3 boehm: Good morning, everybody. I finally figured out a way to get in. All I can suggest is to keep trying various combinations for refreshing and logging in.
4 chirag: Are there any homeworks due on monday?
 9 boehm: 4 Homework 6 is due Oct 15, in its revised form in which the baseline TPS effort is given as 60.5 person-months and $605K. Homework 5 will be due a week after ISD makes the COCOMO II.2000 tool available on SAL125
 
5 Anna: Do we have chat session this morning? I'm getting message that "there are too many people accessing the Web site at this time", and HTTP 403.9 access forbidden. Is it my company proxy or indeed there is a limit for the number of participants in the chat room?
 7 NScape: 5-No Anna there is no limit to the number of participants
 
8 Anna: Chat room works OK now, those messages were only temporary.
10 Anna: To do homework I've used software on CD ROM attached to the class textbook. It did not seem that there were any problems.
 14 boehm: 10 Anna-- you're lucky that you do not depend on campus computing facilites to load and run COCOMO II. We are still working with the campus computing people to enable on-campus people to do this on the SAL125 computers. They have extreemely tight security restrictions of which we were unaware.
 
11 boehm: Q: What are the "pairs" in "technologically efficient pairs" for production functions?
 12 boehm: 11 The pairs refer to pairs of input investment level and output value generated. "Technologically efficient" means that no higher level of output value can be generated at this level of investment input.
 
13 Anna: re: 11: In traditional production function the usual pair is capital and labor. Their mixture would give a specific output based on the technological characteristics of the process.
15 Anna: My recommendation is to load software from CD ROM onto any computer, it can be even 486 version and run it. Software is very efficient, takes very little memory and space and it works just fine.
16 boehm: Q: How do you go from computer-performance quantities like throughput to value quantities like sales and profits?
 19 boehm: 16 This is harder to do for completely new systems. It involves making estimates of your expected market size, market share, and profitability or market captialization (which reflects expectations of future profitability). It is easier for an extension of an existing system, where you have more information about these quantities and about the added value of adding various capabilities.
 
17 Anna: re 16: the best way is to translate everything into dollars or some other utility measure. That will provide common denominator to the different types of production factors 9and outcomes).
 21 boehm: 17, 18 Agree, subject to the qwualifications in 19.
 
18 Anna: IMpact of the throuput could be translated into dollar impact. And actually production function could provide the relationship between certain level of troughput and its impact on the sales and profit.
20 Anna: Another econometric approach to measurement of technical factors, such as throughout, on sales and profit could be probit analysis. Did anyone at CSE try to run some models using that econometric approach?
 22 boehm: 20 Profit analysis also gets into the very interesting and complex topic of pricing of products and services, and whether you are going for low-margin mass markets or high-margin niche markets. We haven't done much specifically here on this topic.
 
23 Anna: Agree. However estimating technology could provide insight into specific type of the relationships between dependent and independent variables. It also could provide insight into what variables are dependent and what are the drivers. So in a sense it could assist in determination of the nature of the relationships between variables.
24 boehm: Q: Why is Return on Investment (ROI) measured as (E-C)/E, where E is Effectiveness or Value Added, and C is Cost?
 26 boehm: 24 This formulation of ROI enables people and organizations to compare the attractiveness of alternative ways of investing their available resources. Projects with higher ROI are better investments of limited resources than projects with lower ROI.
 
25 Anna: Seems that systems dynamics could provide interesting insight into determination of the impact between different factors involved in determination of the software productivity, cost etc. Is CSE working on looking into application of this approach to ascertain environment for software development?
 27 boehm: 25 Ray Madachy and I (primarily Ray) are putting a book together on applications of systems dynamics to software engineering decision analysis. Systems dynamics provides valuable insights on things like the conditions under which inspections and adding people to projects have a net positive payoff.
 
28 Techsupport: /loind temp1
 30 boehm: 28 This is hard to do for software development, but can be done for software or hardware plug&play improvements, where plugging in a more efficient COTS component or processor chip provides an instant improvement in performance.
 
