n.b.: This is more rant and less article. If you expect structure or coherence perhaps this will not be the experience you want tonight.
I believe I must have gotten dumber, for the machines seem to now be smarter. Before they had no ability to recommend, sort, tag, and partition. Now, they Just Work (TM). This simplicity and clarity allows for many to create, consume, and engage. However, I find myself frustrated beyond what can be considered reasonable.
Consider for example the brilliant product Apple Music. It seamlessly combines your existing library with a massive streaming library which allows you to have the best of both worlds. At some point it decided for some of my files that its version was better than my own and replaced it in my library. Now they were likely right! Their file was likely a higher bitrate and probably sounded better, but when you replace an explicit song with a clean one it can be jarring. Especially when the album is Eminem and half the words are gone. However, this was likely my fault as I probably selected the wrong option during setup, or perhaps my file was labeled “wrong.”
Have you ever tried to put a picture in a Word document? It goes exactly where you put it, but later you realize that you should have put it somewhere else. Be careful moving it or your entire document structure may go upside-down. There’s an easy way to fix it by changing the text wrapping of course.
I code for my day job and for fun. However, I will admit that I do not always use the appropriate PEP-8 formatting, however, we are in luck because my IDE has an autopep8 extension that takes care of it for me. Except for the day that it moved an import statement, created an import loop, and it took me hours to find and fix. But hey that was my fault, I should have known I was not following PEP-8 formatting with that import statement moved down like that.
I have found that Netflix gives me a new sense of urgency to watch the shows and movies that I love, because I always have a sense that they are short for this world. Weirdly enough many of them are on DVD right across the room and could be watched whenever.
For some reason this blog now struggles to hold my attention, while my Tweet volume has remained incredible. Luckily I hope I have learned from Shakespeare and kept my wit despite the brevity.
I carry around a slab of glass and metal that is more powerful than machines I once dreamed of owning. It keeps me informed of exactly what I have told it I want to be informed about. I find myself needing to shut it off.
PyTorch Hub is an awesome new step forward in research reproducibility for machine learning and artificial intelligence. Makes it super easy to publish your pre-trained research models and for others to download them and test them. I was shocked how uncommon it was for people to publish their models when I first started reading machine learning papers.
Social media often sucks. The social internet is a magical place full of rich relationships, new connections, intriguing ideas, and true community.
What do I mean when I say the social internet? It is all of the internet designed for sharing, connecting, and engaging with other people. This may include social media, but it also includes personal blogs, forums, discord servers, random IRC channels. In short, places where people congregate around shared interests.
A fascinating look at how the microbiome may affect drug metabolism. Important to remember that the game does not end at pharmacogenomics and we need to be paying attention to the complex interplay of numerous complex systems to understand drug action.
Thanks to the ‘magic’ of deep learning we may be able to better predict which patients are going to respond to immunotherapy in gastrointestinal cancer with cheaper tests. More people treating their cancer certainly sounds good to me.
The gene that was CRISPR-ed in those Chinese babies makes it more likely you die. This isn’t even accounting for the potential off-target effects. Turns out the thing we all knew was unethical is in fact unethical. Who da thunk?
Google did that thing they do again where they make vast steps in artificial intelligence and machine learning. Efficiency of these image recognition networks is up to 10* greater. Most of this gain is because they try to use “scaling coefficients” so that the network scales in a predictable way. I’m just mad because it’s a TensorFlow model and not PyTorch so I can’t drop it into any of my existing image recognition notebooks.
Another Google blog post about how they are doing incredible things. Man what I wouldn’t do to work for Google Brain. (This research is also being done at the same university I am doing my capstone with, so hey maybe they can sneak me in) Okay so in this article they describe a state of the art result for predicting lung cancer using improved volumetric predictions of CT scans. They instead of looking at individual slices in the image are instead reconstructing 3-d structures to improve the accuracy. This both is and is not a crazy step forward. Being able to use the 3-d structure seems to be truly revolutionary, but some of the radiologists performed equally as well as it. Seems that it will be a useful assistance tool for now.
