libreplanet-discuss
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [libreplanet-discuss] Free software is not trusted software


From: Julian Daich
Subject: Re: [libreplanet-discuss] Free software is not trusted software
Date: Sun, 27 Jan 2019 20:51:59 +0100
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.4.0


El 27/1/19 a las 2:07, bill-auger escribió:
> On Sat, 26 Jan 2019 18:35:15 -0500 bill-auger wrote:
>> such rankings could only lead to some projects optimizing
>> for the "score" as to snowball it into the "leader" position; 
> 
> allow me to elaborate on that a bit - that was not merely a vague
> prediction - it already happens - i have experienced it directly and
> it is disturbing
> 

Trustability ranks can be adjusting for not trolling people. Deffining/
ranking software" quality" and user safety are different things. Quality
can be very arbitrary.


> recently, i was informed that one of my scripts had been added to a
> a popular software repo (i do not care to promote it by name) - i looked
> at it's entry on the web and noticed that every package is assigned
> automated "scores" for quality, maintenance, popularity, and so on - my
> script was assigned an extremely low score in all categories, so i
> looked into their criteria out of curiosity - here are some of the more
> ridiculous example of where my script fails so miserably:
> 
> * if the project does not have at least 4 "badges" in its README file
>   on github, it loses points for "code quality"
> * if the project does not use travis-ci, it loses points for "code
>   quality" - (IIRC, some points can be earned only by using premium
>   proprietary web services)
> * if the project does not create an official "release" on github at
>   least once each month, it loses points in the "well maintained"
>   category
> * and IIRC, it actually loses points for not having their specific
>   packaging metadata file prominently the root of the repo master branch
>   (precisely named with their corporate brand, of course); where it is
>   actually just pollution, as packaging metadata serves no purpose in
>   the release tarballs (aka. the git master branch)
> 
> to put that into context, my script has been full-featured and stable
> for probably a longer amount of time than that company has existed
> - my script would not benefit from any of those "essential" prescribed
> webby adornments; and we should hope that no one would be compelled to
> add them, merely to achieve a better score on some gamified
> "leader-board"
> 
> it should be obvious that any developer who puts stock in such rankings
> is going to spend a disproportionate amount of time catering to the
> scoring system rather than getting any real work done; but if people
> treat software development like a game, and put popularity as a priority
> goal, then that is exactly what will happen, and it is actually
> counter-productive to the goal of quality
> 
> that is not to mention how insulting it is to an experienced developer
> to be labeled with such badges of shame, when they know damn well that
> their software is not poor quality; but that ignorant readers of such a
> website which claims to be the authority on the topic are given exactly
> that misleading impression
> 
> so i would say that for the sake of being responsible net-izens, it
> would actually be preferable not to want your favorite software featured
> on such a website at all, and to recommend that no one accepts such
> rankings at face value - it certainly does no favor for otherwise
> responsible developers, and misleads users into valuing only those
> prescribed generic quality criteria - most disturbingly, it rewards
> developers for treating their craft as a game, and punishes the ones
> who take they work more seriously, and who avoid adding unnecessary
> baggage for frivolous "populous" reasons
> 
> _______________________________________________
> libreplanet-discuss mailing list
> address@hidden
> https://lists.libreplanet.org/mailman/listinfo/libreplanet-discuss
> 

-- 
Julian Daich

address@hidden

FCL
www.freecomputerlabs.org



reply via email to

[Prev in Thread] Current Thread [Next in Thread]