[michaelmrose]

1. How do you stop a person who can legally read a book, and thus has the key needed to decrypt it in memory, from obtaining that key and using it to create an unencumbered copy on their hard drive without totally depriving them of ownership of their computer to provide a controlled environment where such copying is impossible?

2. I remember a substantial issue with Etherium wherein someone was able to steal a large amount of currency from other users in 2016. What are the risks of an attacker say stealing a large number of tokens intended to be sold to users from distributor or even the ability to create them? In a decentralized world how do you invalidate them? If you can invalidate them doesn't that mean any seller can take back anything?

This can happen now of course they can steal a copy of your pre production movie but potential to monetize is garbage and effect on your legitimate value proposition may be largely unchanged. In this brave new world their stolen tokens may actually have a legitimate market. Steal 1 million tokens sell all 5 days after release as already read for 10% less than going rate before owner bans tokens from legitimate market.

3. What about the case in which someone breaks into your computer or infects it with malware and effectively loots all your books and music and sells it on craigslist/amazon/facebook. How do you invalidate those purchases without a secondary authority. If you can haven't you recreated central distribution with more steps?

[Sargos]

1. ERC-721 tokens don't replace DRM. Effectively you would buy the book and use any DRM scheme you want as the DRM would verify that you owned the ERC-721 before allowing the media to be used.

2. True, there is always smart contract risk. The DAO hack happened right after Ethereum was founded as nobody really knew what they were doing and tooling was extremely basic. Things are still in the early dial up days for crypto but even now tooling is vastly better, there are dedicated firms for contract/code auditing and the systems being created today are much more robustly designed. With that said, nothing can be certified 100% safe and the only concrete proof of security we have is the passing of time itself. These systems will be new for a while and adoption of this tech will happen slowly over time in the same way the internet grew over two decades into what we know today. There will likely be a system that withstands the test of time and becomes old enough to be trusted by even conservative developers, but until then there will likely be mechanisms built-in where the content creator can destroy and re-create tokens in case of failure. Not every token will be this way and the decision is up to the content creator themselves.

3. This is a problem for wallets in general and lots of people are working on it. The leading solution seems to be social recovery which is where you designate trusted addresses (family members, hardware wallets, friends etc) and configure it to allow your wallet to be recovered if say 3 of the 5 agree that you legitimately lost it. If you ever lose your wallet because you forgot the password or your computer/phone died then you can recover it easily and safely, no complicated and technically hard key backup systems needed which is key for normal non-HN people. You can also set outgoing filters so your ERC-721 tokens couldn't be sent unless it's to a whitelisted address or you verify it with another of your trusted addresses, so even if a hacker got into your wallet they couldn't transfer the tokens and you would just recover your wallet using the system above. It's still early days but Argent is the best example of this.

[michaelmrose]

Are you saying the original owner can destroy the token they already gave you? Regarding social recovery. I wasn't talking about recovery in case of data loss although that is a legitimate concern I was talking about loss through theft.

If everyone suddenly had at least hundreds of dollars of value on their desktop that could be stolen and easily monetized by any software also running on said computer then the entire planet becomes a target rich environment full of badly secured computers. Buy or break into an addon used by 10k people. Steal an average of $200 in digital goods per user and walk away with 2 million dollars.

[Sargos]

>Are you saying the original owner can destroy the token they already gave you?

With most tokens, no. They are yours forever. If the token creator did want to set this up then they can do it though. An example is RealT which tokenizes rental housing. They can revoke a token and re-issue it if they need to (if you lose your token or it gets stolen) and they are up front about this. It is very obvious when a token has a backdoor and most of them do not as it's kind of a scandal when someone does backdoor a token and people don't adopt it.

>Regarding social recovery

I did cover theft in my comment but I'll elaborate. Smart contract wallets are more sophisticated than simply money being stored on a computer. They operate a lot like banks do today with daily spending limits, trusted owners, and the like. With Argent for example you can limit outgoing transfers to non-whitelisted addresses to $100 a day similar to how ATMs are limited in the amount you can withdraw daily. This cuts down on loss due to theft. As soon as you notice $100 missing you would recover your wallet (like replacing your debit card) and the thief wouldn't be able to steal anymore. Also, in case you don't know the technicals, wallets are generally very secure and even if an extension can see your wallet they won't be able to access the funds as they are cryptographically protected and also secured by a password/biometrics. So it's not likely that your funds will be stolen by a run-of-the-mill thief and even if they are successful the loss is limited.

