1: \begin{abstract}

2: We provide a new proof of convergence to motion by mean curvature (MMC)

3: for the Merriman-Bence-Osher (MBO) thresholding algorithm. The proof

4: is elementary and does not rely on maximum principle for the scheme. 

5: The strategy is to construct a natural ansatz of the solution and then

6: estimate the error. The proof thus also provides a convergence rate.

7: Only some weak integrability assumptions of the heat 

8: kernel, but not its positivity, is used. Currently the result is proved

9: in the case when smooth and classical solution of MMC exists.

10: \end{abstract}

11: