FLOSS != GitHub: A Case Study of Linux/BSD Perceptions from Microsoft's Acquisition of GitHub
aa r X i v : . [ c s . S E ] F e b FLOSS != GitHub: A Case Study of Linux/BSDPerceptions from Microsoft’s Acquisition of GitHub
Raula Gaikovina Kula ∗ , Hideki Hata ∗ , Kenichi Matsumoto ∗∗ Nara Institute of Science and Technology, Japan { raula-k, hata, matumoto } @is.naist.jp Abstract —In 2018, the software industry giants Microsoft madea move into the Open Source world by completing the acquisitionof mega Open Source platform, GitHub. This acquisition was notwithout controversy, as it is well-known that the free softwarecommunities includes not only the ability to use software freely,but also the libre nature in Open Source Software. In this study,our aim is to explore these perceptions in FLOSS developers. Weconducted a survey that covered traditional FLOSS source Linux,and BSD communities and received 246 developer responses.The results of the survey confirm that the free community didtrigger some communities to move away from GitHub and raiseddiscussions into free and open software on the GitHub platform.The study reminds us that although GitHub is influential andtrendy, it does not representative all FLOSS communities.
I. I
NTRODUCTION
Microsoft’s GitHub is now reported to support a communitywhere more than 40 million people learn, share, and worktogether to build software. The git-based platform has grownto around 100 million repositories and has released some ofthe world’s most influential technologies, such as React Nativeby Facebook, Tensorflow by Google, and Swift by Apple justto name a few. GitHub is based on social coding, which intro-duces online collaborations to attract and sustain developersto a project [1]. In June of 2019, Tidelift and The New Stackjointly fielded a survey of professional software developers toshow that most developers (84%) view themselves as activecontributors to open-source projects [2].On October 26th 2018, Microsoft announced on its officialblog that it had completed its acquisition of GitHub [3]. In ablog Nat Friedman outlined two principles for GitHub [4]:1)
GitHub will operate independently (from Microsoft) asa community, platform, and business.
This means thatGitHub will retain its developer-first values, distinctivespirit, and open extensibility.2)
GitHub will retain its product philosophy.
We loveGitHub because of the deep care and thoughtfulness thatgoes into every facet of the developer’s experience.With these statements, GitHub confirms its intention to retainand grow its base of Open Source developers.From a traditional point of view, the free software com-munity represents a different aspect of the open source soft-ware community. It is described as a campaign for computerusers’ freedom and the open source camp that only focuson practical benefits of software. To be fair to all camps,we use the term “FLOSS,” meaning “Free/Libre and OpenSource Software” [5]. Although the free software community has had its share of disagreements with Microsoft [6], [7],[8], [9], the only reported negative opinion of free softwarecommunity has different attitudes towards GitHub is the ideaof ‘forking’ so far, as it it is considered as a danger to FLOSSdevelopment [10].In this paper, we report on how external events such asacquisition of the open source platform by a closed sourceorganization triggers a FLOSS developers such the Linux/BSD Free Software communities.II. T
ARGET S UBJECTS AND S URVEY D ESIGN
Table I shows a listing of the targeted subjects for thestudy. It is important to note that our survey results do notrepresent the perceptions of all FLOSS developers. Insteadis only contains a targeted subset of contributors of Linuxdistributions/kernel and BSD families to infer the opinion oftraditional FLOSS development communities. We distributedour survey via two lesser intrusive communication channels: i.Forums.
Our attempt to use the forums was to follow developercommunication channels. In this case, we posted in the moregeneric and random channels. ii. Mailing Lists.
In the absenceof a forum, we reverted to a mailing list. Since the mailinglists are curated, we had to request the mediating curator forpermission before posting the survey.Table II outlines the structure of the questionnaire wedesigned our online form. The survey was broken into threemain sections, that were based on the respondents activitiesafter the acquisition and are listed below: • Remaining on GitHub
This section would be filled out byrespondents that had GitHub projects and would continueto use GitHub after news of the acquisition. • Moved away from GitHub
This section would be filledout by respondents that moved away from GitHub afternews of the acquisition. • Never used GitHub
This section would be filled out byrespondents that have never used GitHub.Finally, we asked all respondents to provide additional open-ended feedback on the acquisition. The intention was to laterclassify whether or not the feedback was positive, neutral ornegative.
