| Author | Message | ||
     Luigi Buglione |
Hi All, if somebody could be interested, an update for a white paper entitled "Some Thoughts on Productivity in ICT Projects" is available at: http://www.geocities.com/lbu_measure/fpa/fsm-prod-120e.pdf Best regards Luigi Buglione, Ph.D., CSMS Level 3 SEMQ: www.geocities.com/lbu_measure/ MIS 3/e: www.francoangeli.it/Ricerca/Scheda_Libro.asp?ID=8054 Luigi Buglione, Ph.D., CSMS Level 3 SEMQ: http://www.geocities.com/lbu_measure MIS 3/e: www.francoangeli.it/Ricerca/Scheda_Libro.asp?ID=8054 | ||
| Patrick Hoornaert |
Dear Luigi, Thanks a lot for sharing your thoughts on Productivity Management by means of your exhaustive but concise and well-founded White Paper. Since this research appears to have been sizzling in your brain for quite a while, I was already familiar with your framing of the subject, so I paid special attention to sections 4, 6 and 7 (respectively WBS-view, 3 Proposals and Impacts). What I retained from the other chapters, is that you disapprove of VAF or any other adjustment factors, as a futile and erroneous attempt to collapse multiple dimensions into a single metric. Your proposal #3 calculating with PSU's will - in my opinion - be very hard to relate to the final product size. As the granularity of a Tasks (in IFPUG addressed by the definition of "Elementary Process") can vary greatly, I also doubt it makes sense for proper benchmarking. Your proposal #2 really is the ideal, but almost unachievable, as is adds Quality, Technical (and why not Organisational) Point Counting in the equation (each requiring their proper sizing method). Therefore #1 is actually the most realistic proposal. Comparison of Productivity versus "Functional Productivity" seems indeed a relevant exercise. However even in this simple scenario, it will be a true challenge to convince Project Managers to spit F vs. NF effort. Adding an attribute to project tasks should suffice to enable aggregation, and the classification you put forward in section 3.4 should aid proper triage. Last but not least benchmarking organisations - such as ISBSG - should take this additional information into account, otherwise service providers will only have their own data to challenge their expert/micro estimations against. Patrick Hoornaert Atos Origin Belgium | ||
| Luigi Buglione |
Dear Patrick, nice to "read" you! Thanks for your comments. I agree with you that proposal #1 is the more feasible for most organizations than the other two (even if - referring to proposal#2 at IFPUG's ISMA2008 in september a draft for 'Technical Points' was announced to be presented at the audience) and it would be great if - little by little - more and more companies would start to adopt it. IMHO to (at least, roughly) split the overall effort into F vs NF cannot be a choice, but a urgent need. Because - as many of us experiment daily - the large part of ICT contracts still use the 'Cost per FP' as the solely economic parameter for paying the work done by a Provider for a software system (product). And the 'boundary' for a project is the overall project effort, as calculated in the nominal 'productivity' figure. Now I've added a more visual example in Section 2 about a glass of wine for explaining it. What about the empty part of the glass (that would be the NF effort within a project)? You're producing effort for the whole volume of the glass within the scope of the glass itself, but you're paying only the volume filled by wine. There must be a proportionality between the real object observed and its representation. Otherwise the 'representation' is not sufficiently granular to be profitably used. Another way to look at this issue is in this position paper appeared on the last 'Software Measurement News' (http://ivs.cs.uni-magdeburg.de/sw-eng/us/giak/SMN-08-1.htm, click on the bottom link for a free download of the journal issue). About ISBSG, yes, the questionnaires should include also a rough split of the overall effort at least into F vs NF. We discussed this point yet last year and will continue to argue on this need for next reports. If somebody else can provide his/her feedback on the wp, I'd be really glad to receive them! PS: I apologize but unfortunately the pdf file for the wp is uploaded on a free website, that has a maximum data transfer usage limit. So, it too much people try to access at the same time to the website (page or file), an error appears. But retrying after a bit, the website and its content will be available as usual. Best regards Luigi Buglione, Ph.D., CSMS Level 3 SEMQ: www.geocities.com/lbu_measure/ MIS 3/e: www.francoangeli.it/Ricerca/Scheda_Libro.asp?ID=8054 Luigi Buglione, Ph.D., CSMS Level 3 SEMQ: http://www.geocities.com/lbu_measure MIS 3/e: www.francoangeli.it/Ricerca/Scheda_Libro.asp?ID=8054 | ||
