模型的唯一真正检验是其预测未来的能力,而未来总是不确定的。
The Erdős discrepancy problem fascinated me because it was so simple to state yet so deep in its implications.