A. Participant Demographics
Table III depicts the total respondents to our survey, totallingto 246 total responses. Considering that we used developercommunication channels for our survey, it was interesting that
Table IT
ARGETED L INUX D ISTRIBUTIONS AND
BSD C
OMMUNITIES
Channel CommunitiesForums Mint, Manjaro,Debian, Solus, Antergos, openSUSE, MX Linux, Zorin, Arch, ReactOS, Lite, PuppyFreeBSD, SparkyLinux, Slackware, Xubuntu, Devuan, Bodhi, Gentoo, Kubuntu, Sabayon, KNOPPIX, 4MLinux, Tiny CoreClearOS, GhostBSD, NixOS, Ubuntu Studio, NuTyX, wattOS, LibreELEC, Trisquel, siduction, Porteus, Elive, ScientificParabola, Maui, Zenwalk, BunsenLabs, Void, Artix, Salix, Pardus, FreeNAS, Pinguy, NAS4Free, IPFireOpenMediaVault, pfSense, Fatdog64, Neptune, SUSE, VyOS, MiniNo, Arya, Runtu, Peach OSI, SalentOS, Zevenet3CX, NethServer, Wifislax, ArchStrike, Porteus Kiosk, Funtoo, KXStudio, Freespire, OviOS, Haiku, Pearl, KaroshiMINIX, Untangle, LinuxBBQ, Refracta, BigLinux, HardenedBSD, PrimTux, EasyNAS, MidnightBSD, Toutou, TurnKey, DietPiKANOTIX, Cucumber, Linspire, AsteriskNOW, RISC, CloudReady, Rebellin, RaspBSD, Springdale, Securepoint, PLD, SME ServerSwift, TalkingArch, NexentaStor, SMS, Ulteo, Volumio, SuliX, Webconverger, DRBL, Dragora, UBports, Liquid LemurAIO, SuperGamer, Namib, ObarunMailing List Debian, Fedora, LXLE, ROSA, DragonFly, Calculate, OpenMandriva, IPFire, SliTaz, NetBSD, Uruk, CRUXheads, Debian Edu, Endian, OSGeo, LuninuX, WhonixAPODIO, Rocks Cluster, Clear, Lunar, Frugalware, GoboLinuxMirOS, Super Grub2, Bio-Linux, GuixSD, Rescatux, gNewSenseExherbo, Thinstation, Vine, BSDRP, OLPC, T2Grml, Swecha Table IIQ
UESTIONS THAT WERE ASKED TO THE R ESPONDENTS
Survey Questions to be answered using a Likert Scale RankingRespondents 1. Do you think it would be a good idea to move the project away from GitHub?that remain on GitHub 2. If the project will be moved away from GitHub to another platform, how muchadditional effort will be required to get accustomed to the new platform?3. If you have any specific comments, please feel free to add here (Optional)Respondents 1. How much do you think this decision to move away from GitHub was related to the acquisition?that moved away 2. Do you agree with the decision to move the distribution away from GitHub?from GitHub 3. Does moving away from GitHub affects your contributions to this project?5. How much additional effort will be required to get accustomed to the new platform?4. If you have any specific comments, please feel free to add here (Optional)Respondents 1. Do you think it would be a good idea to move the project to GitHub?that do not use GitHub 2. If the distribution/kernel will be moved to GitHub, how much additionaleffort will be required to get accustomed to the new platform?3. Apart from Linux and BSD contributions, have you had personal experiences with using GitHub?4. If you have any specific comments, please feel free to add here (Optional)Open-end feedback 1. If you have any generic comments, please feel free to add here (Optional).
Table IIID
EMOGRAPHICS AND P ERSPECTIVE OF L INUX OR
BSD D
ISTRIBUTIONRESPONDENTS . R
ESPONDENTS CAN CHOOSE MORE THAN ONEDEMOGRAPHIC . Demographic (multiple choice allowed)
Casual contributors 149Core contributors 64Project manager/team leader 47Others 16
Perspective
Remain on GitHub 138 (56%)Moved Away from GitHub 75 (31%)Do not use GitHub 33 (13%)TOTAL 246 (100%) the responses were mainly filled out by many causal contrib-utors (149). This was followed by core contributors (64) and then the project manager/leader (47). Before we proceeded, weasked participants on their impressions of GitHub. We foundthat 63% of the respondents were fans of GitHub. A majorityof the respondents either agreed (46%) or had not opinion(20%) whether ‘GitHub appeals with access to over 27 millionusers’ . Likewise many agreed (58%) that GitHub has a usefulset of functions for developers. However, developers were notready to openly compare GitHub when compared to similarplatforms, with 59% of developers not agreeing or having noopinion on whether GitHub is the superior platform.We report that over half of respondents (55%) claim thatthe acquisition would be detrimental to their GitHub projects.Furthermore, most respondents negatively responded to thepossibility of an expansion of Free and Open source contrib-utors (74%), and about half of the respondents (45%) did notthink the acquisition would bring improvements, reliability,and services with the platform.