28 Techsupport: /loind temptest
28 boehm: Q: Can there be systems where the production function does not have an investment segment but starts with a high payoff?
29 Anna: That is highly sophisticated approach that recognizes that rarely, if ever, relationships between different variables are unidirectional. Therefore it might be quite useful in determination of them sociotechnological characteristics of highly inrterdependent environments, and software development seems to be one of those environments (highly interdependent)
 31 boehm: 29 Agree
 
32 Techsupport: /loind barry
32 Anna: I think that one of the huge advantages of your work is classification and categorization of the critical characteristics impacting environment for software productivity. Company can simply take your book at look at what are the characteristics of different factors that are beneficial and what are characteristics that are having negative impact on the productivity of software development.
33 Anna: That provides tremendous aid in uderstanding what is going on and what activities are impacting positively or negatively the productivity and effectiveness, including quality, of software development. I also think that your work to a large extent is portable to areas other than software development.
34 Anna: Application of production functions, instead of simple ratios, such as ROI, enhances ability to reflect nonsimplistic relationships between different "factors of production" and the software development outputs.
[Ref: 32,33]35 boehm: 32,33 Some of the work applies well to vaious classes of system engineering. We have a new project to develop and calibrate a systems engineering cost estimation model called COSYSMO. Boeing along with Lockheed Martin, Raytheon, SAIC, and TRW have been active in formulating it.
 36 Anna: 35: very intersting. What stage those activities are at? How soon do you expect to see results?
 
37 boehm: Q: How is the S-curve production function on pare 193 of SW Engr. Economics related to the results of stakeholder Easy WinWin negotiations?
 39 boehm: 37 Easy WinWin has a prioritization step in which stakeholders rate each win condition on a scale of 0 to 10 in terms of its business importance (correlated with value) and its degree of difficulty (correlated to cost). The results of the assessments are color-coded to indicate degrees of consensus on the ratings, and people can discuss and change their ratings to provide a best-consensus set of priorities for win conditions (features, levels of service, platform constraints, etc.)
 
38 Anna: It provides objective documentation of the relationships and therefore may assist in sometimes difficult negotiations. It de-emphasises emotionality, subjective assessment and personal feelings (and ego) involved any negotiations and it emphasises "FACTS OF LIFE" - that is the way how things are. Depending on our decision making make-up, we might benefit (and use) the results of the estimating and produciton functions in our decisions and trade-offs.
40 boehm: Q: What do you do if in-project changes invalidate assumptions, evaluations, or priorities that you have built into your initial project decision analysis (new technology, changing competitive market place, staff turnover, strategic partnership defaults, budget cuts, etc.)
 42 boehm: 40 You need to monitor the status of all of these factors and reconsider decisions if they look like they are leading you in the wrong direction. This may involve re-running your business-case analysis or components of it such as COCOMO II, based on the new values of cost and value factors. In some cases, they may lead you to cancel the project (for example, if you assumed that you would be first to market with a new product and you find halfway through the project that others have already captured most of the market share).
 
41 Bob Robinson: Are there ways that affiliates can download and have access to the latest models (including COQUALMO, COSYSMO, and other unreleased extensions) to evaluate for corporate (Boeing in my case) use in SPI efforts? Who is the best contact at USC to arrange access?
 43 boehm: 41 We have a discussion of this coming up in our USC COCOMO II project meeting at 10am today, to determine which of the emerging extensions are mature enough to release in beta-test form to our Affiliates, which include Boeing, by the time of the COCOMO Forum Oct 23-26. I'll let you know the results. COQUALMO is a possibility; COSYSMO has yet to achieve even a functional form with agreed-on definitions of parameters such as the size of the system being system-engineered.
 
44 Bob Robinson: · In adding product capabilities to intermediate prototypes in successive spiral iterations, are there quantitative methods to factor in the relative value of marginal software features in assessing cost and schedule risks? Do any of the USC model frameworks accept scalar (or weighted vector) inputs corresponding to capability benefits and other inputs corresponding to perceived risks, then mathematically correlates the input sets to produce a relatively objective, weighted output? Without totally precluding the human decision factors in assessing intangible risks, such a model would provide valuable insight and be a useful project risk assessment/mitigation tool.
 45 boehm: 44 The closest we have come to this is in determining the content of a system's core capability and the nature of the system's architecture which enables you to easily add or subtract features to meet your available schedule. We have a couple of recet tech reports on this that we will be posting on the sunset web site (LiGuo -- please post the two recent SAIV papers as tech reports on the web site).
 
46 boehm: Q; Under what conditions is a feature gold-plating?
 47 Anna: 46: When feature is not listed as critical in user's required features priority list, but it finds its included in functionality of the system that is being developed
48 boehm: 46 This is very situation-dependent. Speech understanding may be gold-plating for a simple query system, but not if you're blind. It also depends on the level of maturity of the technology and the degree of sophistication required (size of vocabulary, degree of recognizer training provided to understand people's special accents, noisiness of inputs, etc.)
 
49 boehm: Time to wrap up and go to the discussion session in OHE 136.