> Assuming e.g. a 50% chance of failure, having two teams reduces this chance to 25%.
...if failure is independent, sure. Not only is independence generally a dangerous assumption, but in particular when it comes to software engineering, it's been shown that problems are not independent. I don't have the reference at hand, but if I did, I could show you that completely separate teams tend to make the same mistakes.
This makes sense. Both teams must work with the same organisational rules, the same infrastructure, the same 'enterprise' technology strategy, often the same geographic constraints, the same marketing organisation, the same customer preconceptions about the company, etc, etc.
Nono it doesn’t make sense. What are the chances that both 50% failure chances of both teams overlap on 49% of each other? Bad luck! but it is rarely random: Start at the wrong time of the economic cycle for example. Also what are the chances each team poaches resources that the other needs? Then you increase failure rate to 75%…
and if the product's market only allows for say 2-3 dominant actors, the lack of focus might even mean that you lowered your chances of success quite a bit. just look at Google's multitude of misc chat and meeting solutions.
...if failure is independent, sure. Not only is independence generally a dangerous assumption, but in particular when it comes to software engineering, it's been shown that problems are not independent. I don't have the reference at hand, but if I did, I could show you that completely separate teams tend to make the same mistakes.