http://www.oochina.org/articles/200410/379/index.jsp
From: Dr. NECO Electronic Journal
(1) Why do you need a peer review?
For the first reason:
The reason why technical work needs to be accepted is like a pencil, please remember:
Everyone is wrong!
The second reason why we are inseparable from peer review is:
Although people are good at finding some of their own mistakes, a large number of various mistakes are still easy to escape the eyes of the creator relative to anyone else. This is called thinking.
Third cause:
Many defects are introduced in the early stage. As shown below :
Source: Applied Software Measurement, CAPERS JONES
The thinep the defect found, the higher the correction cost. As shown below:
Source: Boehm, IBM, 1981
And the cause of the number of defects, each of the defects that enter the next step may cause multiple defects in the next step, resulting in a sharp increase in the cost of the displete.
(2) Which one is more cost-effective?
Of course, it is more cost-effective. It is impossible to replace the test, and the purpose of doing peer review is to reduce the defects of leakage to test phases. From the economic perspective, many defects are introduced in the early stage, the sooner the defect, the more economical, and some early stages can be reviewed, but it is impossible to test. It took a lot of time in peer review, but saved time in the test. Test and peer review compared to a large amount of time, testing cannot find some specific types of defects (such as violation programming specifications).
(3) Will the peer review delay the cycle of the project?
The peer reviews have reduced the workload of later modifications and rework, so it is actually shortened the cycle of the project.
(4) Do you have contributed to the ability to enhance developers?
Even in the case where the technical product does not have errors, the peer review activities have the necessary education for the participant's educational value. If there is no peer review activity within the organization, the communication and communication within large technical institutions becomes slow and unreliable. Experience, a programmer who regularly participates in the review and the work of the work, and the accumulation speed of experience is three times the programmer who works alone!
The review activities also assume the heavy responsibility of a wide range of exchange technology information, which supplements or replaces the role of formal written exchange.
(5) Why is the peer review activities that are different from our management departments undergoing budget and progress review?
When we talk about the right to review, we are talking about the contents of the project material, rather than these material related content, such as schedule and budget. The problem that needs to be answered in the peer review:
Will this product work properly as designed?
Of course, if the cost of the product is too high or from the progress, then it certainly can't talk properly;
Progress and cost are issues on management, not technical issues. Programming progress and cost issues may also include technical factors, and things are almost like this. Peer review provides cost, advancement, and other related information concerning the management department. There is no information provided by a reliable peer review, and the management department is like building a building on the sand sand. The work is lacking firm foundation.
(6) Is there a difference between the progress report we use now with peer review activities?
Peer reviews are similar and progress reports. It can be considered that the peer review is a mechanism for a certain reporting progress? The institutional balance measures to ensure progress information reliability is present in this system.
There is a very large mechanism for reporting "progress" in the program development project. Some mechanisms are responsible for reporting a specific project segment has spent how much time, while others report the number of lines that have been completed and the number of pages of the document. Then, depending on the progress table and resource allocation, these "progress" parameters will be made, prompting the management department to pay attention to potential issues. If the progress reported by these mechanisms is true, they will be very effective. However, if there is no such agency balance measures, we will not guarantee that the completed lines have been completed, and the programming industry has long been used to this situation: a project's "progress" report indicates that the project has been completed. 99%, and in fact this project is less than 10%; why? Because the code of thousands of rows may be wrong, it may be wrong, or only one or two small mistakes ?? But covering this two errors may cost 2000 workload.
Therefore, it is necessary to facilitate peer review mechanisms to ensure that the progress report chart has practical significance. If there is no quality supervision and control measures, use quantitative monitoring measures may be more dangerous than do not take measures.
(7) How much role in peer reviews?
For large systems that save exact records, a complete set of peer review systems can reduce the error in each test phase to 90%. In this way, even in the case of comprehensively considering the cost of peer review activities, peer review activities will also reduce test costs by 50% to 80%.
The peer review activities will also bring a decline in maintenance costs. If a programmer accepts a complete peer review in the development phase, and its changes in the maintenance process have also been reviewed, then its maintenance cost will be reduced by 80 %.
In the long run, peer review activities will also significantly improve employee ability, which is difficult to explain with numbers. In general, we will find that the employee's morale is constantly rising. They become more dedicated. The project's maintenance costs have been reduced, and the project valuation and progress plan becomes more reliable. The management department has obtained the role of ensuring the success of the project. More affirmation, at the same time, peer review activities also enable management to make correct evaluation of employees' individual and overall ability.
(8) The peer evaluation process seems to have the trend of universal combat existing in the technicians, will it cause this consequence?
On the contrary, no matter whether people prefer or, the review process will force everyone to work in an open environment. Once people know that their work should be accepted, they will make their work in public to be supervised.
(9) The peer review sounds good, how should we implement?
Peer review is a relatively biased method, and it is necessary to pay attention to many details. As the saying goes: Details determine success or failure! Listening to experts before the implementation of peers, the mature experience of the industry is not going to take a lot of detours, so that work is effective, causing leaders to pay enough attention and firm support.
