterns can predict impending doom.

"Email logs can provide advance warning of an organisation reaching crisis point. That's the tantalising suggestion to emerge from the pattern of messages exchanged by Enron employees.

"After US energy giant Enron collapsed in December 2001, federal investigators obtained records of emails sent by around 150 senior staff during the company's final 18 months. The logs, which record 517,000 emails sent to around 15,000 employees, provide a rare insight into how communication within an organisation changes during stressful times." From the New Scientist.
posted by WPW (18 comments total) 6 users marked this as a favorite

 
Oh noes! The general public may be onto us! MeFites in the Cabal, I've sent you all emails with further instructions. You know who you are.
posted by iamkimiam at 8:31 AM on June 25, 2009


From an HR standpoint, it's kind of obvious- when things start falling part, cliques form quickly and faster- people start realigning their alliances and looking to cover their asses. Because typically fears and long standing gripes aren't welcome, people begin venting and sharing in small groups, and morale-wise, it increases the splinter effect amongst the group.

The avalanche effect kicks in once people start cutting off people, withholding resources, start trying to offer each other up, or attempt to make power plays all at once. Then your org goes to shit very quickly.
posted by yeloson at 8:32 AM on June 25, 2009 [2 favorites]


People communicate a lot in times of crisis. Film at 11.
posted by Ironmouth at 8:35 AM on June 25, 2009 [2 favorites]


I've analyzed five years' worth of emails from my dad, and the results would seem to indicate that we're approaching a crisis point in regard to the intelligence of blonde women.
posted by The Card Cheat at 8:38 AM on June 25, 2009 [11 favorites]


Isn't there another bit of lore/wisdom on late-night pizza deliveries in Washington DC? That their rapid increase portends some Big Event?

awesome title, btw
posted by jquinby at 8:42 AM on June 25, 2009 [1 favorite]


Maybe we should take this to MetaTalk.
posted by orme at 8:49 AM on June 25, 2009


I am sure there is a serious scholarly endeavor in there, but looking at e-mail logs to predict problems in the workplace seems ass-backward in some way. Reading that made me think of my partner's father. My partner grew up in North Dakota, and his father used to make fun of people who said things like, "You know how to tell it's really cold? When the snow squeaks," or "You can tell it's really cold when you have to drive a mile before the flat spots on your car's tires thaw and they're round again."

Partner's dad would say, "You know how you tell when it's really cold? You walk outside and it's really cold."

This reminded me of that.
posted by not that girl at 9:18 AM on June 25, 2009 [1 favorite]


People communicate a lot in times of crisis. Film at 11.

No, the finding isn't that people communicate a lot, it's that the graph of who they communicate with looks different.

And while the changes to the graph may be obvious from an HR standpoint, as with many such things, the opposite could also be post-hoc called obvious (oh people communicate more widely before a crisis because they're seeking information to understand what's going on and in a desperate attempt to coordinate to prevent the inevitable). The only way to mediate between the two obvious explanation is to actually measure what's going on, see which pattern actually holds, write up the whole thing, go through peer review, and then have it covered in the media so other people can say it's obvious.
posted by If only I had a penguin... at 9:21 AM on June 25, 2009 [5 favorites]


Things like this probably seem obvious to people who've worked in small organizations, or in parts of large organizations that have actually fallen apart. Within the actual "disaster area," the problem is pretty obvious to all concerned — that's why people are frantically emailing each other. That's not really what the finding is about, or where the interest is.

Imagine you're in the HR department of BigCo, which has tens of thousands of employees and tens or hundreds of offices all over the country. You're sitting in your cozy little cube, and you have no clue at all of the shit that's going down in the Memphis office, where 80% of the employees have gotten together and decided to quit en masse on Monday because the boss is a raving lunatic and nobody listens to them and they can't stand it anymore. The first thing you know of it is when all their resignations suddenly show up in your personnel system, or maybe when someone further up the chain of command asks "what the hell happened and why don't we have a Memphis office anymore?"

I've seen this happen. I've seen entire engineering teams, whole projects, suddenly disappear. Poof. To some people — the people involved — it won't have been sudden, it'll have been brewing for months or years. But to management, in a lot of cases it just happens, apparently apropos of nothing, because internal communication is so bad, or some people are failing to pass information along for various reasons.

To someone in a situation like that, being able to keep track of the pulse of an organization automatically, maybe receive some sort of alert when email volume or communication patterns change suddenly in a way that's indicative of a problem, this is really neat. Maybe instead of only learning about the Memphis office that's about to go off the radar after it's gone, you can get a hint that something is about to go sideways a week or a month earlier, and take preemptive action. (Of course in an organization as dysfunctional as the hypothetical one I've created, they probably wouldn't be able to do anything effective, but that doesn't mean that the monitoring tool doesn't look really awesome.)

I'm sure there's a huge potential market for tools like this, if you could make them work in the real world with any degree of accuracy. However I question whether that's feasible: there are a lot of non-catastrophic situations that could produce false positives, and false positives are the Achilles' heel of an early-warning system. Nobody is going to pay attention to the Real Thing if the system is spewing out warnings of impending doom every few minutes.
posted by Kadin2048 at 9:45 AM on June 25, 2009 [2 favorites]


Obviously.
posted by disclaimer at 9:47 AM on June 25, 2009


Isn't there another bit of lore/wisdom on late-night pizza deliveries in Washington DC?

