• I've mapped one of the road to Ojaveda maps and calculated that it's total area is around 725k square coordinate units.
• I know of 4 outdoor areas that I could see this whole idea of dynamic paths making sense (2 roads to Ojaveda and 2 roads to bronze doors)
• From the Tech Talk we know that the bottleneck currently lies with the bandwidth. So the focus of these deliberations will lie with that.
• Since the server knows where every character is at all times and has CPU and Memory to spare the analysis of Player movements can be done server side
• 1 coordinate unit is about 2-3 times as wide as a character
• Making paths is a rather slow process

• On the basis of the grass shown in the dev blog about 8 degrees for the strength of a path would suffice. 16 would be hardly noticeable and 4 could be a little rough but might still work.
• 1 square coordinate unit looks like na adequately fine grained tile size to grow paths so lets run with that for now.
• Using that tile size we would have a total or around 3 million tiles for all the four road areas. At 3 bits per tile that would make about 1.25 MB (~300kb per road map) for a complete download of the grassmap.
• These grassmaps could arguably be cached for at least a day due to the expected slow build up of paths. That would leave us with an an absolute maximum of 1.25 MB per active player per day
• Aside for standard compression algorithms the size of the grass map could be further reduced by defining areas where grass cannot grow like rocky terrain, very steep slopes or stalactites. All the tiles in those areas could simply be omitted from the grassmap. All that would be required would be a static file client side to indicate which tiles are missing.
• Even finer paths could be extrapolated client side by projecting the grass map onto a 4 times finer grid and using something like a anti aliasing algorithm to smooth the paths

• The weight of any player's steps would have to be put into relation to the amount of inhabitants that can be expected to live in these few and relatively small cities.
• Also instead of weighting the steps of all players the server could simply just monitor only a few random players to reduce the cpu load
• The server could monitor the paths and notify the dev/gm team about paths that remain over a prolonged time so that they can be moved from the dynamic model and into to static map.
• The server would only have to generate the grass map file once for every cycle. And with a simple hash comparison the clients could determine whether they need the new file or not.

Strength
Your physical power. More strength means higher damage in combat and a higher carrying weight.

Light armor
Your proficiency in fighting with no armor or armor made from light materials such as leather.
A higher skill will allow you to dodge more attacks despite wearing armor and block more attacks with it.

Armor Repair
Skill at repairing armor. A higher skill will reduce the chance to degrade the item and allow you to repair more, faster as well as repair more complex items

Yes and no. I don't think it's wanted that you can have 200 poly lines with 200 lines to "overcome the 200 lines issue" which is meant as a "feature".
So assume that each line of a poly line counts as a "line", what are the advantages you hope to get from the "polylines"?