| Mauricio Aguiar |
Luigi, Even though I generally like to read your posts there is something I have wanted to say for some time now. There is a flaw in your reasoning about the relationship between functional size and effort, as expressed in sentences like “You're producing effort for the whole volume of the glass within the scope of the glass itself, but you're paying only the volume filled by wine”. This does not happen as described and, if true, would never be accepted and used by the IT industry. In most projects, total effort holds a fairly strong correlation with functional size. That is what made the “price per function point” business practice possible at all. What one actually does is: 1. Estimate/measure the functional size of the project using IFPUG Function Points – the industry “de facto” standard 2. Use some method/tool (COCOMO II, SLIM, SEER-SEM, ad-hoc, etc.) to estimate/obtain the TOTAL effort of the project (not only the functional part of it) based on historical data and project characteristics 3. Apply the knowledge obtained in step 2 to establish price(s) per FP or productivity value(s) in hours per FP So, the “but you're paying only the volume filled by wine” part of your sentence does not reflect actual practice. One does in fact pay for the whole thing. Hope this helps to clarify the discussion. Regards, Mauricio Aguiar IFPUG Immediate Past President | ||
| Luigi Buglione |
Hi Mauricio, thanks for sharing your comments. Unfortunately the 'Cost per FP' and a 'nominal' productivity using FP are quite spread figures used in ICT contracts without often other complementary metrics for the NF part (at least in the Italian experience, but not only limited in this context). Anyway the estimation method (BTW, in your three points it seems to me you're assuming a constant proportionality between the F-NF parts, since you could estimate the overall 'project' effort by the regression analysis done on the functional product 'size', right?), the point stressed by the glass of wine was about the proportionality among the parts of a project with its related cost (and selling price). So, my question is: which sense has to put in relation units referred to different entities? In the 'cost per FP', the upper part of the formula refers to the 'project' entity, while FP are a product unit (only covering the F part); in the 'nominal' productivity formula, the lower part of the formula is the overall 'project' effort while in the upper part is a (partial) product-level unit of measure...could it be more realistic to compare 'apples with apples' and 'oranges with oranges' instead of mixing parts? (BTW, could you give me an opinion about the position paper on 'Software Measurement News'?). Of course, taking care of some measure more will cost a bit more but will assure a better representation at all from several viewpoints. That's why the origin of the wp with the goal to stimulate a debate and a discussion for eventual misuses of measures within the project management process. Best regards, Luigi Buglione, Ph.D., CSMS Level 3 SEMQ: www.geocities.com/lbu_measure/ MIS 3/e: www.francoangeli.it/Ricerca/Scheda_Libro.asp?ID=8054 Luigi Buglione, Ph.D., CSMS Level 3 SEMQ: http://www.geocities.com/lbu_measure MIS 3/e: www.francoangeli.it/Ricerca/Scheda_Libro.asp?ID=8054 | ||
| Mauricio Aguiar |
Luigi, Thank you for the reply. Some additional comments: 1) The variance in productivity and price per FP can be reduced by classifying projects into properly defined categories and having different productivities/prices for each category. Analysis of Variance can help in doing this. 2) Constant proportionality between the F-NF parts is a reasonable assumption provided project categories are properly defined (within ranges where the F-NF effort relationship is approximately linear). Also, notice that models such as COCOMO II and the others I mentioned use cost drivers to account for non-functional project characteristics. 3) Your sentence was, “You're producing effort for the whole volume of the glass within the scope of the glass itself, but you're paying only the volume filled by wine” and that was the object of my comment. If you intended to say something else, of course that is now understood. I apologize if I did not catch your intended meaning. 4) As for your third paragraph, of course it would be better to have a direct measure of the non-functional aspects of projects. However, the current industry practice – to use correlations & relationships among variables to produce estimates – is also a common and accepted practice in the social sciences. We will have to do with what we have until we find a better (probably not perfect) solution. There is no “silver bullet” as we all know. I will be honored to provide you with an opinion on your paper as soon as I can. I believe you are doing good work – I just wanted to clarify some points that seemed overrated/overstated in the discussion. Regards, Mauricio. | ||