With Capital in Panic, Pizza Deliveries Soar
posted by dhartung at 9:48 AM on June 25, 2009 [2 favorites]


People communicate a lot in times of crisis. Film at 11.

I think the actual point of the New Scientist article is actually that while communication increases amongst certain cliques or groups prior to a crash, this increased communication occurs a full cycle or year before the crash, and that information sharing between groups or cliques declines.
posted by KokuRyu at 9:56 AM on June 25, 2009


While I totally see the value in better management, the unfortunate part is that by the time you reach the factionalizing, the environment has already gotten so toxic that it's generally beyond repair.

What's sadly true, in all of these cases, is that the biggest problems are the ones that receive the least communication- the problems are silenced early on and made clear they're taboo subjects - whether it's an abusive person, poor worker, or major problem in the overall system.

It might be better to look at stuff like when people start taking more sick days, tardiness, time wasted on the web, or other signs of morale falling off... but again, that's usually after your hard workers have been burned.
posted by yeloson at 10:45 AM on June 25, 2009


To apply not that girl's rule of thumb to Kadin2048's example...
You know how you tell when the employees in Memphis are going to quit because their boss is incompetent and they're being ignored? You go to Memphis and the boss is incompetent and the employees are being ignored.
I wish organizations would just supervise their middle management...

Kadin2048: But to management, in a lot of cases it just happens, apparently apropos of nothing, because internal communication is so bad, or some people are failing to pass information along for various reasons.

...A social network warning system would be neat, but the real meaning of any (non-false) alarm is that upper management has failed. E.g. Why is internal communication bad? Why are people hiding information? Because the organization has inadequately supported and/or disincentivized accurate assessment, accurate reporting, and appropriate response.

Social network analysis produces some interesting findings but unfortunately the researchers tend to operate without reference to many non-trivial social forces. This is rather anti-humanist. I hope that whenever a group falls to its knees because of incompetence at the top, people point to the guilty and not just the needle on the warning meter.

On preview: similar to what yeloson said.
posted by halonine at 11:05 AM on June 25, 2009 [1 favorite]


the real meaning of any (non-false) alarm is that upper management has failed. E.g. Why is internal communication bad? Why are people hiding information? Because the organization has inadequately supported and/or disincentivized accurate assessment, accurate reporting, and appropriate response.

I agree completely. And I think most people — even executives — in big firms would agree. It's not like they're unaware that they suck at internal communication, they just can't figure out how to fix it. In many cases they're probably trying, but it's a hard problem: it's easy to say that a company should incentivize accurate reporting, but the "how" is the rub. I've seen companies spend millions of dollars to very little effect, trying to get people to talk to each other. I've seen them bring in consultants and motivational speakers and anthropologists. People are stubborn, they're sometimes irrational, they have motivations that aren't obvious or that they're disinclined to share, occasionally they're just plain weird.

Even in the best-case scenario (and I've worked at a large company that I think is probably the best-managed and most organizationally open-minded large company around) I'm not sure it's feasible to try and make internal communication in a large organization dependably non-dysfunctional. Dysfunctionality is, in my opinion, a defining characteristic of large organizations once they surpass the point where everyone knows everyone else. But because large organizations have some economies of scale, they live on and are profitable despite not really working that well. (In other words, a big dysfunctional company can out-compete a small functional one if the benefits of being big outweigh the benefits of poor internal communication, conflicting objectives, etc.)

That's not to say that large organizations shouldn't continually try to fix themselves, because in doing so they might get rid of the really bad issues, but just that I don't think they'll ever be as good as small organizations that use personal loyalty to their advantage rather than their detriment. And since they're guaranteed to always be fucked up in some way, there will always be a market for tools that sound an alarm whenever something bad happens or is about to happen.
posted by Kadin2048 at 12:09 PM on June 25, 2009


Kadin2048's point can be summarized as "if only we knew what we know", and it's one of the dreams of the internet (and of a lot of SF) to solve that problem. Heck, there's a Metafilter subsite that's devoted to querying 'the hive mind'. It's a fiendishly hard problem, even when all the participants are cooperating!
posted by Fraxas at 12:32 PM on June 25, 2009 [1 favorite]


Any disgruntled organizational cabal worth its salt will conduct its communications via personal email sent off-site. Everyone in a company should know that their work email is not private.
posted by oozy rat in a sanitary zoo at 1:44 PM on June 25, 2009


The issue is that self correcting measures are generally as open to abuse as anything- and there's the rub of it- people have personal politics, ambitions, rivalries, abusive behavior, friendships, nepotism and a billion and one possible things that cause them to selectively enforce or ignore policies and decisions.

And the sad truth is that the objective measurements- profit, efficiency, etc. can be increased for a period of time by burning bridges and creating toxic work environments - which inevitably collapses after a point.

It's easier to find people who, when you put them in a department, can "turn it around" and you see numbers jump up, but it's harder to figure out who in a department is keeping the numbers from falling, and ultimately those are the people you need to consider and protect. Toss in a few abusive folks, overwork these people, or reward them inadequately, and they lose morale, burn out, or move on, and then things start going to shit.
posted by yeloson at 1:47 PM on June 25, 2009


« Older The evolution of a Programmer, a classic of the ea...  |  John Dillinger... Newer »


This thread has been archived and is closed to new comments