1/ Many dapp tokens are unnecessary, add friction, & were created just to sell in ICOs.
2/ Many dapp ICOs were likely unregistered illegal securities offerings & their tokens are likely securities. We also learned from the SEC that ICO classification != current security classification (see ETH). ICO tokens that look OK may not be in the clear from U.S. regulators.
3/ I believe crypto & games are a great combination -- trading & tracking scarce game item ownership w blockchains & tokens makes a ton of sense. The scarcity of rare game items implies the corresponding tokens should have at least some value while the game is popular.
4/ If ‘optimal’ cryptocurrency security is a winner take all or nearly all model (similar & related to the winning crypto SoV), then the security of the winning blockchain will be magnitudes larger than all others.
5/ Thus, why secure a valuable item (like a $100m game item) on the #2 chain when the ratio of #1 / #2 blockchain security skyrockets? 
6/ Perhaps because of high fees before a robust Layer 2 for smaller value items & transactions.
7/ Even more nefarious gambling & high risk dapps risk limited adoption if there is concern that pot sizes could outweigh the cost to break the security model of the network via buying stake, hash rate, paying miners, etc. (see Fomo3D)
8/ Thus, dapps securing high value stakes should conglomerate around the most secure chain in the long-run.
9/ Not only should all tokens & dapps leverage the most secure infrastructure they can for security, but using consistent in-game currencies also helps tremendously with UX.
10/ Just as Ebay does not force users to buy 'Ebay coin' to use their marketplace, dapps should not force users to buy their own native token to use their marketplace.
11/ Many Ethereum dapps require ERC20 tokens to perform functions within the dapp that could be done instead just with ETH. The result would be much cheaper and easier.
12/ Some dapps (see Dexs, Augur), don’t require native dapp tokens for usage, and some tokens give access to additional optional functionality like governance or income, but usage is not as open for all dapps.
13/ If there exists sound rationale as to why the dapp token provides more benefits than negatives in the system, then of course the addition of the token could be worthwhile.
14/ But in many cases, the token seems to be the direct result of needing to sell something through an ICO to raise capital & trying to back into a use for the token later.
15/ Forcing a token to be used as the currency within a marketplace can create price appreciation, as many tactics can in the short-term. But short-term trading != long-term investing.
16/ If there is strong buyer demand for a good with limited supply, a forced intermediate token that also has limited supply will almost certainly be pushed up.
17/ A good example of many of these ideas is the project Decentraland -- a project that had an ICO & 2 has Ethereum tokens (non-fungible LAND & the fungible ERC20 MANA).
18/ LAND represents a scarce asset -- real estate in a new VR world with a bounded map. MANA is the currency you must use to buy LAND or list LAND for sale in.
19/ MANA was sold in an ICO. The ICO raised $24m & some MANA was allocated to Decentraland. (see bravenewcoin.com/insights/decen…)
20/ The MANA sale undoubtedly gave the Decentraland team years of runway (hopefully more than a couple if they sold their crypto & manage their treasury well), but at what expense? They have now done an ICO & MANA exists.
21/ Concerns on the ICO front (around proper registrations, each investor, etc.) are more obvious, but on the token front: in the long-run, would Decentraland be better with just LAND & no MANA, and should LAND be secured by whatever blockchain has the best security?
22/ To replace an ICO, Decentraland could have raised funding by selling traditional equity (for $) to VCs (accredited investors), in an undoubtedly more legally clear fashion.
23/ Revenue could be generated by charging transaction fees on LAND buys/sells, renting LAND, etc. & profits could be distributed back to shareholders (via dividends), in an undoubtedly more economically clear fashion.
24/ To replace MANA, Decentraland could denominate LAND in ETH, leading to an undoubtedly more clear UX for buyers & sellers.
25/ In the long-run, ETH should be replaced with whatever the leading cryptocurrency is & valuable LAND (& all valuable assets) should be undoubtedly be secured by the most secure blockchain.
26/ Right now, Decentraland (and many other dapps) has a strong idea, tech, & 1 token (LAND), and a questionable legal standing (ICO) & 2nd token (MANA)
