There's generally more than 1 cook in the kitchen, especially when you consider the direct and indirect use of third-party code. Besides, "good programmers" is not the same as "perfect programmers." They are people, and they naturally prone to honest mistakes and oversights like ourselves and everyone else. I also can't imagine anything applications you have in mind (given your descriptions of browsers, word processors, pdf readers, etc...) is limited to anything even close to a single developer.
Additionally, there is error with everything that is made. Even something coming off a factory line isn't the same every time, and has room for error/failure. The failure rate is a large part what determines whether something is designed & manufactured well, not the fact that any failure is inevitable. Even so, the quality of product support in the event that failure does occur is equally as important as initial prevention measures.
Now when it comes to source code specifically, "safe" is relative to a number of things. Whether or not code is safe from being exploited to compromise a system is one thing, but there is also security of the information in which the application deals with, security of the intellectual property (hence why obfuscators are commonly used with managed languages), etc... Nothing is ever completely safe from every risk and hazard out there.
Last week, TechCrunch reported that Google would add number portability later this year to Google Voice, which would let users keep one of their existing phone numbers as their Google Voice number. For example, users could make their cell phone number their Google Voice number.
It means a bunch of drunk drivers will be on the streets free to run whoever they want over... Meanwhile, prosecution of drunk driving will go down, and more drunk drivers will be on the streets.
I respectfully disagree. First, Florida has a waiver form a driver suspected of being impaired beyond their normal faculties may sign denying the breathalyzer, blood, or urine tests. One might initially argue this would have the same effect. However, the consequence of not taking this test is having your license suspended for 1 year. So even if this somehow helps you manage to avoid the normal 2 weeks of county jail time, you have a difficult year ahead of you.
You also need to consider that if you go to court and have a trial for a DUI charge, you're at the mercy of the jury. Reasonable doubt is a high standard, but that isn't synonymous with any doubt (especially hypothetical). I'm optimistic that if you really were impaired beyond your normal faculties and you took a field sobriety and had it recorded on a in-dash camera, there's a good chance a jury would find you guilty. Frankly, as long as police follow protocol and have someone present capturing the sobriety test on camera, proving a breathalyzer's accuracy is not necessarily required to meet the burden of proof.
On the other hand, imagine if you were on the receiving end and were falsely accused of driving under the influence of a substance and impaired beyond your normal faculties. Wouldn't you want the state to be required to do everything possible to meet its burden of proof? False readings are possible (though I suspect not probable), and as for alternative explanations to apparent impairment (i.e. swerving erratically), this could include medical conditions such as untreated/undiagnosed (and unknowning) diabetics. Combine that coincidence with a false reading and you've got yourself a great reason to change your opinion. ^_^ Granted, I wouldn't expect this hypothetical to have any significant probability of happening, but still, it's a good "what if" to ask.
Real Users find the one combination of bizarre input values that shuts down the system for days.