Why do we need google for that?
I am not sure what you are asking. It's almost like you're being obstinate for the sake of it.
Let me add something to what TheCreator wrote.
When you distribute your binary to someone, you generally distribute the release version. This does not come with debugging symbols, perhaps because you do not want someone to reverse engineer it, and also perhaps because of the smaller file size. Now if one of your users has a crash, they would before systems like Windows minidump and Google Breakpad were around, just post a report saying "when I am in the west hall and I am attacking somewhere in the attack it crashed". The programmer who received this would scratch their head.
However, with a minidump, the release crash metadata is written out. And in the cases where it is supported, sent to a remote server (reported). The programmer can look at this minidump and reconcile it to the debugging information they have for the given progam, and generate a callstack and perhaps even see values of variables. They can see exactly where it crashed, and perhaps even the reason why. I've even worked at a company where these dumps are automatically associated with an issue in the company's issue tracking system, and you don't even need to process them, the call stacks and debugging information are automatically added to the given issue and duplicates removed. So by the time you might get a reported issue, it might have a huge amount of automatically generated hugely useful information to help track down the cause.
The benefits are less if a scripting language is used. In that case you'd really need a custom system that extracted and appended scripting language frames into the callstack to show more than just virtual machine operation.
