I'm sorry about your personal situation. That's awful, and highlights the need for human review and for flexibility in public documents, especially highly publi

I'm sorry about your personal situation. That's awful, and highlights the need for human review and for flexibility in public documents, especially highly public documents like those displayed by Google. However, what you've clearly demonstrated is that there is neither human review nor flexibility in these systems. Google doesn't have a malicious employee with a grievance against you (or against the other Andreas affected) who is cruelly searching out incorrect images to display next to your name and your story to cause harm. You were close to the root of the problem when you wrote "...to have a large corporation come along and continuously and consistently misrepresent my work and cause distress..." - the thing which is misrepresenting you and causing distress is an unmoral computer program. (Unmoral: Lacking awareness of moral standards, in contrast to immoral, intentionally rejecting morality, or amoral, aware of but acting without regard to morals.) The computer program doesn't have a concept of distress or harm. All it knows to do is to display its best guess of a portrait of an author when someone searches for the author's name. It has no ability - was not programmed with the ability - to understand that sometimes there may not be a photo of the person or that sometimes this is not good. That's still anthropomorphizing the program - it does not understand anything and does not know what is good. It just creates knowledge boxes with photos to accompany searches for author names, always. When you contacted Google, the person who responded to you may not have been a person - instead, the entity behind your email was likely another computer program, designed to handle complaints about faulty knowledge boxes. If it was a human being (unlikely), they did not go to the source code file for the website google.com/search?q=andrea and delete the \<img\> tag, nor did they add an understanding of the concept that some authors don't have public photos and some knowledge boxes ought not include a photo to the knowledge box algorithm. Either the program or the human just flagged the first image as incorrect, and let the knowledge box program run as it did before and find another image. There is not (yet) a fault mode that indicates "this person's knowledge box ought not be accompanied by a portrait" or "this person's knowledge box ought not exist". What other comments seem to be suggesting is that tech companies ought not run algorithms that work well 99% of the time, nor fault handling algorithms that handle 99% of the faults effectively. I'd take a slightly more nuanced approach and suggest that instead, programmers and spec authors should consider fault handling first, and expected operation second. It's very simple to follow the expected path, when you have to handle forks in the road it does get exponentially more complicated. Eventually, you always need a "defer to human" option. It would be prohibitively difficult (expensive) to defer to a human every time, or even 1% of the time, but that option needs to be available.