Interesting. Looking at Palmer's claim: "Try saying anything negative about the 五毛, or even mentioning them at all. Your comment will last about 30 seconds and get deleted without warning or notice, CCP-censor style." This seems to be evidence against the broader claim that anything gets deleted, rather than just negative comments.
This sounds like yet another anti-China thread that comes the same day as the new security law for Hong Kong (I'm sure it's just a coincidence).
I wonder: does China comment on repressive laws approved in other countries? Isn't Hong Kong a part of China? Why wouldn't they approve any laws they see fit? Why are the people that are concerned about this never concerned about repressive laws approved in US-friendly countries like Turkey or Saudi Arabia or the Emirates?
And why all of a sudden is everyone so concerned with what China has been doing for decades (something that, while definitely authoritarian, is not exceedingly nefarious either)?
It's a serious question. We all knew how China works, and even if we thought it's something that goes against some of our values, we never considered it bad enough to be a deal breaker.
So what happened that made us all of a sudden become so fixated about it?
> There's no way to instantly convert the stablecoin back into the collateral asset, you have to find someone with a CDP who is willing to buy the stablecoin from you.
This is incorrect and actually doesn't make sense. An important part of DeFi are DEXes (decentralised exchanges) such as Uniswap and Kyber. There is a liquidity pool where you can instantly buy/sell assets. There is no 'waiting' to find someone (i.e. no problems with coincidence of wants). You definitely do not need to find someone with a CDP. You can hold DAI without opening a CDP. Contracts can hold DAI (as DEXes do).
> Similarly, there's no way to unlock your collateralized asset, you have to find someone who is willing to sell you the stablecoin in order to open up your vault and get your eth out.
Also incorrect. You need to pay back the debt of your CDP with DAI. You can do this easily by buying DAI on a DEX, paying back the debt, then releasing your collateral. Some services exist to do this in 1 transaction, so you don't need to actually 'buy' any other asset. You just send the transaction to a contract and they take care of the details.
The DeFi, specifically the Ethereum space, has moved very quickly in a short amount of time, so there are a lot of new concepts and instruments out there. I think a more traditional finance person will have trouble understanding it all as in some cases, there are no analogies to the traditional finance system (e.g. flashloans).
> I think a more traditional finance person will have trouble understanding it all as in some cases, there are no analogies to the traditional finance system (e.g. flashloans).
All "cryptocurrencies" are, in effect, complex securities that are derivatives of greater market factors. The crisis of 2007 showed the risk of using complex investment vehicles that were poorly understood by investors -- and the CDOs that were sold in the mid 00s were far more transparent and predictable than cryptocurrency. Given that crypto will always be one asset class among many, if the traditional finance system (which already operates outside any single fiat currency) can't understand it, they won't use it for anything more than speculation.
Using crypto as an investment vehicle requires a reasonably accurate assessment of risk. In the case of fiat currencies, central banks manage that risk so that investors can rely on the liquidity of the overall system without wiping out deposits. DeFi has no such mechanism, and no central bank to absorb a big hit temporarily in the case of a black swan event.
I still feel that cryptocurrency is just the 21st century version of penny stocks and junk bonds. Fiat currency works because its power as currency is secured by a government able to mobilize military and industrial power to solve market problems. The most heavily traded currencies (RMB, USD, EUR) are those backed by large industrial and military powers because those countries have the scale and political power to manage market risk. Centralized governance is a feature, not a bug.
If you hold Dai without opening a CDP, the only way to convert that Dai back to Eth is to find someone who opened a CDP and is willing to buy from you.
Similarly, if you opened a CDP and then sold the Dai, the only way to get your Eth back out of the CDP is to find someone who is willing to sell you Dai. The Maker liquidity crisis yesterday happened because there were more people trying to scoop up Dai and get their Eth out of their CDPs than there were people selling Dai, which meant that a lot of people were stuck holding leveraged positions on Eth that they couldn't exit. Even worse, the auction system was malfunctioning, so it appeared as though those people may not even get a fair value for their Eth if they did get liquidated (not to mention, they'd also have to pay the 13% fee for being liquidated, even if they got a fair price in the auction).
> An important part of DeFi are DEXes (decentralised exchanges) such as Uniswap and Kyber. There is a liquidity pool where you can instantly buy/sell assets.
This only works if the total number of buyers and sellers are balanced. The way Maker is set up right now, its possible for a large percentage of your Eth pool (or Dai pool) to be completely unavailable because the holders have not listed the Dai on an exchange. Simply having a decentralized exchange does not automatically guarantee liquidity - people have to agree to sell their assets on that exchange.
> You can do this easily by buying DAI on a DEX
The entire problem yesterday is that all of the Dai was scooped up from all of the Dexes. There was a period yesterday where the Dai price was >$1.11, meaning that CDP holders were paying an 11% premium to exit their positions. The Maker system had no exit valve for people stuck in this position.
> The DeFi, specifically the Ethereum space, has moved very quickly in a short amount of time, so there are a lot of new concepts and instruments out there.
There are also a lot of old concepts and well understood financial relationships that are being ignored, and a lot of highly predictable failure modes that are being forecasted as "black swans". The best team is likely composed both of people who have a very solid background in cryptocurrency as well as people who have a very solid background in traditional finance.