I promise this won’t be all Google, but again what they are doing right here is incredibly cool and a little bit scary. They have found a way to approximate the 3-d size, shape, and depth of moving people even when the camera is moving. This work has really cool implications for AR and VR and a little bit terrifying uses for a potential police state. There are many places where face recognition has been banned or people are considering banning it, however, combining a 3-d map of a person with existing effective identification techniques like gait tracking can serve as a proxy for facial recognition in those areas. Combined with facial recognition it could provide an even stronger match limiting false positives, and avoiding false negatives.
Okay we are finally away from Google, but into something even more terrifying. This neural network when fed a small sample of speech is able to generate a qualitatively accurate facial guess. The model seems quite adept at identifying both race and gender. Scary stuff.
This fun parody site created by Joshua Davis, Kyle Gibson, and the pseudonymous Cas Piancey mercilessly lampoons the tomfoolery of Kik’s attempt to challenge the SEC. For the record, I do not think promising an Ethereum public DApp and delivering a one node Stellar fork is a good thing.
Bitfinex is still being hesitant to hand over documents to the NYAG. They have struggled to get access to documents relating to the transfer from Tether to Bitfinex, and this suggests to me that either the documents don’t exist or there is a very good reason they are not being shared.
In 2015 Tether openly admits to exchanging Bitcoins for Tethers without KYC. Now it is possible, but in my opinion unlikely that they still had sufficient fiat reserves at that point, but I think it is plausible to doubt that and to believe that Bitcoins have often been a part of the backing.
In conclusion: Tether has paid executives dividends out of non-segregated accounts, does not feel a need to keep cash reserves, is buying bitcoins with reserves, and cannot handle a rush to redeem. Their largest claimed transaction is also smaller than multiple apparent redemptions on the blockchain.
There was a recent comment by Joseph Stiglitz in which he expressed his desire to “shut down the cryptocurrencies”. This prompted me to seriously consider how one would go about shutting down or seriously damaging Bitcoin. This is not an easy problem because Bitcoin is designed to be a remarkably resilient system, however, I do think with coordinated international action a significant amount of damage could be done. My proposed plan relies partially on how Bitcoin does difficulty adjustments.
Every 2016 blocks (~2 weeks) the difficulty of Bitcoin mining adjusts so that blocks continue to come approximately every 10 minutes. This is important because the hashrate dedicated to mining Bitcoins has varied significantly over time, and so this helps keep block time relatively stable. However, major swings in hashpower can significantly change the time between blocks.
Furthermore, it is important to remember that Bitcoin mining tends to run with a pretty narrow profit margin, and as such major swings in price can significantly affect the profitability of miners on the network.
So knowing these things, how do we attack Bitcoin? Step 1 is to start buying up old mining hardware. Because mining demands a high level of efficiency to be profitable old generations of miners are rapidly abandoned as miners with access to more efficient technology can reap larger rewards. However, if you do not care about profitability you can acquire these miners. Now make sure you keep them off the network, you do not want people to know that you are acquiring hashpower as it would cause suspicion. Right after a difficulty adjust bring all of your hash online and wait until the next difficulty adjustment.
The next step is to try to negatively affect the price as much as possible in as short of a time as possible. These steps need to occur right after the difficulty adjustment after you brought your hash online. You will accomplish this price with drop with two primary techniques. The first, as a major state level actor you have seized and safeguarded Bitcoin and other cryptocurrencies in the normal course of law enforcement actions. You will now sell it all, or as much of it as humanly possible. Instead of trying to maximize your potential monetary gains from these sales you will instead try to sell them on the lowest liquidity places you can access, with the goal of throwing off indices and inciting further selling from major holders. While you are doing this you will simultaneously try to pull a major source of liquidity from the market. If there exists for example a poorly regulated exchange and stablecoin who combine for a significant portion of the liquidity in the market you will seize them right as you begin to sell. Allow as many people as possible to stampede for the exits.
