3 ms·

I don't think it comes from management books. The problem is deeper, it's psychological. At the level where full time, professional management begins it starts

I don't think it comes from management books. The problem is deeper, it's psychological. At the level where full time, professional management begins it starts to become very unclear how to judge a manager's success over short/medium periods of time. Yeah in the very long run "is the product a success" can be used to judge, but that captures a lot of people who aren't the manager and may take years to figure out. You need to evaluate their performance before then. So people come up with heuristics, like "does this manager meet their commitments". But software is inherently unpredictable so the answer to this question is always random. This results in managers trying to look like they're doing a good job by forcing early releases: they are thus seen to be "meeting their commitments" and must be good managers, with the quality issues showing up much later at which point they can just shrug and say, well, all software has bugs. Unless a layer of management higher up digs into individual bug reports and investigates really deeply to conclude, no, you forced an early release before it was ready, they won't be held accountable.

cookiecaper 7y ago
It's the principal-agent problem coming home to roost in a world without principals. Everyone is effectively an agent because the principals are sufficiently "diversified" that they don't really have substantial direct investment in long-term organizational success. What matters most is the short-term boost that you can use to dazzle the next employer. Every employee knows that they're going to stay for 5 years max because companies have made it impossible for there not to be a massive market disparity if you keep stable employment. There are no perks to long-term loyalty and thus no reason to consider an employer's long-term interests as identical to your own. The outcome of this is the predictable situation we see now. Like it or not, the company's culture comes from the top down. If the boss doesn't know or care whether the product is substantial, people who do will eventually be replaced at least until the boss is duly insulated from them. No one wants to rain on the parade of the guy who holds hire/fire authority. There's a wider thing here too, which is that the market doesn't know or care about any of the technical details either. You need the touted features to work correctly just enough to create sufficient ambiguity, probably about 10-15% of the time. If you invest enough engineering effort to get them working correctly 90%+ of the time, that's a great deal more money and time spent on engineering for no market benefit. A competitor who dumps that money into marketing and sales will come out far ahead because technical quality and reliability simply doesn't drive sales.

cookiecaper 7y ago

It's the principal-agent problem coming home to roost in a world without principals. Everyone is effectively an agent because the principals are sufficiently "diversified" that they don't really have substantial direct investment in long-term organizational success. What matters most is the short-term boost that you can use to dazzle the next employer. Every employee knows that they're going to stay for 5 years max because companies have made it impossible for there not to be a massive market disparity if you keep stable employment. There are no perks to long-term loyalty and thus no reason to consider an employer's long-term interests as identical to your own. The outcome of this is the predictable situation we see now. Like it or not, the company's culture comes from the top down. If the boss doesn't know or care whether the product is substantial, people who do will eventually be replaced at least until the boss is duly insulated from them. No one wants to rain on the parade of the guy who holds hire/fire authority. There's a wider thing here too, which is that the market doesn't know or care about any of the technical details either. You need the touted features to work correctly just enough to create sufficient ambiguity, probably about 10-15% of the time. If you invest enough engineering effort to get them working correctly 90%+ of the time, that's a great deal more money and time spent on engineering for no market benefit. A competitor who dumps that money into marketing and sales will come out far ahead because technical quality and reliability simply doesn't drive sales.