Making decisions about technical debt is hard. Anybody who tells you otherwise is someone you want to stay away from in a work environment. Every single one of the decisions you make is context dependent, sometimes massively so. You need to know where you are and where you need to get to at every step. Only then can you make good decisions. For what it's worth, here is my list for the places where most people get it wrong.
1) People lie about what is required. Anyone who has been in the industry for more than a token amount of time has experienced the PM who says that they need X+100 when they really need X because "programmers are lazy". Similarly everyone has worked on a team where one or more of the programmers are more interested in building their beloved architecture than solving the problem at hand.
2)People don't listen to the needs of others. Very often they assume problem #1 exists ("They don't actually need that") and blindly forge ahead without paying attention. They play whatever political game that is required to force the other side to follow their approach.
3) People make mistakes. Often this is due to a lack of experience, but sometimes it's just because the problem is hard. Some people make mistakes more often than others. Good management will recognize those that are making good choices, but sometimes you have to let people crash a few projects before you realize that they don't know what they are doing.
4) Choices are made at the wrong level. With all of the other points I've made, it is exceptionally tempting to move all the decision making to someone who acts truthfully, listens well and has a good track record. However, the information required to make a good decision can be at many levels. Sometimes you need 100% test coverage of something and only the programmer will know when this is going to happen. Sometimes you need to deliver tomorrow even if it doesn't work at all. Only the PM will know this.
In order to build a team that executes at the highest possible level, you need to have people who work well together, trust each other and are competent. If you do not have this, then success will likely be a matter of chance. No amount of heroics from a single person can reliably fix the issue -- and often such heroics just make matters worse because they result in the behaviour listed above.
1) People lie about what is required. Anyone who has been in the industry for more than a token amount of time has experienced the PM who says that they need X+100 when they really need X because "programmers are lazy". Similarly everyone has worked on a team where one or more of the programmers are more interested in building their beloved architecture than solving the problem at hand.
2)People don't listen to the needs of others. Very often they assume problem #1 exists ("They don't actually need that") and blindly forge ahead without paying attention. They play whatever political game that is required to force the other side to follow their approach.
3) People make mistakes. Often this is due to a lack of experience, but sometimes it's just because the problem is hard. Some people make mistakes more often than others. Good management will recognize those that are making good choices, but sometimes you have to let people crash a few projects before you realize that they don't know what they are doing.
4) Choices are made at the wrong level. With all of the other points I've made, it is exceptionally tempting to move all the decision making to someone who acts truthfully, listens well and has a good track record. However, the information required to make a good decision can be at many levels. Sometimes you need 100% test coverage of something and only the programmer will know when this is going to happen. Sometimes you need to deliver tomorrow even if it doesn't work at all. Only the PM will know this.
In order to build a team that executes at the highest possible level, you need to have people who work well together, trust each other and are competent. If you do not have this, then success will likely be a matter of chance. No amount of heroics from a single person can reliably fix the issue -- and often such heroics just make matters worse because they result in the behaviour listed above.