The author discusses the problematic issues of solving inverse tasks of analytical photogrammetry: triangulation, photo-triangulation, and self-calibration of images. It is noted that the said matters do not have an exact solution and can only be solved through mathematical statistics. In particular, the method of least squares is used. In the geometric interpretation, this technique consists in finding the minimum point of a positive definite quadratic form. Its topology can be significantly different depending on various factors that distort the solution. It is noted that the main factors are: poor conditioning of the normal equations system, inadequacy of the survey’s mathematical model to the real physical process, errors in the initial measurement data, presence of many local minima of the quadratic form, and lack of reliable information on the accuracy of the measurements. The individual impact of each factor on the topology of the minimized quadratic form is considered and clearly shown. The negative consequence of the quadratic shape surface distortion is considered and described. The main known ways of overcoming each of the considered defects of the minimized surface are explained