| Luigi Buglione |
Mauricio, about your counterpoints, I mostly agree on them (it's a question of sharing assumption, of course!). But I do not totally agree on the assumptions provided for #2, because some NF project characteristics can be managed for 'adjusting' the overall effort (e.g. the analysts experience), but in many cases other ones (e.g. documentation) may have a stronger impact on the project WBS, producing a relevant effort that would be made lower and lower than the real one (verified at the end of the project by task) when included only in terms of a multiplier in an exponential regression analysis formula. In any case, thanks for sharing again your thoughts and viewpoint to me: I'll wait for your comments on the position paper(muito obrigado)! Best regards, Luigi Luigi Buglione, Ph.D., CSMS Level 3 SEMQ: http://www.geocities.com/lbu_measure MIS 3/e: www.francoangeli.it/Ricerca/Scheda_Libro.asp?ID=8054 | ||
| Mauricio Aguiar |
Luigi, As you said, this involves making assumptions – since we are not dealing with a particular case – so our discussion could go on forever. Sadly enough, it seems that both of us are working under time constraints. It is true that documentation or, more generally, the level of ceremony will impact effort. Therefore projects using processes with different levels of ceremony should be placed in different categories for the purpose of estimation – keeping apples with apples and oranges with oranges as you said. But this is just one of many details. In the real world, some "manual adjustments" (and negotiations!) will always be necessary. Remember George Box - “All models are wrong, some are useful”. I will get back to you on the position paper as soon as my “real job” allows me. Thanks for this interesting discussion. Regards, Mauricio. | ||
| Syed Raashid Ahmed |
Gentlemen, Reading through the thread, allow me to get clarified on a fundamental question... Do all of the 'Productivity' estimates go up in vapours, if 'Personal / Psychological factors' (like sadness, extreme joy, anger, frustration)affecting the performance of a FTE is not considered? I remember having read an article by Capers Jones on 'What makes up Productivity'....but unfortunately did not archive it....Kindly guide me on how to wrap in these 'Psychological' factors to calculate the 'Productivity'.. Syed Raashid | ||
| Patrick Hoornaert |
Dear Syed Raashid, I still see a fundamenal difference between "psychological" factors and quantification of Non-Functional (NF) size = Quality + Technical (+ Organisational). Unless we employ robots instead of humans, we can't control the psyche (let's say we can address this issue by recruiting capable and mentally stable people we can "count" on). Quality, Technical, Organisational Requirements can be scoped, analysed and designed, however they are even more intangible compared to FPs. In this case having an accurate WBS + effort estimation for all NF project activities seems sufficient to me, so I would like to align with Luigi's Proposal # 1. Correct me if I am wrong, but I don't directly see a real contraction between that proposition and Mauricio's arguments. Patrick Hoornaert Atos Origin Belgium | ||
| Syed Raashid Ahmed |
Patrick, You are right. While there is a faint corelation to Mauricio's and Luigi's topics, the issue of 'Psychological' factors was running on my mind concurrently. I would like to plug in 'Nation indicators' (productivity factors of FTE for most countries, that were derived after intensive research). Will share it to the forum, when I lay my hands on them}.... However, your judgement of discounting the intangible factors is logical...Syed Raashid | ||
| Patrick Hoornaert |
Dear Syed Raashid, The establishment of "nation" indicators is an altogether different but nevertheless extremely interesting discussion. (Why don't you start a new discussion thread?) Thus outsourcing becomes a matter of cost and productivity combined. However when this factor comes into play, I fear competition will overtake peer2peer comparison as the main driver behind the industry benchmarks. Remember the catastrophic results of the Great Leap Forward during 1958 in P.R.China ... inflation of production-related metrics will more then likely catapult us into chaos. Patrick Hoornaert Atos Origin Belgium |
