Program lifetime: a vital statistic for maintenance

Some widely held views about software maintenance expenditures are examined and it is argued that in the light of recent evidence the views must be modified. The implications for those wishing to monitor improvements in their maintenance process are discussed, and it is concluded that measurement of average program lifetime is likely to be a better indicator of maintenance improvements than overall maintenance costs. The author also contrasts data processing and embedded systems and suggests that the latter are likely to obey different rules. Embedded systems are therefore excluded from the conclusions drawn.<>