I started to watch this thread after around 5 minutes of OP.
After a while it was clear it's not a scam (In the wake of Nxt there were like three annoncements like this a day), but also that it will be a long development. I'm okay with that.
However after the IPO announcement a few weeks ago (or months now?) I don't have any intention to buy the coin at IPO, the risk/reward ratio doesn't look good. I'm still watching because I want to see what technology it will bring, maybe buy some coints at some point, but the IPO format removed any incentive to invest in it.

Skycoin will have a rolling IPO. The IPO will be very small, then a small number of coins will be released on to an exchange each day. If the price decreases, the number of newly issued coins will be decreased to stabilize price. If the price goes too low, the developers will begin buying back coins.

The majority of distribution will go to developers, people involved in marketing and people buying and operating a mesh network nodes. We will distribute between $100 and $150 dollars per meshnet node deployment. If a user spends $200 to deploy a node and receives $150 in Skycoin, this distribution is unlikely to negatively impact price because of Proof of Burn. Distribution an exchange or airdrop, in contrast is likely to drive down price and should be minimized.

The problem with Maidsafe is that it does not have users and does not have volume. To succeed, Skycoin must have a large and rapidly growing userbase and there must be a reason for people to be buying and using Skycoin.

Coins that no one uses and which have no user base are purely speculative. Many people have made money from speculative investments, but there is no rational basis for the price to increase over time. Dogecoin and Bitcoin should experience appreciation because of increasing adaption and a growing userbase. If the Bitcoin userbase doubles and Bitcoin transaction volumes doubles, we would expect Bitcoin price to approximately double, or at least increase.

It is very difficult to understand why Maidsafe would increase from the IPO, without a growing userbase.

Just an idea at this stage. Found an improvement. Each Obelisk node, has a public key. We hash the public key into an address and then it stores 10 coins in an output owned by that address.

It does not add a cost. It just proves that you own 10 coins. It proves you know the private key, for a public key, whose address has 10 coins in it. You can still spend the coins.

The idea is that it upper bounds the number of nodes. If 10 coins must be held and there are 100 million coins, then it upper bounds the network at 10 million nodes. The upper bound does not appear to be mathematically useful right now, but is something we should keep in mind.

When a new Obelisk node is run, it will "trust" some random peers. The user can also add a few nodes by hand that it trusts (exchanges or trusted community members). A node is identified by its public key hash and found by DHT. Its not like Bitcoin where nodes are IP:port pairs. You can move your computer around and the identity of the node does not depend on its IP address.

We want the network to be secure with random nodes being chosen. We dont want a situation like Ripple, where the three developers nodes control the network. However, we wanted to prevent a situation, where someone runs 200,000 nodes and tries to collect the trust relationships from new users. These Sybil attacks nodes, still cannot 51% attack generally, but anything that increases the cost of the attack is still useful.

Maybe, we restrict it so that new user will only randomly trust nodes that have a coin balance. Trust relationships wont be severed if the node does not have a coin balance, but they just wont get new random users.

The connectivity graph for trust relationships, is supposed to be a fully connected random graph. A few nodes (trusted community members, exchanges, websites, organizations) will have more trust relationships and that helps the convergence time for block consensus a bit. It reduces the network diameter a bit.  Some nodes will be used to verify consensus (you choose a bunch of exchanges or different public keys), these nodes do not affect consensus decisions, but are "consensus oracles" to check if your node has converged with network.

If two large exchanges have different consensus for a particular, block, that is a problem. It could indicate a netsplit or an attack on the network. Exchanges may want to suspend trading until the issue is resolved.

Why should we trust them?

A dev holding 50% of the coins is a good defence against a 51% attack.