Cost of Attrition

I’ve come to think a certain amount of attrition is needed to keep a product and the individuals working on it healthy. Back in my consulting days I worked with a few enterprise software companies that had a lot of highly tenured employees (think 10+ years). All the long-timers I met were skilled fiefdom builders with a strong bias toward opposing every change and maintaining the technical status quo because a lot of their value was in their vast knowledge of the way things already are. Without attrition you end up accumulating people you don’t want. Overall it seems like a path to ossification&#x2F;stagnation for your people and product.<p>From the individual perspective, staying too long slows down your rate of learning because you aren’t coming into contact with new people and technology at the same rate. Anecdotal, but lately I’ve interviewed some 8+ year tenured candidates that I wouldn’t rate above SWE II because it was literally a “one year of experience 10 times” situation. I try to change things up every 4 years or so to avoid ending up this way myself.

While the point seems worthwhile, the animations don&#x27;t help at all.<p>Stuff moves around for no reason (some graph layouting algorithm with easing I guess), the looping is bad and it&#x27;s annoying to wait for the changes and as they happen instantly<p>A simple picture for each transition would have been simpler to implement and actually showed something.

The author&#x27;s points are absolutely correct.<p>However, I don&#x27;t think that it makes any difference.<p>The entire industry has now shaped itself into a transient, mercenary, loyalty-free community.<p>It will take a long time to change that.<p>A lot of the trouble is the &quot;You go first.&quot; mentality. Who will be the one to stay at a company for many years, getting only 3% raises; regardless of their performance, as their company&#x27;s CEO keeps raking in millions of dollars, and lives a lavish, high-profile life?<p>Who will be the company that starts to treat their employees in a manner that proves they are worth staying at? This may mean higher pay raises, the CEO taking some of their profit (and the shareholders and VCs), and sharing it with the employees. Letting employees unionize, etc.<p>As people or companies are doing that, their competitors are running riot; acting as selfish, destructive and greedy as always. Many times, the competitors can crush the people trying to do the right thing.<p>So that generally means that governments need to step in, and help the people and companies to do the right thing.<p>As everyone knows, that&#x27;s pretty much a non-starter, these days.<p>The tech industry makes crazy money. When an industry makes money like that, everyone &quot;looks the other way,&quot; at truly awful behavior. The finance industry has been like that, for decades. Whereas industries that don&#x27;t make much money, like public education, social services, etc., are regulated up the wazoo, with an iron fist.<p>I was a manager for over 25 years. I feel that I was a good one. My employees seemed to agree. I kept many of them on board for decades, and these were folks that could walk out the door, and get huge pay raises (my company paid &quot;competitive&quot; salaries). I certainly never made that much, compared to what people are doing, these days. many new hires out of college make more than I ever did, as a senior manager.<p>I worked hard at being a good manager; and that often meant working around a company with a fairly rapacious HR policy (HR was run by lawyers). Most folks here, would (and have) sneer at me, for staying so long, and for doing the things that I needed to do, in order to be a good manager.<p>In my case, it was personal Integrity thing. I have a <i>really</i> stringent Personal Code. I know that&#x27;s unusual, and we can&#x27;t expect it from most managers.

Doesn&#x27;t mention the benefits of attrition. A network can be robust since none of the pieces leaves, but what is even more robust is a network that constantly renews itself and gracefully handles the process of swapping out parts. Constant attrition means that your company never starts relying on individuals, instead the process survives and thrives on its own, hiring new people who then hire new people before they leave etc. Such a process isn&#x27;t very attractive to individuals, of course, but they are very attractive to the owners.

Nice. But remember: the cost of non attrition can be keeping idiot coworkers around. Who actively destroy work with their anti-work, and waste time with idiotic discussions.

“This post is also available as a Twitter Thread” struck me as nonsensical as a restaurant saying “This meal is also available as a taxi ride”.

The other part of why people leave is they end up doing lots of support. Often when things go seriously wrong or even just regular problems the main people called on to fix it are the people that have been there a while and know the system best. They end up missing out on the interesting new projects. Its a good incentive to move.

Just look at FAANG: Retain developers at (almost) any cost.<p>Pay developers like it&#x27;s monopoly money.<p>You could call this &quot;The benefit of retention&quot;, though that may be a bit reductive.

Arguments of attrition aside, I don&#x27;t agree with this model.<p>Maybe I&#x27;m not understanding what an edge is supposed to be, but the author implies that &quot;something&quot; is lost when the edge disappears. That doesn&#x27;t make sense. Knowledge live in the node, not the edge. You don&#x27;t just forget a project you worked on just because someone left. If you lose access to knowledge when someone leaves, that&#x27;s because you did a bad job of preventing silos.<p>Further, someone who has deep domain knowledge is valuable even with 0 connections.

&gt;Beware looking at teams on a spreadsheet.<p>Beware looking at teams as a social network graph or org chart! Human teams are nuanced and require up-keep.. hence the existence of HR and management.

I have to ask one important question, even though talent is more valuable than money, companies still don&#x27;t have infinite amounts of money to simply rise pay for everyone.<p>Some companies operate in not that profitable markets or have bunch of other issues.<p>Cost of attrition is still dispersed over the time and while you loose the knowledge at hand it still might be cheaper to build it back over a year than drop cash on people right here right now.<p>There is such thing as time-value of money and company might prioritize other investments that in long term will outweigh knowledge lost as knowledge can be rebuilt and gains from other investment might not.<p>Of course one can say - there will be some knowledge that will be gone. From practical point of view, if that knowledge that is lost would be so valuable it would most likely resurface or would be rediscovered quickly by new hires. Maybe not ideally but still in a way that company can continue.

Counter-point: Tesla&#x2F;SpaceX have extremely high attrition rates, but are so successful they&#x27;ll likely be remembered in 100 years for their achievements.<p>(To be fair the article just points out that attrition has a higher cost than we realize, not that attrition is never worth it).

This feels like an organizational twin of an old idea of personal relationship, which is called &quot;weak tie&quot;.<p><a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Interpersonal_ties" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Interpersonal_ties</a>

The article doesn&#x27;t mention the double whammy of losing well connected engineers: morale and ripple effect. When a high performed with a good network leaves a company, it usually encourage others to leave as well.

&gt; Teams can be growing but have dropping tenure.<p>It seems like any team that’s growing (by a non-trivial amount) <i>would</i> have dropping tenure.

The cost of attrition is very real, probably higher than the cost of keeping low-performers on the team for large companies. For small startups the cost of attrition is probably negligible. Maybe the solution for big companies is to break them up into small independent startup like entities.

I don&#x27;t know. This is a nice piece of thinking, although I did get a bit bored after the first few points, but it perhaps ignores a couple of key realities:<p>1. Most obvious, people are going to leave, so you need to take that into account. You can&#x27;t keep everyone and I&#x27;d argue you shouldn&#x27;t try.<p>2. The impact of someone leaving is highly variable. Sometimes it&#x27;s a bad thing, sometimes it&#x27;s a good thing. Sometimes it&#x27;s pretty neutral. Different people add different amounts of value, and not all relationships are equal.<p>Of course high attrition can be a yellow or red flag - certainly a warning that you need to look at how management and leadership work - but some amount of attrition is a reality, and the outcome of that reality is not universally a &quot;cost&quot;.

If it&#x27;s enshrined woke culture that&#x27;s driving people away, mere $$$ won&#x27;t fix it.

Thank you for posting the blog version