Goal 1 is clearly subjective in the sense that it strongly depends on a given observer and the way he or she interprets the (possibly informal) specifications. What ``looks right" to an observer from one (sub)culture may ``look wrong" to an observer from another (sub)culture (or another time).
Goal 2a depends on the nature of the computer running the algorithm. In what follows, this dependency will be ignored. Ultimately this is justified by the above-mentioned invariance theorem.
Like Goal 1, Goal 2b depends on the observer. But in a sense, Goal 2b is less subjective than Goal 1. This is because intelligent human observers, in principle, can learn to compute anything a digital computer can compute (the reverse is a matter of controversy). In particular, a short algorithm running on a conventional digital machine can be quickly taught to an intelligent human being. Note that if the human observer was another universal computer, then we could immediately apply the invariance theorem, thus (ultimately) eliminating subjectivity from Goal 2b. Then the only remaining subjective aspect of low-complexity art would be that of Goal 1.