III. S
URVEY F INDINGS
Based on the questionnaire, we first analysed all answersaccording the perspective and then conducted a deeper analysisof the open-ended optional comments.
A. Feedback based on Perspective
As shown in Table III, we found that over half of ourrespondents did not move away from GitHub (138, 56%),while a third of the respondents already had their projectsmove away from GitHub (75, 31%). Finally, there was asmaller number of respondents that did not host any of theirprojects on GitHub (33, 13%).For developers that are currently contributing to GitHub,33% of respondents thought it was a bad idea for their projectsto move. In terms of the additional effort required to move,there was mixed responses, with no clear majority.For the developers that had any of their projects migrateaway, the majority completely agreed (65%) with the decisionto move and were sure that the acquisition was the keymotivation (59%). Interestingly, when asked about how themove would affect their contributions, developers were lessforthcoming with 53% either neutral or having no opinion.Still developers were more optimistic with the move, with amajority of 36% confident that no additional effort would berequired to get accustomed to the new platform.For the developers that never contributed to any GitHubprojects, 80% thought that moving to GitHub was not a goodidea. In terms of the effort required, a significant number ofrespondents (29%) had no opinion, while 48% reported that itwould take them much effort to get accustomed. That beingsaid, a majority (65%) of these developers had other opensource of personal projects on GitHub.
Takeaway 1:
Some of the responded developers ofthe Linux and BSD distribution had left GitHub orhad not used GitHub, although the majority continuedusing GitHub. Developers that stayed with GitHub didnot cite that additional effort is not a reason why theyremained with GitHub.
B. Open-end Feedback
We also collected all open-end comments (optional) andmanually categorized the reasons based on their polarity oftheir sentiments (i.e., positive, neutral or negative). Out of the70 responses, we were removed 9 responses that were notrelevant, leaving 61 sentences. a) Positive Responses : We find 3 respondents that werein favor of the acquisition, stating that ‘Microsoft brings money, money means stable andreliable hosting of our source code’
As has been reported in previous studies, innovative servicesare the key to sustainable projects [11]. Other sentimentsincludes the statement that ‘It’s useful for mirrors in orderto make it easy for drive-by contributors’ and that it isconvenient, stating that ‘I don’t see anything wrong with buying Github by Microsoft. For example, we use Skype sofar and we will continue to use it until it suits our needs, oruntil we do something relatively convenient for us.’b)
Neutral Responses : For the 28 respondents, they werenot as concerned of the acquisition as much, citing otheralternative platforms such as GitLab or that ‘It’s easy tomigrate to other git hosting platforms, but that’s not neededright now for me.’
From this point of view, respondents didcite the benefits of using GitHub because of its huge userbase, ‘The problem is, if too much devs move to otherplatforms, then the greatest benefit of GitHib (easycross-contribution and linking between differentprojects) will be lost. Thus follow the herd ;).’
Another respondent expressed that the strong support ‘Githubhas strong support for SaaS integration (CI/CD services inparticular), but these services are provided by third parties.Github’s competitors .. provide CI/CD functionality as part oftheir packages which would probably make me decide to usethem instead of Github if I had to set up a new user accountright now.” and yet another ‘Imho GitLab is superior in partsof functionality but is still lacking the userbase.’
Other respondents cited that GiHub was not their primarysource, thus did not see any immediate threat. For example,one respondent stated that ‘FreeBSD doesn’t use GitHub as itsprimary source anyway’ and that it was too early to see anyaffect, ‘Practically speaking, nothing has changed yet. We arenot even talking about predictions, but about “gut feelings”’ .As mentioned by a Debian developer: ‘The Debian project had always been in favour ofnot using services running on infrastructure that isnot managed by Debian. Impact to the project hasbeen minimal as we had already our own infrastruc-ture (Alioth recently migrated to salsa.debian.org,our own gitlab infrastructure). Debian is used tohandling upstream developers using many differentforges (Sourceforge, Savannah, GitHub, Gitlab) andthe impact of upstream moving from GitHub toGitSalsa has not been significant.’
This sentiment can be seen in other responses that state that ‘The ownership of GitHub is not relevant at all.’ . Overall, thereis divided feelings, even within the teams: ‘Many team members are not concerned about theuse of Github, and the convenience of the features onGithub are useful. Visibility of the project (exampleMapserver) is very important to some team leaders,and there is no replacement. We at OSGeo dot orghave already built our own git server, and it ispopular in our group.. the content of the projectsis perfectly mirrored, but the tickets and PR historydoes not port. Some people have found tools tomigrate away from Github, but it has not happenedin general. The opinions about this are very divided,with strong feelings on different sides.’
