It is well-known that SL(n,Z) has Kazhdan's property (T) if n is at least 3. I will present a new proof of it: this proof does not appeal to ANY specialty of the ring Z (such as lattice structures or bounded generation). This proof may be applied in much broader context. For instance, we re-prove the celebrated theorem by Ershov and Jaikin-Zapirain (but with no estimates of Kazhdan constants), which has a corollary on the "unbounded rank expander problem." We can proceed even further.