Economic value of open source


Published on

My slides for the talk "the economic value of open source software" given at SvSCon 2012 in Bozen

Economic value of open source

  1. 1. The Economic value of Free/Libre Open Source Software Carlo Daffara European Working Group on Libre Software CloudWeaversSfsConf 2012
  2. 2. The Economic value of Free/Libre Open Source Software (for Europe) Carlo Daffara European Working Group on Libre Software Conecta ResearchSfsConf 2012
  3. 3. “The GPL effectively prevents profit-making firms from using any of thecode since all derivative products must also bedistributed under the GPL license” (Evans, D.,in “Government policy toward open sourcesoftware”, R.W.Hahn, editor, AEI-BrookingsJCRS)SfsConf 2012
  4. 4. “[..] the aim of free software is not to enable ahealthy business on software but rather tomake it even impossible to make anyincome on software as a commercialproduc t.” (Thomas Lutz, Microsoftrepresentative at Tunis WSIS, 2005)SfsConf 2012
  5. 5. “Open-source software is deliberatelydeveloped outside of market mechanisms... thenonmarket coordination mechanism fails tocontribute to the creation of value indevelopment, as opposed to the commercialsoftware market. [It] does not generate profit,income, jobs or taxes … In the end, thedeveloped software cannot be used togenerate profit.” (Kooths S., Lagenfurth M.“Open Source-Software: An EconomicAssessment” University of Muenster, MuensterInstitute for Computational Economics)SfsConf 2012
  6. 6. “[Open Source] ... suppresses qualitycompetition between OS firms and restrictstheir output much as an agreement to suppresscompetition on quality would. .. We find thatthe first-best solution in our model is to taxOS firms and grant tax breaks to[proprietar y sw] firms .” (Engelhardt,Maurer, 2010 Goldman School of PublicPolicy)SfsConf 2012
  7. 7. “Rail travel at high speed is notpossible because passengers, unable tobreathe, would die of asphyxia.” Dr. DionysusLardner (1793-1859), Professor of NaturalPhilosophy and Astronomy at UniversityCollege, London.“Heavier-than-air flying machines areimpossible.” Lord Kelvin (1824-1907), ca.1895, British mathematician and physicistSfsConf 2012
  8. 8. Measuring value is complex. A bad way ofdoing it: “...First we listed the major opensource products. Then we looked at thecommercial equivalents. Next we looked at theaverage cost of both the open source productsand the commercial products, giving us a netcommercial cost. We then multiplied the netcost of the commercial product by our opensource shipping estimates.” (Jim Johnson,Standish group)SfsConf 2012
  9. 9. Some groups measured the total revenues ofFLOSS firms; so Pierre Audoin Consultantsfound a total market of 8B€ in 2008.Unfortunately, HP alone made 2.5B$ in Linux-related consulting in 2003, while IBM made4.5B$ in OSS-related revenues in 2005 (as anexample, the OSS PBX market alone is 1.2B$alone.)In fact, the majority of FLOSS-relatedrevenues are not made by FLOSS companies atall.And the software market is not that easy todefine as well.SfsConf 2012
  10. 10. SfsConf 2012
  11. 11. SfsConf 2012
  12. 12. This provides us with an overall IT spending estimatefor Europe: 492B€approximately 24% is hardwaresoftware and services market: 374B€software market alone: 244B€SfsConf 2012
  13. 13. How much FLOSS is inside the average codebase?SfsConf 2012
  14. 14. SfsConf 2012
  15. 15. How much FLOSS is inside the average codebase?● On average, 30% of implemented functionalities is based on reused OSS code (Sojer M., Henkel J. “Code reuse in Open Source Software Development”)● Gartner reported that among the surveyed customers, 26% of the code deployed was Open Source● The Koders survey in 2010 found that 44% of all code was Open SourceSfsConf 2012
  16. 16. ● Black Duck analysis of large code projects (avg. 700MB of code): 22% is FLOSS, up to 80% of new development is avoided through FLOSS● “sampling continues to find that between 30% and 70% of code submitted is .. in the form of OSS components and commercial libraries” (Veracode, “State of Software Security Report volume 3”, 2011)● Sampling shows that FLOSS use increases with time → average usage for last 5 years: 35%SfsConf 2012
  17. 17. SfsConf 2012
