Hacker News Reader

Comment by walterbell | Hacker News Reader

Why isn't a search engine sufficient for discovering <foo-language> packages and resources? Historically, human-edited lists don't have a long archival lifetime, e.g. original Yahoo, DMOZ, Mahalo.

If curation is better than search, why is a curated list better hosted at github than Wikipedia, which has a lower technical cost of editing? If the problem is Wikipedia moderation, who will reconcile multiple github curated lists?

Is the expectation that github's "awesome-xyz" repos will rank higher in Google search results than a curated Wikipedia page? Or is this github-specific SEO?

View on HN

For what it's worth, I wrote https://lua-toolbox.com/ (a website referencing all Lua rocks which users can endorse).

A search engine will allow you to discover tools you are looking for, but sometimes you do not know exactly what to search. There are way fewer Lua modules than e.g. Python or Ruby libraries so exhaustive lists are still manageable.

Also, sometimes you want to know what people are actually using. A bit of "social proof" is always good when you are looking for Open Source libraries.

There is more to say about this and the Lua modules ecosystem, which is why I gave an entire talk about that topic at Lua Workshop last year http://files.catwell.info/presentations/2013-11-lua-workshop... :)

A search engine is technically sufficient, but takes more time and is less accessible. A search engine won't put things in front of you for you to peruse and explore; you have to know what you're looking for. Beyond that, the contributors and I have done research to find and decide on the entries on this list, which should help readers avoid repeating the same process.

A git repo has an explicitly responsible maintainer, while wikipedia articles do not. This helps keep the quality high.

Regarding SEO - even now, the same day I first pushed this, searching 'lua awesome' returns this as result #5 on Google. 'awesome-<language>' is a known convention for curated lists like this one, and people familiar with that convention who are looking for Lua resources will easily find this.

> the contributors and I have done research to find and decide on the entries on this list

It would be helpful to have a summary of selection criteria at the top of the list, in addition to the longer version on the Contributing file.

If such curated lists were also available in JSON, developers could meta-curate a locally searchable union of all their upstream projects. It would then be easy to update, annotate or cross-reference lists.

Maybe the way to judge curation is by comparing curation+search to search alone. With good curation effort, i think that curation+search is a winner , at a relatively minor cost(needing more time).

Also curation has the benefit of being a useful discovery tool.

In theory, https://www.openhub.net/p?r&q=lua could construct/curate such lists from project activity analytics.

Personally i think human curation is much better than curation by activity analytics, and it can offer better context.

No question. But it usually doesn't last, as people move onto other projects. Even within this thread, the new list is replacing a similarly named, now unmaintained list.

DMOZ/Wikipedia at least try to provide long-term continuity and a standardized ontology for discovery. This github list is depending on the searcher already knowing that they should preface their dev discovery queries with "awesome".

If this list format were standardized, it could be dynamically merged with data from OpenHub. Or GitHub could add analytics features (similar to OpenHub) to assist the human curator editing this GitHub repo.

This list replaces a centralized list. Maybe with the magic of git this will work better. Or maybe we need better community building and connection around such list.

But since such a knowledge source is highly useful for people, and very valuable and people could get lots of credit for helping(in theory), i could see something like this built up.

On the other hand comparing just the current list(5 days of operation) , with the best openhub or github search could offer for the queries "python cryptography" and "python xml", i don't think analytics alone should do. But sure analytics can(and are through Google/Github) being used to guide curators.

Reply on news.ycombinator.com