Comment by thesumofall | Hacker News Reader

Comment by thesumofall | Hacker News Reader

Parent Full thread

Why would we see a 80/20 distribution when most human abilities are distributed on a bell curve? I don’t think we really have the tools to reliably measure performance, because performance can mean so many things in so many different contexts. And yes, this can include “process.” The referred McKinsey study talks about politics. Is success in democratic politics really defined by content or by your ability to build a majority around certain ideas that may or may not be your own? Or: should we really dismiss the success of Toyota’s Lean Mgmt at its time which was all about process? My experience is that having “process top performers” sprinkled into a typical team can make all the difference. They ensure that the team can apply its abilities to the right topics in the right sequence avoiding costly detours or entire failures. Or how often do teams deliver output but somewhere along the lines forgot to ask what the initial objectives were actually all about? Once more: process

I’m not downplaying the relevance of content, but I’m simply arguing that performance is much more nuanced and diverse than just being good at content

Because we don’t know how to measure output properly and corporations often misattribute success. I’ve seen it happen a few times where someone delivering crap quickly is praised (oh, look how productive he is, so many tickets closed) or someone who has good autopromotion skills (does little but talks a lot about it) or someone who simply led the project, and not the guys who did the high quality work. Then there is often a group of people fixing stuff after a super star moves to another sexy project.

There is a saying „if you want to go fast, go alone, but if you want to go far, you need a team”.

> Why would we see a 80/20 distribution when most human abilities are distributed on a bell curve?

Generally you get normal distributions (bell curves) when you add together random variables. For example, if you roll 10 dices and add their numbers, the probability distribution will look somewhat like a bell curve. And if overall employee performance was the sum of their normally distributed human abilities, you would get another bell curve.

But it only works when add numbers. If you multiplied numbers instead of added them, you would get a log-normal distribution instead. Other processes can get you Pareto distributions (80/20).

Employee performance is a not just a sum of their individual abilities, there are some synergies happening, giving a multiplicative effect.

In other words, seeing a 80/20 type distribution shows that people are more than the sum of their parts, literally.

Because ability does not mean desire to work :P

> Why would we see a 80/20 distribution...?

You're mis-quoting. Here's the quote:

> Organizational researchers have shown, similar to the 80/20 rule, the majority of your company's output comes from a minority of "superstar" performers: What's known as a power-law distribution.

The cutoff is arbitrary, e.g. one of the cited studies uses a 5/95 cutoff. [0]

[0] https://www.mckinsey.com/capabilities/people-and-organizatio...

But doesn’t change that it isn’t a bell distribution

I hesitate to write this, but I think a counterbalance is needed. In software development, there are things you can measure. For example, you can put a new starter on a big pile of bugs and say “get through these”. You can measure throughput of work tasks, or story points, or lines of code. Can people game this stuff? Yes, but managers can be the check on such things (and have managers of their own).

I know someone that works in a sales oriented profession and who manages a large number of people. Their company is ruthless, you get only a couple of weeks to demonstrate you can hit the targets, and after that maybe a couple of weeks of grace after the hard talking to. Less than 50% of new hires make it through probation. Staff that survive all have the same attributes: excellent sales ability and focused determination. It also allows in a high percentage of narcissistic bastards though.

I don’t like the idea, but I wonder if a similar metric based probation would achieve the same in this industry. If you want people who can pump out quality content at a higher than average rate, set the bar high and cull ruthlessly during probation.

It has always amazed me how differently sales people are treated compared to tech.

One of the historical reasons has been that it’s very hard to find and hire tech people - I wonder if that is changing?

It’s also been fairly easy for tech teams to obfuscate performance and bamboozle managers who don’t understand tech.

With sales the performance criteria tends to be very simple and there aren’t really any good excuses for failure (at least none that seem to be accepted).

I think this is because sales tends to be a one-person job in most settings (except the very highest value deals). One sales person manages a number of accounts. Measuring the results of that person is then quite straightforward. Tech tends to be a collaborative job even in quite simple settings, which makes it immediately more difficult to measure. Add to that that the true results are often only visible after months if not years of work and you have a real challenge at hand. But I agree: there are also simple settings where you can measure a programmer’s productivity. Those are probably just not the settings in which people are that concerned about attracting top talents, though

> It’s also been fairly easy for tech teams to obfuscate performance and bamboozle managers who don’t understand tech.

Just as much a problem for sales oriented roles. Most “line managers” are experts in the field, so I don’t know if this is a valid point. In the anecdotal case, the managers still have sales targets and got to be managers by being the most effective sellers.

> With sales the performance criteria tends to be very simple and there aren’t really any good excuses for failure (at least none that seem to be accepted).

Here we are talking about producing, which by its nature is the measurable stuff. I think we don’t have it because we (some we) don’t want it, not because it wouldn’t work.

"Less than 50% of new hires make it through probation. Staff that survive all have the same attributes: excellent sales ability and focused determination."

Sales is cut throat. They take in new people to extract all their first rate contacts (friends family). Once those are sucked out, they don't need you.

In some areas like hft it was effectively achieved with engineering. For a time… I hear it’s changed now.

The problem is in most software businesses it can be hard to objectively measure who is providing business value and who is just lots of busywork

Reply on news.ycombinator.com