I would ask @greebie to take a closer look, as he's an expert in network analysis, but my sense of this after reviewing the documentation and having worked in the digital humanities side of the house:

• I think both the Gephi and GraphML derivatives cover the same use cases - you have granular timestamp, as you want, and then it indicates the origin and destination links. The LGA might be slightly more space efficient, but it's also not as readable for spot checking or quickly reviewing data.
• Part of the project goal is taking WARCs and making them accessible... GEXF and GraphML are standardized file formats that are usable by a wide variety of network projects (including Gephi and SigmaJS), whereas to my knowledge, LGA is only used by IA and Archive-It services. I could be wrong on the latter but they're certainly not as widely used as GEXF and GraphML.

So I think since we cover the use cases, we shouldn't go down this road. But happy to hear other opinions.

I remember taking a look at this before. At this stage, until we start seeing the LGA format being used widely by big data analysis players (Gephi in particular), I say it's not a huge priority now.

I can see why it could offer some useful benefits for, say, a D3.js visualization, but I don't see a tonne of work out there playing with derivatives in this way yet.

OK, I think we can probably close this now (removed stupid link).