In fact, respondents were quick to remark that GitHub ispropriety. ‘The problem is NOT GitHUB-Microsoft deal butmore general, a small example: in the past we (nearly) all use usenet. There are TONS of newserveraround the world, anyone can host another if heor she want. There are plenty of groups some fun,some about sport, some about work, some to helpothers, some to discuss new ideas, ... today we haveFEW non-integrated proprietary platforms (from Fbto StackExchange to Reddit to 4Chan, YC, ... allof them are POOR replacements of ancient ng. Allof them are proprietary and depend on a singlecompany.’
However, being propriety itself had mixed feelings, with onerespondent stating that: ‘This has always been a proprietaryservice using open source technologies. The owner doesn’treally matter to me as long as they don’t mess with code orvisibility of projects for business purposes’ . c) Negative Responses : We found 30 negative responses,which expressed a distrust of a closed source to merge withan open source. For example, ‘A Platform to develop OpenSource software cannotbe controlled by King of closed source software. It’sa contradiction.’
Similar responses where expressed by respondents, even ifthere was no grounds. For example, ‘It’s quite hard for me to trust Microsoft, they’reprobably still planning to EEE stuff - even a ”harm-less” tiger raised in captivity will still bite’
Ideas of freedom, in terms of both legal and that it could havean effect on the quality of the code. Such concerns arisedwith a respondent that stated that ‘.. the ability to change thefoundations and the fundamentals of your project by usingterms and conditions will result in a poorer product and morepeople worried about legal implications instead of writingcode.’ . Another respondent was concerned with the idea ofa monoculture, stating that ‘Monocultures are to be avoided, GitHub shouldsimply be one of a number of similar projects tobe used by software projects.’
This is supported by another respondent that ‘”GitHub isa centralised, proprietary platform and always has been’ .Another respondent was abit less direct, stating that ‘I preferfree and open tools’ . Another yet, stated how it impedescompetitive software (sw), ‘MS will impede development ofcompetitive sw to MS products’ .Finally, there were statements that directly referenced thefreedom of open source in their response, stating that: ‘Today and not since today our freedom is get-ting smaller and smaller from the hardware (UEFI(in)secure-boot etc) to services (how many peopleinstead of have a local maildir sync-ed, for instance,only live on webmails?) to the society in general’
Takeaway 2:
We found that the theme of the free-nature of Open Source, especially in regards to GitHubOpen Source projects was raised during the discussionof the acquisition. IV. A Y
EAR F OLLOW - UP We contacted a subset of the developers on November 12th2019 to confirm the results and survey their stance, a yearafter the acquisition and first survey. The main goal was toconfirm our findings by asking the single question:
After oneyear, has your perception on Microsoft’s acquisition of GitHubchanged?
Out of the 16 emails sent out to the respondents, wereceived 4 responses.We found that all respondents had unchanged sentimentstowards the acquisition, with stronger conviction on the vi-olation of freedom. One interesting response we found thatGitHub should not be a reflection of all Open Source: ‘All development of open source is happening onGithub - therefore Github = Open Source and thatI have heard not once but many times and I havealso heard Git = Github’
Furthermore, the social impact of GitHub on the OpenSource community is a concern for a free community devel-oper. ‘What many open source developers see in Githuband many other services is a free lunch which theyare so it’s hard to argue about that. It’s a freelunch that encourages you to market the free lunchand that you are almost an outcast if you aren’t.“Hey why aren’t you putting the github link on yourpage? Don’t you know all open source happens onGitHub?”’
Takeaway 3:
The acquisition provides a reminder toacknowledge the existence of FLOSS Communitiesthat reside outside the GitHub Platform.V. C
LOSING R EMARKS
We understand and acknowledge that our survey may onlyrepresent a small fraction of the free community of traditionalOpen Source software, but it is a good indicator that the free-dom community is still active. Looking forward, we recognizethat although GitHub hosts influential and exciting part ofOpen Source, it is a reminder that GitHub is not always arepresentative FLOSS. R
EFERENCES[1] R. G. Kula and G. Robles,
The Life and Death of Software Ecosystems
Empirical Softw.Engg. , vol. 22, no. 1, p. 547–578, Feb. 2017. [Online]. Available:https://doi.org/10.1007/s10664-016-9436-6[11] H. Hata, T. Todo, S. Onoue, and K. Matsumoto, “Characteristics ofsustainable oss projects: A theoretical and empirical study,” in2015IEEE/ACM 8th International Workshop on Cooperative and HumanAspects of Software Engineering