The thesis project that I proposed, failed outright. It was quite ambitious, and was a sort of win-or-lose proposition. It took me 3 years to reach the point where it was evident that I wouldn't observe the signal needed to complete the work. My own abilities were doubtlessly a factor.
A lab mate of mine, who had already accepted a professorship at another university, gave me one of his start-up projects. By agreement, as I built the experiment, I sent him copies of my drawings, circuit diagrams, and source code, letting him reproduce my work almost concurrently.
That became my thesis, and the technical work became my ticket to industry.
Often, dissertations bear little resemblance to the proposed project.
This fact is telling of the level of ambition and innovative thinking that is considered to be acceptable in students proposing their thesis. A "safe thesis" is good for the business of education not for the advancement of human knowledge. If they want safe thesis level work how about a "by reproduction" track so we still get "cheap PhDs" and we get an actual benefit to science?
You may be speaking across purposes, just because "thesis proposal" means different things in different places. In some institutions, it's a proposal for a whole thesis project, from experiment design through conducting experiments to drawing conclusions and writing up. In other institutions/countries, it's done as the project is drawing to a close, with experiments completed and data gathered: the candidate then proposes the thesis that this data supports, which they will write up and defend. The latter is more of a formality, because it happens at a different step in the process. The point where innovation and ambition incur risk has mostly passed by then.
In my case, the thesis proposal was at the beginning of an open ended and fairly long term project. You were supposed to conduct a literature review, show that the project had a good chance of working, and that you had a good enough grasp of the subject matter to handle the work. It was also a point where you could opt to finish with a Master's instead, having completed your coursework.
There was high attrition in between this step and getting the PhD.
The other end, where the student is basically done except for the writing, is less risky. One reason is that professors are strongly discouraged from letting a student fail a thesis. When I was ready to defend my thesis, I had to get permission from my committee to do so, and their signature indicated that they considered my work to be defensible. So the defense was in fact a formality. Failures at that point, that I know about, had to do with students getting caught fudging their work.
A lab mate of mine, who had already accepted a professorship at another university, gave me one of his start-up projects. By agreement, as I built the experiment, I sent him copies of my drawings, circuit diagrams, and source code, letting him reproduce my work almost concurrently.
That became my thesis, and the technical work became my ticket to industry.
Often, dissertations bear little resemblance to the proposed project.