As said before, why do it right when you can do it wrong.
There are so many "smart" solutions floating around such as the above, RS232 without a 232 transciever, programming chips with "fast" homemade programmers (Fast because minimum pulse width per data sheet not met) and RC reset. All these "solutions" end up costing more than the right solution such as using supervisors, watchdog chips or chips with internal watchdog, MAX232 or equivalent transcievers, ISP or a programmer from a reputable manufacturer.

Another aspect is: can the use of "smart" solutions be guaranteed to work some time from now?. Examples kitchen table programmers fail to program for full data retention and can not work when the Gatesians make a new release. The "no 232 transceiver needed" fail when you use another PC. The 555 "watchdog" makes one false trigger. The RC reset screw up your boot vector.

If you really want to be smart, use your smarts to make innovative products with innovative features, not to "prove" that the standard solution is wrong.

Erik