Free software communities
Open source, or free software as it was originally called, has become in recent years one of the most talked about phenomena in the ICT world. This is remarkable, not only for the usual reasons that open source has been around for many years as a volunteer driven success story before being discovered by big business and now government, but also because it has largely developed on its own without the headline coverage and glare of international attention that it now receives.

This in turn makes it more attractive to governments and policy makers. Yet what is the special value of open source software, and how can it be harnessed? The Free/

Libre Open Source Software (FLOSS) study in 2002, a comprehensive study of developers and users, showed that the most important reason for developers to participate in open source communities was to learn new skills — ‘for free’. These skills are valuable, help developers get jobs and can help create and sustain small businesses. Meanwhile, the most important reasons given by users of OSS were not the lower costs but the higher security and better performance as compared to proprietary software. Therefore, the open source method of development is clearly seen by users as being innovative and providing the same or better quality.

The FLOSS survey also showed that while just over 30 per cent of all developers earn directly from their support, development or administration of OSS, a further 20 per cent earn indirectly, most of whom reported being given a job because of their experience in developing OSS. This indicates that employers value the skills that are learnt through participation in the developer community.

This finding was also supported by the FLOSS survey of user organisations, of which 36 per cent reported allowing their employees to participate in open source projects during their time at work.

To a considerable degree, therefore, the OSS community must be regarded as an informal and skill development environment that provides good training, and competitive advantages on the labour market. It is ‘costless’ in that the costs for training are not explicitly borne in monetary terms by any of the parties benefiting from the new skills made available in the market. Universities are not paying for this training (only 20 per cent of developers are students), and nor, explicitly are companies, it is the individual developers themselves who are contributing their time and effort to learn and teach others in an informal ‘apprenticeship’ system. As these costs are social costs distributed widely across individual participants, they are effectively a subsidy for sectors of society which do not or cannot explicitly pay for skill development.

This skill development process is particularly valuable for small businesses and for less wealthy regions and countries, where the high direct costs of training ICT professionals may otherwise hinder the development of a local information economy. Participation in the global open source community can help offset such costs by effectively providing a (voluntary) subsidy from the global community.

The FLOSS study showed that developers who provided ‘learning new skills’ as their reason for joining the community often show ‘sharing skills’ as an equally or more important reason for continuing their community participation. This is correlated with the duration of their participation in the community, naturally, and represents a shift from ‘apprenticeship’ to ‘mentor’ roles. In a reflection of the development process for individuals, countries that profit most from open source are those that contribute back to the community and knowledge base, and there is a built-in incentive (and low barriers) for a shift from being a recipient of skills to being a skills donor. So the process of ‘subsidy’ is very dynamic, and is likely to lead not to a dependency relationship but rather to a equal relationship based on, among other things, local specialisations for locally relevant issues.

Such skill development extends to the creation of new, local businesses, which are able to provide commercial support for and build upon OSS in a way not possible with proprietary software. This effect is heightened by any public support of the open source software sector. For example, the taking up of open source by the Extremadura region in Spain through its support for the LinEx project (a localised, Spanish-language version of the GNU/Linux operating environment) has led to an economic regeneration in a relatively poor region of the EU. This has not just allowed the implementation of activities for a lower price, but also enabled activities especially in education and training which were simply not possible with proprietary software. It has also led to the growth of a number of small businesses to provide commercial support, since with open source there is no need to approach one sole vendor for support — approaching local entrepreneurs is possible and an obvious choice.

Total cost of ownership and low labour costs
Inexpensive skill development is an important reason for developing countries to promote OSS. But, in contrast to the situation in richer countries, another reason is simply cost. Total Cost of Ownership (TCO) studies show varying results in rich countries, where labour costs are high, the relative low license fee of OSS need not necessarily reduce total costs of using and maintaining systems. When labour costs are high, labour-intensive components of the total cost (such as support, customisation, and integration — i.e. everything other than the software license fee, communication and hardware costs) represent a high share of the total cost, making the license fee itself (which is not present in the case of OSS) less crucial. In contrast, when labour costs are low, the share of license fee in the TCO is much more significant, even prohibitively so.