With all that said, a lot of people will likely still use banks to store and use their crypto as those existing trusted intermediaries still work just fine. People just have the option of storing their digital money themselves now.

[michaelmrose]

What is the point of a token if its actually centrally controlled like rental properties? You can't actually use the token to revoke a rental agreement you use a court and the court looks at supporting docs like the rental agreement. At best it acts as official motorization of what the exact agreement is which isn't normally the point of dispute.

Presumably those spending limits and indeed ANY limits are user configurable with a token that could itself be stolen or just used by software instead of a person on the machine in question. If the person can do it the software can after all. For example by faking keyboard and mouse input and keystroke logging the user to steal their password.

We can't secure simple things now I cannot imagine how you can believe we can secure complex things with even higher stakes.

Credit card transactions are tolerable because the financial sector takes in a huge chunk of the value of the economy and writes off losses out of that huge pile of money.

In a decentralized situation where the only money in the kitty is users actual money how does this actually work?

[Sargos]

>What is the point of a token if its actually centrally controlled like rental properties?

This goes down the rabbit hole of crypto a bit and I don't want to drone on too much if it gets annoying, sorry in advance.

Most tokens will not be centrally controlled in the future (I can elaborate if you're interested). During the transition period there will be centralized tokens like RealT which have similar properties to the paper forms we use today but even in that state they do offer improvements such as integration with DeFi products (RealT tokens can traded on Uniswap and will likely be integrated into Maker allowing for using your property shares in a collatorized loan), easier accountability/auditability, and other use cases (Let's go a bit sci-fi and say your car lock will open for you as long as you prove you own the car's ERC-721).

>Presumably those spending limits and indeed ANY limits are user configurable with a token that could itself be stolen or just used by software instead of a person on the machine in question.

This stuff does exist and there's some really good documentation out there that's not just me rambling :) Changing or turning off the spending limits or whitelists comes with a user-defined waiting period, so even if a virus somehow accomplishes this feat you would still be notified and have a day or so to cancel the changes and recover your wallet. Your original example was a browser extension which cannot fake keyboard or mouse input to your wallet. It would need to be a real virus of some kind which is rare on desktops in today's world and basically impossible on mobile platforms. I'm not saying the risk is 0% but it's not like mass numbers of people are going to just get their crypto stolen willy nilly, especially if they use their phones like the majority of the people in the world do.

>Credit card transactions are tolerable because the financial sector takes in a huge chunk of the value of the economy and writes off losses out of that huge pile of money.

Credit cards will still exist in a crypto world. Banks will still exist too. You will be able to reverse those payments just like you can today even if the credit card company backs their operations with ETH/DAI/BTC instead of USD. On top of that, it's actually really easy to develop a payment system on top of Ethereum that contains the ability to reverse transactions for a period of time with complex logic such as different amounts of time for different merchants based on a calculated risk score. For end users this isn't much different at all in the short to medium term.

[michaelmrose]

The point about changes requiring time to implement is well taken and I can even imagine people relying on notifications to catch bad behavior because I see some people doing that now.

Malware isn't impossible on mobile. Android install security is in a fashion crap and androids are 90% of the market worldwide.

OEMs base their installs on old kernel versions that support their custom never to be upstreamed modules needed to boot their board then after 0 through a few updates stop providing updates for their phones. This is so because their is no stable interface for kernel modules.

My own phone is running 3.18 while my laptop is running 5.6.

Ultimately a lot of people are running around with devices that are actually vulnerable to things we already know are broken in addition to the known vulnerabilities upcoming that will be known to attackers long before its known to OEMS.

You can posit a better more secure future but the honest truth is that our present work is crap and we have no particular reason to believe the future isn't also largely composed of crap. It's trivial to imagine that in a situation where you can derive an increasing payout for breaking security that the attackers wont keep pace with those trying to secure the future.

This pessimism has been the correct answer from the moment computers were networked to one another through today. Given that we have been bad at securing networked computers for 5 decades it behooves the optimistic to prove it.

Maybe in 5 years we will all be running devices running Sel4 with only substantially audited code but I would bet on more steaming piles of insecurity instead.

Thanks for the interesting discussion though.