For the time for peak hashrate calculation (1 year) I think it is good because it is the time needed for a full rotation of the Earth about the Sun. Yes maybe in the Northern hemisphere, summers in June-August may make mining more costly, but we should encourage decentralization, thus that should be offset by mining in the Southern hemisphere. As for EMP/UFO/asteroid attacks, we have no obligation to protect from them in the protocol, but of course we should adapt to changing conditions. If all continents except Australia get attacked by "aliens" then you are left with a centralized group of miners in Australia, and if the peak hashrate requirement dissipates quickly, they will have no incentive to help the other parts of the world to rebuild and make mining competitive again. They will probably attack the other parts of the world (take out the competition). I don't know why I am still responding to these obvious/troll questions.

What I can work on is to develop an automated system for changing algos. I think we should keep 8 algos, and just change 1 at time in an automated way using a rigorous voting process and deterministically compiled binaries for the algo specification. In order to replace a new algo, we should have 75% vote from all 3 parties: users, investors, and miners. We look back at the last 5000 blocks (1 week), and this is in line with Bitcoin which uses 1000 blocks for fork activation and 10 minute block time. The blocks/transactions used for voting will include the hash of a deterministically compiled binary that specifies the algorithm, so that it is absolutely clear what is being voted on. Also, the weight to assign to the algo should be included (set some limits on this). The source code should be distributed in a way that anyone can exactly reproduce the binary (that's what I mean by deterministic). Miners/validators don't have to work with the same binary, but should be able to prove it's the same algorithm by comparing their source code with the source code that we distribute for producing the deterministic binary.

Once a new algo is activated, the block reward should be equal to the average of the block rewards of the other algos for maybe 5000 blocks, to prevent abuse by people creating new algos just to have easy mining.

If all 3 parties agree to replacing an algo with some new algo that no other chain mines (thus is basically native), then it can happen. It would look bad on Bitmark in my view if this became abused, but at least it is done within a fair system, and the algos can improve in the future. We already have a lot of native mining going on and actually I would propose to replace Cryptonight with the new cryptonight that monero will use in October. That would give us a lot more honest hashing power in my view.

Basically, with mining you want to maximize honest hashing power. A miner is providing honest hashing power if he doesn't attack (double spend, censor transactions...). The expected value of the honest hashing power is h*P, where h is the hashrate and P is the probability that a miner is honest. For native mining, it is not hard to imagine P being greater on average, though h is very small and there can be quick burst attacks where dishonest miners suddenly flood in, so variance is high on honest hashpower and expected value is low when you are a small coin. That's why I think merge mining is safer, so if you're going to propose a replacement algorithm, you should have a justification for why it will increase the honest hashing power.

Another thing I would support is reducing the max block size to 200 kB to prevent spam, allow a fee market to develop, and ensure good decentralization. Bitcoin uses 1 MB per 10 min, so for 2 min, 200 kB makes sense, and Segwit can be added if needed. This would also be useful if we have a user voting system, as currently fee requirements are minimal, so it is cheap for people to create many transactions that support their vote.