  18. 18. What value does FLOSS reuse brings in?(Abts, Boehm, Bailey Clark “Empiricalobservations on COTS software integrationeffort based on the initial COCOTS calibrationdatabase”)SfsConf 2012
  19. 19. 35% of code reuse provide a reduction inactual costs of 31%: 75B€/yearSfsConf 2012
  20. 20. “Figures suppor t the idea that FOSSsolu tions are more innovative thanproprietar y ones: indeed, in all the threedimensions, experts’ evaluations are higher forFOSS than for proprietary software. … FOSSsoftware not only show different levels ofinnovativity, but, as far as, new to the worldproducts are concerned, they are also shapedby different innovation processes: radicalinnovation in the FOSS vs. incrementalinnovation in proprietary field.” (Rossi,Lorenzi, “Innovativeness of Free/Open Sourcesolutions”)SfsConf 2012
  21. 21. "The growing rate, or the number of functionsadded, was greater in the open source projectsthan in the closed source projects. Thisindicates that the OSS approach may be ableto provide more features over time than byusing the closed source approach. (Paulson,Succi, Eberlein “An Empirical Study of OpenSource and Closed Source SoftwareProducts”)SfsConf 2012
  22. 22. "Findings indicate that community Open Sourceapplications show a slower growth ofmaintenance effort over time.” (Capra,Francalanci, Merlo “The Economics ofCommunity Open Source Software Projects:An Empirical Analysis of MaintenanceEffort”)“The fourth law of software evolution,implying constant incremental effort, might beviolated (Koch “Evolution of Open SourceSoftware Systems – A Large-ScaleInvestigation”)SfsConf 2012
  23. 23. (Mohagheghi, Conradi, Killi and Schwarz “AnEmpirical Study of Software Reuse vs. Defect-Density and Stability”)SfsConf 2012
  24. 24. Project failure data:● Jones :“the cancellation rate for applications in the 10,000 function point size range is about 31%. The average cost for these cancelled projects is about $35,000,000”● Standish group, 2009: 24% of projects are canceled before deployment● Sauer & Cuthbertson, in an Oxford university survey of 2003: 10%● Dynamic Markets Limited: 25%+ of all software and services projects are canceled before completionSfsConf 2012
  25. 25. Project success data:Size People Time success rate< 750K$ 6 6 55%750K to 1.5M 12 9 33%1.5M to 3M 25 12 25%3 to 6 40 18 15%6 to 10 250+ 24+ 8%>10M 500+ 36+ 0%SfsConf 2012
  26. 26. By reducing effort, staffing and duration the35% code reuse introduces a reduction onthese parameters of 10% → a reduction in thefailure rate of 2% → 4.9B€/yearSfsConf 2012
  27. 27. “While IBM initially contributed software that wasvalued at 40M$, external contributors to the projectcreated software representing a value of roughly1.7B$ over the examined period.” (Spaeth,Stuermer, von Krogh “Enabling knowledge creationthrough outsiders: towards a push model of openinnovation”)SfsConf 2012
  28. 28. ● OSS maintenance effort is substantially lower than the average (Capra E., Francalanci C., Merlo F., “The Economics of Community Open Source Software Projects: An Empirical Analysis of Maintenance Effort”)● Using a model by Jones and Bonsignour, traditional code does have a cost of 2000$ per function point, while code shared or developed using best practices costs 1200$ per FP● the shared code in a reused OSS project introduces an additional reduction in maintenance and dev. effort of 14%SfsConf 2012
  29. 29. 14% reduction in maintenance anddevelopment costs → 34B€/yearSfsConf 2012
  30. 30. Deshpande, Riehle “The Total Growth of Open Source”SfsConf 2012
  31. 31. Source: Dirk Riehle, “The open source big bang”SfsConf 2012
  32. 32. Total value of OSS reuse per year: 114B€SfsConf 2012
  33. 33. ● What is the second order effect?● We know that most of these savings are reinvested:SfsConf 2012
  34. 34. SfsConf 2012
  35. 35. ● “The principal results from this econometric analysis are: 1) the measured output contribution of computerization in the short- run are approximately equal to computer capital costs, 2) the measured long-run contributions of computerization are significantly above computer capital costs (a factor of five or more in point estimates), and 3) that the estimated contributions steadily increase as we move from short to long differences. (“Computing productivity: firm-level evidence”, Erik Brynjolfsson, Lorin M. Hitt; Review of Economics and Statistics, November, 2003 )SfsConf 2012
