Hacker News Reader

Comment by xorand | Hacker News Reader

>...an article puts the abandonment rate of open source projects on Github at about 98% - meaning that there is no activity on 98% of projects after a year [0]. This has coined a phrase - abandonware.

I'm not a programmer, I'm a researcher. Let me tell you a fact: about 100% of research articles are not updated.

Somehow the link [0] cited in the article makes an indirect similar point, but what's wrong with programs which are not updated? After all there is a (small) group of people who keep telling that programs are the next level in mathematics, (ideally they are) better than proofs.

[0] https://www.techrepublic.com/article/open-source-failure-is-...

View on HN

Moreover he's using the large amount of wreckage to argue that the Darwinian process of selecting "category killers" isn't working, which is completely backwards.

If "category killer" is a winner-take-all competition, then there are going to be a vast number of losers, and we do see a vast number of unmaintained projects. It's just that no one is bothering to remove them from view when they're defeated.

I'm unaware that the goal of evolution is to have one lion win the lion competition and eat all other lions.

Maybe we should stop using bad metaphors? Writing software is not a Darwinian process, neither is competition on the market place. Unless you got your OS by randomly twiddling bits in the binary file, in which case why are we even talking about open source?

> have one lion win the lion competition and eat all other lions

If that improved the chances of that lion's genes surviving, it would absolutely be selected for. Obviously it generally doesn't because there's safety in numbers, etc.

> the goal of evolution

Minor nit: evolution has no goal. It is undirected, random, capricious, etc.

The analogy might be clearer if instead of GitHub we used Fossil.

He does talk about how huge corporations use open source and exploit open source developers for their own profit...

Which is an important advantage to open-source. Abandoned, failed close-source software is almost never available. Talk about wasted productivity... If an abandoned github repo has any redeemable value, it will get found and used.

Not to mention that abandoned github repos wind up in google searches and I have learned something from looking at them in such a context, even if untouched for years.

Don't delete your repos because they are "old". You could be doing a favor for some random person with a google search, looking at something you looked at too.

> This has coined a phrase - abandonware.

Huh. "Abandonware" is a very old word as far as internet culture goes, referring to something entirely different.

The concept does need some qualification. “Small software” can probably become “done.” Grep, as an example, is done. Frameworks, libraries, and applications all have a steering wheel requirement of support.

I’d suggest that a fair amount of github open source is resume padding, or at least an attempt at it rather than “scratching a developer’s itch” which the original ESR paper mentioned as a reason.

I suspect most software is overdone, I was looking at Node Express the other day and it struck me they should have stopped when promises and async/await came along and moved on to another project.

>Somehow the link [0] cited in the article makes an indirect similar point, but what's wrong with programs which are not updated?

Bit rot, for one (not keeping up with changing OSes, APIs they talk to, environments, libs they use, security concerns, etc), but also the fact that when they stopped updating they were still incomplete and more often than not half arsed.

Are there any numbers on closed source projects to compare this with? Say, if closed source projects have a 99% abandonment rate, what do these numbers tell us?

I'm not a researcher, I'm a programmer.

The same abandonment happens within healthy corporations with proprietary source code.

There are even times when the source code is abandoned but the object code is in active use :-)

emulators make the object code portable :)

I agree. There are plenty of good points to consider in the article, but this is not one of them. Lots and lots of things on Github are little experiments that someone tried out and made available so that others can benefit from what they learned or tried. That can be called 'abandonware' in order to reinforce a straw man, or it could also be called a source of community learning.

I thought proofs and programs were the same thing

Constructive proofs are equivalent to programs. But until somebody writes the program, there are always details, context, style, etc which make a proof understandable by a human but not a program. A program is more rigorous than a proof, in this sense.

Reply on news.ycombinator.com