The problem of counting nodal curves on algebraic surfaces has been studied since the nineteenth century. On the projective surface, it asks how many curves defined by homogeneous degree d polynomials have only nodes as singularities and pass through points in general position. On $K3$ surfaces, the number of rational nodal curves was predicted by the famous Yau-Zaslow formula. Goettsche conjectured that for sufficiently ample line bundles $L$ on algebraic surfaces, the numbers of nodal curves in |L| are given by universal polynomials in four topological numbers. Furthermore, based on the Yau-Zaslow formula he gave a conjectural generating function in terms of quasi -modular forms. The formula is consistent with many existing results on projective surface, $K3$, and curves with at most 8 nodes on general surfaces. In this talk, I will discuss how degeneration methods can be applied to count nodal curves and sketch my proof of Goettsche's conjecture.