  36. 36. With a 3 years investment discount period,model based on linear growth in efficiency dueto reinvestment → 342B€/yearSfsConf 2012
  37. 37. Revenue per employee rating (FLOSS firms vs. Industry average) Computer Equipment 182% Software consultancy and supply 427% Services (excl. software cons. and supply) 211% Manufacturing (excl. computer equip.) 136% Other 204% ALL: 221% Source: MERITSfsConf 2012
  38. 38. Revenue ratio: FLOSS firms vs. Industry average (FLOSS firms vs. Industry average) Computer Equipment 1115% Software consultancy and supply 262% Services (excl. software cons. and supply) 177% Manufacturing (excl. computer equip.) 4501% Other 1045% ALL: 758% Source: MERITSfsConf 2012
  39. 39. SfsConf 2012
  40. 40. Source: Venice International University TEDIS studySfsConf 2012
  41. 41. DrupalCon 2010, CopenhagenSfsConf 2012
  42. 42. ● Does the increase in efficiency reduces local revenues for incumbents?● For commercial products (that is, proprietary products that embed OSS): the producer reduces its production costs with no other impact on the business itself, so it can either increase its margins or pass some of the savings to the customer.● For internally developed products, the savings are direct, with no other effects on the external market● OSS reinvestment are mainly localSfsConf 2012
  43. 43. With proprietary software, 86% of SW spending goes outside of Europe-and reduces local company margins Ecosystem Revenues compared with MS revenues by partner type Product- Services- Retail Logistics Logistics-Oriented Oriented Oriented Value-Added Partner Partner (e.g., LargeMicrosoft Partner (e.g., Large Partner (e.g., Partner (e.g., (e.g., VAR) Retail Electronics Account Reseller) ISV, IHV) SI, Hoster) Store) $1 $4.09 $2.44 $2.30 $2.70 $2.93 1 24% 40.9% 43.5% 37% 34%Source: Partner Opportunity in the Microsoft Ecosystem, IDC 2011; analysis by Daffara SfsConf 2012
  44. 44. ● Still missing in the model: “pull” adoption● More difficult to assess – huge variability in outcomes● In desktops, with successful migrations, TCO reduction ranges from 10% to 25% (typical) up to 50% (for high-uniformity environments)● Movement towards web-apps is changing substantially the economics of moving from/to a different platform, reducing the transition cost → requires a move away from “pure substitution” to “reengineering for the future”SfsConf 2012
  45. 45. From: "The future of computing: indispensable or unsustainable?" Royal Academy of Engineering, 2011SfsConf 2012
  46. 46. SfsConf 2012
  47. 47. ● Innovation from end-users:SfsConf 2012
  48. 48. ● Non-code contributions: value deriving from anything that is not directly compilable● “[non-code] outside contributions are signicant. Open Cascade estimates that they represent about 20 % of the value of the software. Matra Datavision had to inject approximately 2M€ per year to continue to develop its tools. In 2000, the company limited the costs to 1.2 million.” (Jullien, Clement- Fontaine, Dalle “New Economic Models, New Software Industry Economy”)SfsConf 2012
  49. 49. Thanks! Carlo Daffara Twitter: @cdaffaraSfsConf 2012
  50. 50. SfsConf 2012
  51. 51. “A study carried out between January andJune 2010 shows that despite the desiredaffirmative action for open source products, inalmos t half (47.5%) of the tendersthere is s till a preference for closedsource vendors or products. Thispreference inevitably results in not givingvendors of FLOSS a fair chance to win the bid.(Mathieu Paapst, Center for Law and IT,University of Groningen, the Netherlands)SfsConf 2012
  52. 52. ● The vendor must employ MS certified employees.● Asking for an operating system to be used together with the Microsoft Campus Agreement.● If your bid is open source you should give extra guarantees concerning the stability of the open source community.● Not allowing “zero-price” licenses.● Demanding that offered applications must be certified by Microsoft, are Oracle 10 compliant and using the official Microsoft style guide as much as possible.SfsConf 2012
  53. 53. SfsConf 2012
  54. 54. SfsConf 2012
  1. A particular slide catching your eye?

    Clipping is a handy way to collect important slides you want to go back to later.