There's difficulty in translating these new concepts into examples using existing real-world concepts but as a broad generalisation: a DAO is a "company" that the shareholders govern and the DAO's code is the shareholder agreement which has zero implicit rights or behaviours.
The DAO itself could have code that enables something like, "the code for this DAO can be updated as if 50% of token holders vote yes" and then 50% of token holders could vote yes to a code change that appoints a CEO who has absolute authority or they could vote to change the code so that no vote could ever take place in future, and the code becomes "stuck" forever.
The code for a DAO lives and runs on the blockchain, so the integrity of the DAO is linked to the integrity of the network on which it runs: although there's no absolutes, in the case of a network like Ethereum, it is for all intents and purposes, secure, so deployment and execution is not a network-level attack vector.
Does that help?
The reason DAOs are considered _the future_ by some is the implicit assumption that perfect code is possible to produce. Many non-software engineers believe that _if we have the integrity of the network to guarantee the code cannot be changed without consent, then we can have absolute faith in the code_... but of course, as software engineers, we know code is very fallible, whether it's unintended side effects or malicious backdoors or just an honest misunderstanding of what the code is meant to do, there's millions of ways for code to go wrong long before we need to worry about code integrity.
You have to accept the premise that code as an absolute authority is a good thing, if you accept that premise, a DAO makes a lot of sense and can deliver a lot of value: no longer are we beholden to the weakness of corruptible man, we are now empowered by the strength of noble technology.
of course, as a software engineer, you know that is a hellish nightmare because the code we write is fallible so this entire thing makes no sense whatsoever.
The DAO itself could have code that enables something like, "the code for this DAO can be updated as if 50% of token holders vote yes" and then 50% of token holders could vote yes to a code change that appoints a CEO who has absolute authority or they could vote to change the code so that no vote could ever take place in future, and the code becomes "stuck" forever.
The code for a DAO lives and runs on the blockchain, so the integrity of the DAO is linked to the integrity of the network on which it runs: although there's no absolutes, in the case of a network like Ethereum, it is for all intents and purposes, secure, so deployment and execution is not a network-level attack vector.
Does that help?
The reason DAOs are considered _the future_ by some is the implicit assumption that perfect code is possible to produce. Many non-software engineers believe that _if we have the integrity of the network to guarantee the code cannot be changed without consent, then we can have absolute faith in the code_... but of course, as software engineers, we know code is very fallible, whether it's unintended side effects or malicious backdoors or just an honest misunderstanding of what the code is meant to do, there's millions of ways for code to go wrong long before we need to worry about code integrity.