This relationship is neatly demonstrated by comparing license fee with a country’s GDP per capita (i.e. the average individual income). As is quickly apparent, in developing countries, even after software price provide discounts, the price tag for proprietary software is enormous in purchasing power terms. The price of a typical, basic proprietary toolset required for any ICT infrastructure, Windows XP together with Office XP, is US$560 in the U.S. [2]. This is over 2.5 months of GDP/capita in South Africa and over 16 months of GDP/capita in Vietnam. This is the equivalent of charging a single-user license fee in the U.S. of US$7,541 and US$48,011 respectively, which is clearly unaffordable. Moreover, no likely discount would significantly reduce this cost, and in any case the simple fact that a single vendor controls any single proprietary software application means that there can never be a guarantee that any discount offered is intended to be sustained for the long term, rather than as a temporary measure used to tempt consumers into a lock-in situation at which point in time the discount can be reduced.

This simple calculation is presented in the table for 176 countries, together with 10 geographical and political aggregates. The table also includes the piracy figures published by the Business Software Alliance (BSA). It should be noted that there is a correlation between the piracy rate and the effective software license fee, that is, the more expensive the software is, the higher the piracy rate. This is common sense, but does not seem to be reflected in the BSA estimates of the ‘losses’ to the software industry based on piracy, which assume that all the estimated unlicensed copies of software in a country should (or could) be replaced with paid licensed copies. Ironically, the logical conclusion of the increasingly stringent international campaign for strong enforcement of copyright is the reduction of piracy rates not through taking up licensed, proprietary software, but through the use of OSS. Anecdotal evidence shows that this is the case in Argentina, Peru and other countries especially in Latin America, where a campaign for strong copyright enforcement has coincided with poor economic conditions.

Table 1: License fee relative to GDP/capita.

