I have seen some information about CMM, I just felt some empty, and it was very complicated, and I couldn't be implemented in China's software companies. However, for so many years, I have experienced a lot of projects, and many projects have also been leading, and I found new understanding of CMM. CMM's key problem domain is a lot of failures and many successful examples, maybe it is complicated, requires high, but if we really understand these key issues, then we should be from one project I came out in failure, let the project succeed in the spirit of the scientific spirit, let the programmer no longer 'kick', let them live like ordinary people, rest, vacation, 行, don't have to have a big pressure now. Speaking of actual, now a project, a three-year project for a lot of reasons, really can be tagged with failure. From the start of the company, the company is trying to use this project to enter a certain user of a certain industry, it is destined to the route and failure of the project. Since the company hopes to take this project to make a martial stone, the high-level is swaying on the attitude towards this project, and the beginning is ignored. I hope to invest the least people, do the least functional point, look forward to the backward project. When the user is no longer interested in the latter large-scale project, this project suddenly discovered that this project became a possible potential product, so it began to pay attention to this project. However, due to the preliminary attitude, the functional demand of this project from the beginning is blurred, and the extension of the user's demand is constantly extension. At the same time, the user requirements are actually evaluated due to competition. This will add more chaos to this item that has been chaotic. Users have proposed more than 400 function points, and for us for only three months. If the company's high-level can be carefully evaluated, it can be seen that the risk of failure is very large. The demand given by the user is very simple, and for the reasons for competition, there is no opportunity to investigate and investigate the needs of users; It has been in the process of continuous development; there is no more attention to the data volume and the complexity of the data involved in the system, resulting in almost unbearable background processing times; there is no test plan; there is no stable between the project team and customers. Effective communication mechanism. If you recognize these risks, we have the opportunity to remedy. But the high level did not do this, the result is that everyone will continue to overtime for a past year, but the bosses are still very unsatisfactory because the user is not satisfied. If someone taking out any of the books that introduce CMM, then our next field may not be so tragic. The key issues in the CMM2 level will point out how dangerous, demand management, no, risk management, no, plan, no, nothing, only programmers are desperately encoded, edit those destined to be overturned Code, know that it is necessary to fail, but can only continue to walk, so tragic ending, the programs of the defect.