It's important to note that the vast majority of people who will own and use Dai have never and will never interact with the Maker protocol or CDP holders. You trade ETH to Dai on Uniswap and go about your day. You trade the Dai to ETH on Uniswap later. I'd expect most people just to go from fiat straight to Dai and back as well.
Maker itself is a system for the professionals and most users won't even know it exists and won't have to.
This is great and potentially solves a challenge i'm having right now working with geth. However, I'm not comfortable sending private keys to your API end point. Do you guys have a solution to solve this concern? I'm sure others would probably feel the same...
create a separate wallet on myEtherWallet and fund it with a little bit of eth to run the calls that you're making. you'd have to be stupid to send the private key of a wallet with a non-trivial amount of eth to an endpoint of a closed source, un-attributable API. which again limits the usefulness, imo.
Thanks! The private key concern is sensible. Our plan is to open source the code and make the hosted version auditable in real time to prove the code being run is 'honest'. Bit of work to do on this, still. I'll post an update when we get there.
There are many of these coming out. Mobius (https://mobius.network/) is one that provides very easy API to create smart contracts and apps, which you can choose to post on for their live DApp Store (https://mobius.network/store). It's based on Stellar but other ERC20 tokens like Credo/Bitbounce are using it now. Here's a link to the docs: https://mobius.network/docs/
I think the issue is you need parallel corpuses of (sentences with grammatically errors) -> (same sentence but correct). Gutenberg has a lot of only the latter I think.
The author can probably get a lot of traction by doing random perturbation, but they won't be getting things in the correct ratios (i.e. lots of dropped hyphens but not many dropped apostrophes) and also won't probably make all the types of errors that humans actually make. It will work, but a huge part of machine learning and doing NLP is getting those ratios right. This is one reason Google slays with their translator, they have huge corpuses that allow them to get fine grained distinctions in their models.
This is great. Well done to OP! I've been considering creating something like this in the future as well.
You mentioned that you used the lyrics from 50,000 rap songs. Did you have any filters for quality or certain sub-genres? What era were the rap songs from? Maybe the 'quality' would be higher if the data set was narrowed to what you consider quality hip hop lyrics.
Also from your post: "You can see that D-Prime does internally rhyme to a small extent, but making the model better in this regard seemed like it was going to be tough, and I didn’t pursue it."
What were your main challenges to improving/pursing better internal rhyme schemes?
Hey man, thanks! I didn't try to filter to any subgenres, but I did try to give the model as easy a time as possible by trying to stick verses which stuck to the most common words (lowering the perplexity of the corpus). I would call this genre 'mediocre', and may explain why I had such a hard time with the sexism thing.
Internal rhyming / assonance and other stuff could be done by rewarding the model for using it during the beam search. I suspect it's going to be hard to tune how to set the reward so it still makes sense, it only just makes sense already. I also think at this point you would really need to sit down and figure out what rhymes and what doesn't. I did the quickest, hackiest thing I could think of, which was good enough for generating end rhymes, but if you want to have internal rhymes and not massively constrain the model, I think you need to work it out purely from analysing the syllables.
For 'm of n transactions', you can choose to have multiple signatures required to execute a 'contract'. This can be 2 out of 3 signatures, or 10 out of 100 signatures, what ever you decide to set.
This is as good (and better) than current escrow services as it is decentralised, between the contracting parties only, and can have very fine grained conditional statements or scenarios before a 'contract' is executed.
In the example you gave, the escrow could consists of you, me, and a computer program. You sign to say you've sent it, the computer program I used to open it signs it, and I also sign it when i view it. So in this scenario, you could get paid when 2/3 parties have signed, or have a staggered payment structure of 50% when 2 have signed, then the other 50% when the remaining person has signed. This would be escrow, but escrow based on the quality of a digital good.
Bitcoin is not anonymous and not meant to be. It never will be and shouldn't be to gain widespread adoption.
For the average person, I can list a few reasons:
- near immediate transfer of money across international borders
- close to zero fees or greatly reduced fees when transferring money internationally
- safe and secure store of value (for the unbanked)
Then there are the future killer apps that will be created
- tipping on the internet (changeTip / Coinbase)
- programatic money
- smart contracts
- pure play digital banking
- ??
It's hard to see how bitcoin could apply to the general public sitting at your computer in a first world country.
But keep in mind there are nearly 5 billion unbanked or unserved people in the world where bitcoin (or similar) will change their life.
Fast transfer, agreed. Cost per transaction is currently between 10 and 20 dollars and has been close to 100 dollars. I don't know what a wire transfer costs but I don't think the advantage is huge if you are only transferring a couple hundred or a few thousand dollars. Safe and secure store - taking malware, hacked exchanges, price fluctuations and the like into account, I am not sure it looks really bright. Future killer apps - let's talk again, when they are real. And tipping, for example, does not look like it could work - transactions are way to expensive to be used to tip a dollar. And when it comes to unbanked people I am not sure that it would be better or easier for them to adopt Bitcoin instead of a traditional banking system.
9 hours later it is still there... ️