Country	GDP/cap	PCs ('000s)	Piracy	WinXP Cost(3)
				Effective $	GDP Months
Albania	1300	24	n.a.	15196	5.17
Algeria	1773	220	n.a.	11140	3.79
Angola	701	17	n.a.	28184	9.59
Antigua and Barbuda	9961	n.a.	n.a.	1983	0.67
Argentina	7166	3415	62%	2757	0.94
Armenia	686	24	n.a.	28806	9.80
Australia	19019	10000	27%	1039	0.35
Austria	23186	2727	33%	852	0.29
Azerbaijan	688	n.a.	n.a.	28708	9.77
Bahrain	12189	92	77%	1621	0.55
Bangladesh	350	254	n.a.	56401	19.19
Barbados	10281	25	n.a.	1921	0.65
Belarus	1226	n.a.	n.a.	16120	5.48
Belgium	22323	2394	n.a.	885	0.30
Belize	3258	33	n.a.	6064	2.06
Benin	368	11	n.a.	53613	18.24
Bhutan	644	5	n.a.	30668	10.43
Bolivia	936	175	77%	21109	7.18
Bosnia and Herzegovina	1175	n.a.	n.a.	16818	5.72
Botswana	3066	66	n.a.	6444	2.19
Brazil	2915	10835	56%	6777	2.31
Bulgaria	1713	n.a.	75%	11534	3.92
Burkina Faso	215	17	n.a.	91801	31.23
Burundi	99	n.a.	n.a.	198864	67.65
Cambodia	278	18	n.a.	71184	24.21
Cameroon	559	60	n.a.	35319	12.01
Canada	22343	14294	38%	884	0.30
Cape Verde	1317	31	n.a.	14998	5.10
Central African Republic	257	7	n.a.	76998	26.19
Chad	202	12	n.a.	97728	33.24
Chile	4314	1640	51%	4579	1.56
China	911	24222	92%	21678	7.37
Colombia	1915	1810	52%	10316	3.51
Comoros	386	3	n.a.	51208	17.42
Congo, Dem. Rep.	99	n.a.	n.a.	199394	67.83
Congo, Rep.	886	12	n.a.	22288	7.58
Costa Rica	4159	659	64%	4750	1.62
Cote d’Ivoire	634	118	n.a.	31140	10.59
Croatia	4625	376	67%	4272	1.45
Cyprus	12004	188	61%	1646	0.56
Czech Republic	5554	1490	43%	3557	1.21
Denmark	30144	2896	26%	655	0.22
Djibouti	894	7	n.a.	22107	7.52
Dominica	3661	5	n.a.	5396	1.84
Dominican Republic	2494	n.a.	64%	7922	2.69
Ecuador	1396	300	62%	14149	4.81
Egypt, Arab Rep.	1511	1010	58%	13075	4.45
El Salvador	2147	140	73%	9203	3.13
Equatorial Guinea	3935	2	n.a.	5021	1.71
Eritrea	164	8	n.a.	120613	41.03
Estonia	4051	238	53%	4877	1.66
Ethiopia	95	75	n.a.	208612	70.96
Fiji 2061	50	n.a.	9584	3.26
Finland	23295	2197	27%	848	0.29
France	22129	19949	46%	893	0.30
Gabon	3437	15	n.a.	5747	1.96
Gambia	291	17	n.a.	67847	23.08
Georgia	601	n.a.	n.a.	32884	11.19
Germany	22422	31471	34%	881	0.30
Ghana	269	66	n.a.	73442	24.98
Greece	11063	860	64%	1786	0.61
Grenada	3965	13	n.a.	4982	1.69
Guatemala	1754	150	73%	11261	3.83
Guinea	394	30	n.a.	50090	17.04
Guinea-Bissau	162	n.a.	n.a.	121634	41.38
Guyana	912	20	n.a.	21670	7.37
Haiti	460	n.a.	n.a.	42984	14.62
Honduras	970	80	68%	20371	6.93
Hong Kong, China	24074	2600	53%	821	0.28
Hungary	5097	1021	48%	3876	1.32
Iceland	27312	118	n.a.	723	0.25
India	462	6031	70%	42725	14.53
Indonesia	695	2298	88%	28412	9.66
Iran, Islamic Rep.	1767	4495	n.a.	11177	3.80
Ireland	26908	1500	42%	734	0.25
Israel	17024	1564	40%	1160	0.39
Italy	18788	11286	45%	1051	0.36
Jamaica	3005	130	n.a.	6573	2.24
Japan	32601	44311	37%	606	0.21
Jordan	1755	165	67%	11257	3.83
Kazakhstan	1503	n.a.	n.a.	13143	4.47
Kenya	371	172	77%	53283	18.12
Kiribati	430	2	n.a.	45919	15.62
Korea, Rep.	8917	12142	48%	2215	0.75
Kuwait	16048	270	76%	1231	0.42
Kyrgyz Republic	308	n.a.	n.a.	64178	21.83
Lao PDR	326	16	n.a.	60625	20.62
Latvia	3200	361	59%	6173	2.10
Lebanon	3811	247	79%	5184	1.76
Lesotho	386	n.a.	n.a.	51122	17.39
Liberia	163	n.a.	n.a.	121417	41.30
Lithuania	3444	246	56%	5736	1.95
Luxembourg	42041	228	n.a.	470	0.16
Macao, China	14089	79	n.a.	1402	0.48
Macedonia, FYR	1684	n.a.	n.a.	11735	3.99
Madagascar	288	39	n.a.	68550	23.32
Malawi	166	13	n.a.	118904	40.45
Malaysia	3699	3000	70%	5341	1.82
Maldives	2082	6	n.a.	9487	3.23
Mali	239	13	n.a.	82801	28.17
Malta	9172	91	53%	2154	0.73
Marshall Islands	1830	3	n.a.	10795	3.67
Mauritania	366	28	n.a.	53959	18.35
Mauritius	3750	131	65%	5268	1.79
Mexico	6214	6835	55%	3179	1.08
Micronesia, Fed. Sts.	1973	n.a.	n.a.	10012	3.41
Moldova	346	68	n.a.	57020	19.40
Mongolia	433	35	n.a.	45598	15.51
Morocco	1173	400	61%	16840	5.73
Mozambique	200	63	n.a.	98978	33.67
Namibia	1730	65	n.a.	11420	3.88
Nepal	236	83	n.a.	83770	28.50
Netherlands	23701	6872	39%	834	0.28
New Zealand	13101	1511	26%	1508	0.51
Niger	175	6	n.a.	113078	38.46
Nigeria	319	889	71%	62014	21.09
Norway	36815	2292	34%	537	0.18
Pakistan	415	585	83%	47630	16.20
Palau	6280	n.a.	n.a.	3146	1.07
Panama	3511	110	61%	5627	1.91
Papua New Guinea	563	298	n.a.	35071	11.93
Paraguay	1337	76	72%	14777	5.03
Peru	2051	1262	60%	9630	3.28
Philippines	912	1702	63%	21658	7.37
Poland	4561	3301	53%	4331	1.47
Portugal	10954	1177	43%	1803	0.61
Puerto Rico	17682	n.a.	47%	1117	0.38
Romania	1728	801	75%	11433	3.89
Russian Federation	2141	7200	87%	9226	3.14
Rwanda	215	n.a.	n.a.	92034	31.31
Samoa	1465	1	n.a.	13485	4.59
Sao Tome and Principe	311	n.a.	n.a.	63600	21.63
Saudi Arabia	8711	1343	52%	2268	0.77
Senegal	476	182	n.a.	41539	14.13
Seychelles	6912	12	n.a.	2858	0.97
Sierra Leone	146	n.a.	n.a.	135380	46.05
Singapore	20733	2100	51%	953	0.32
Slovak Republic	3786	800	46%	5218	1.77
Slovenia	9443	549	60%	2092	0.71
Solomon Islands	614	22	n.a.	32173	10.94
South Africa	2620	2962	38%	7541	2.57
Spain	14150	6916	49%	1396	0.47
Sri Lanka	849	175	n.a.	23257	7.91
St. Kitts and Nevis	7609	8	n.a.	2596	0.88
St. Lucia	4222	23	n.a.	4679	1.59
St. Vincent and the Grenadines	3047	13	n.a.	6483	2.21
Sudan	395	115	n.a.	49990	17.00
Suriname	1803	19	n.a.	10955	3.73
Swaziland	1175	n.a.	n.a.	16816	5.72
Sweden	23590	4991	31%	837	0.28
Switzerland	34171	3906	33%	578	0.20
Syrian Arab Republic	1175	270	n.a.	16815	5.72
Tajikistan	169	n.a.	n.a.	116879	39.76
Tanzania	271	115	n.a.	72860	24.78
Thailand	1874	1698	77%	10540	3.59
Timor-Leste	517	n.a.	n.a.	38212	13.00
Togo	270	100	n.a.	73033	24.84
Tonga	1406	n.a.	n.a.	14054	4.78
Trinidad and Tobago	6752	91	n.a.	2926	1.00
Tunisia	2066	229	n.a.	9560	3.25
Turkey	2155	2792	58%	9167	3.12
Turkmenistan	1097	n.a.	n.a.	18010	6.13
Uganda	249	71	n.a.	79324	26.98
Ukraine	766	898	86%	25802	8.78
United Kingdom	24219	21533	25%	816	0.28
United States	35277	178326	25%	560	0.19
Uruguay	5554	370	63%	3557	1.21
Uzbekistan	450	n.a.	n.a.	43943	14.95
Vanuatu	1058	n.a.	n.a.	18677	6.35
Venezuela, RB	5073	1300	55%	3895	1.32
Vietnam	411	933	94%	48011	16.33
West Bank and Gaza	1286	n.a.	n.a.	15366	5.23
Yemen, Rep.	514	35	n.a.	38434	13.07
Yugoslavia, Fed. Rep.	1020	249	n.a.	19373	6.59
Zambia	354	72	n.a.	55824	18.99
Zimbabwe	706	155	68%	27965	9.51
Regional Aggregates [4]
European Union	20863	116997	n.a.	947	0.32
EU Accession countries	4840	8286	n.a.	4082	1.39
EU applicant countries	2023	3592	n.a.	9766	3.32
The Caribbean	4560	308	n.a.	4332	1.47
Latin America	4335	18703	n.a.	4557	1.55
Africa	652	7636	n.a.	30297	10.31
Middle East	2679	9708	n.a.	7375	2.51
Asia	2128	102229	n.a.	9282	3.16
Oceania	13946	11886	n.a.	1417	0.48

