The real problem is how difficult we can reasonably make winter without killing every character who starts during the period. The general problem is that with scaling characters -- you get better the more you play -- comes the need for scaling challenges -- from newb to blaze -- otherwise newbs will die and blaze will get bored. If challenges are to scale -- from newb to blaze -- the challenges must be localized in the game world, or newbs will bump into them when they can't handle them. Dangerous creatures this way, kindergarten that. If the challenges are localized in the game world, the game world must have an up and a down. If the game has an up and a down, there is less sand in the sandbox. Another problem with localization is that one wants to avoid the RoB situation, where the RoB is a blasted wasteland of deforestation, of trial and error, and of early newb mistakes. The reason for the RoB situation is that all spawning is located to one particular point in the world. Perhaps this isn't a problem? Perhaps one could randomly seed the map with dangerous areas. If food is nigh on impossible to get during winter time -- as, it could be argued, it should -- then what about those who can't even fish? The cycle -- Planting during spring, filling up stores during summer and fall, and simply trying to survive during the winter while engaging in simple crafts -- would be great fun for an established player who knows the game, but all players aren't like that. Some don't know the game, and how do I handle those?
I could give them a little world of their own to play with -- with quests and building and whatnot -- before letting them out into the real world. Fine, but then survival in the real world still cannot be dependent on skill levels, as they still wouldn't have too many of those. As it stands, though, it should perhaps be said that survival isn't really dependent on high skill levels. The only thing that depends on skill levels is quality of the stuff you make. You can still make it at more or less any level. I think this is as it should be. There are the learning skills, though. Also, it is unreasonable that newbs should have to go through more or less the entire game content, and then take an exam on the subject, before being allowed to play with others. Learning from others ingame has its charm.
Animals scaling isn't entirely convincing. It leads, as bright minds have pointed out, to a situation where animals follow the normal MMO pattern of being mobs that are to be aggroed. While this has its charm, I quite like killing animals in H&H, it isn't entirely realistic. Perhaps this could be avoided if one changed the model for how characters scale. If a bear is always 50 units of danger, and a newb hearthling is 10, then Blaze (An insanely high level player with the best gear there is to have) could perhaps be 40-50-60 (Assume linear relation between the units)? That is certainly more realistic. Even a trained spec ops navy seal KGB/SAS/Mossad ninja will -- more or less -- get beaten up if there are three or four determined guys attacking him at once in a situation where he can't run. The model would then be that five newb hearthlings who gang up on a bear would have a chance of taking it out, but only a high level player has any real chance of doing it himself. Under a representation like that, where levling basically comes with a steep, steep amount of diminishing returns -- you would probably always consider the bear somewhat of a challenge, and they do not have to scale per se -- though they could still have a bit of natural variation. This also leaves room for lindworms and trolls and leshys and god knows what else. One obvious risk here is that players do not feel that they get rewarded for their efforts. It could, potentially, make character development into a long boring grind. Another problem is that it might make throwaway newb assasins into a problem. People could exploit the relatively slow pace of development. That could perhaps be solved by giving a defender a massive advantage in all forms of pvp. Assuming this relation animals should probably be less frequent by an order of magnitude. A problem with that is that animals are one of the few things that make the world feel alive. Perhaps some simple things, animated patches of grass blowing in the wind, could improve on that.
Perhaps one could have a special flag on newbs, a baby blue eyes effect, which makes animals not attack them. The flag could be removed once newbs learn some critical fighting skill or other. If one had something like that, one could spawn newbs pretty much anywhere in the wilderness. There'd still be the problem of linking up people who want to play with each other, but that could probably be solved through some sort of invite system.
So, what about winter? Assuming a model like that solves the above mentioned issues, could it be dangerous? Maybe. If the seasons gently roll across the map from one end to the other, then winter could be like a murdering polar bear of pain that kills, maims and destroys hearthlings, crops, trees and buildings. Number one this could help clean out abandoned settlements, and keep the map fresh. Number two, you would only really have to play through it if you had some important value or other to protect in the area in question. Number three, newbs could always be spawned in the spring area, giving them a couple of IRL weeks or perhaps even months to learn the ropes and get set up before winter hits them. Number four, if they still aren't up for it they could always move with the season, as humans obviously have done for large parts of their history.
I guess the model would be that if an object gets covered in snow, it will decay more or less in an instant if the decay timer hits it. You would perhaps not have to repair stuff per se, but you would have to clear snow off of it. Also, you'd have to wear warm clothes or die instantly.
Maybe that could work...