Source: World Bank World Development Indicators Database, 2001; Piracy data from Business Software Alliance GDP/capita in US$, Windows + Office XP cost in effective US$ equivalent.

Conclusion
There is much current debate on whether governments should promote or encourage adoption of OSS in their economies, such as through public procurement practices. While there are a number of reasons cited for and against such action, any policy based purely on value–for–money considerations would, faced with a choice of spending either 0 or 16 months of GDP, necessarily prefer the former.

ICTs are supposed to be an ‘enabler’ for growth in developing countries. Such growth cannot spread much beyond a very small elite if the basic enabling software infrastructure requires the investment of several months’ worth of GDP on software license fee, repeatedly, every few years in an upgrade cycle beyond the control of users.

Moreover, economic growth driven by ICT depends on the wide dissemination of ICT usage and competences. The skill development aspects of open source encourage this, provide support for the generation of local ICT industries, and furthermore facilitate a reciprocal relationship where developing economies and local players can quickly start contributing to the global software developer community, and hence to the global economy.

To conclude, in the interest of sustainable, long-term and widespread economic growth and ICT development, developing countries need to seriously consider the adoption, and promote OSS in order to develop local skills and businesses, actively participate in the global ICT economy, and avoid unnecessary expenditure.

Notes MERIT/Informics and Berlecon Research 2002, “Free/Libre/Open Source Software Study– Final Report,” at http:// www.flossproject.org/report/ Price from amazon.com in June 2003. Windows + Office XP equivalent US$ cost calculation = $560* (country GDP per capita/U.S. GDP per capita).< /li >