Now we must do everything we can to reduce hashpower on the network. Turn off all of the hashpower you brought onto the network, and simultaneously coordinate with China to convince them to cutoff as many of their miners as possible. Seize any mining hardware you reasonably can and keep it off the network for now. This drop in hashpower, combined with a major drop in price will cause many previous profitable miners to now be mining at a loss. Many will choose to turn off their machines rather than lose money continuing to mine. The lower the hashrate goes here the less usable the network is. Block times will lengthen and people will grow increasingly frustrated and apt to sell, perpetuating the cycle. Whenever anyone sells mining hardware buy it.
Now you wait for the next difficulty adjustment. This will take longer than it normally will because the time between bloks has vastly increased. However, Bitcoin is a stubborn beast and there are likely some people who have isolated themselves reasonably well from state intervention and will mine to keep it alive. Once the difficulty adjusts again, you will again deploy your hash which should now be even more. If you control the majority of hash you will mine empty blocks making the network entirely useless. The only transactions you will allow through are those meant to help you sell your block rewards. You will sell your entire block rewards with the goal of continuing to push the price even lower.
If you do not control the majority of hash, but do control >25% you will selfish mine and continue to try to identify other miners on the network and seize their hardware.
Regardless of whether or not you are controlling the majority of the hash the next difficulty adjustment should come quick, in less than two weeks. Again you will withdraw your hash and let block times lengthen, but now transactions will be going through, giving desperate people a chance to sell.
Lather, rinse, and repeat. Eventually you will control the majority of hashpower and once this happens you will force the community to make incredibly tough decisions. These could include changing the hashing algorithm or changing how difficulty adjustments work. These decisions are likely to be contentious and will therefore further fragment the community. If these contentious issues result in forks you will see people on each side of the fork trying to dump the other side of the fork, further depressing the price. Even if some vague semblance of a Bitcoin is left standing after this attack it will be a shell of its former self and you will have demonstrated to people that it is more vulnerable than they have ever believed. Enjoy your continued monopoly on money printing.
Note: Typos have been fixed from an early edition of this article
Many Bitcoin proponents, chief among them Nick Szabo, laud Bitcoin for its social scalability. In order to discuss this issue fairly we must first define social scalability. In the famous blogpost where it was first used, Szabo defines it as:
“Social scalability is the ability of an institution –- a relationship or shared endeavor, in which multiple people repeatedly participate, and featuring customs, rules, or other features which constrain or motivate participants’ behaviors — to overcome shortcomings in human minds and in the motivating or constraining aspects of said institution that limit who or how many can successfully participate. Social scalability is about the ways and extents to which participants can think about and respond to institutions and fellow participants as the variety and numbers of participants in those institutions or relationships grow. It’s about human limitations, not about technological limitations or physical resource constraints.” (http://unenumerated.blogspot.com/2017/02/money-blockchains-and-social-scalability.html)
First, I must acknowledge that there are significant benefits to Bitcoin’s design that enable social scalability. Among these are the expense required to censor a transaction, the prevention of double spend without a centralized entity, and the issuance of rewards without a central entity. Each of these has contributed significantly to the success of Bitcoin and are what make it such a compelling piece of technology to me. However, certain design decisions have created a significant and hard to rectify argument surface that may limit future growth. The most important of these, in my opinion, is the choice of a finite, hard cap.
Challenging this hard cap is challenging many of the fundamental ideas held by Bitcoiners and as such I’ll belabor certain points in order to ensure they’re addressed thoroughly. First of all there is a conception among Bitcoiner’s that inflation is inevitable in our modern fiat system, and that this inflation will be bad either for them individually, or for society as a whole. I am willing to concede among these points that inflation may sometimes be bad for the individual, however I contend it is often still a net-positive. Furthermore, I want to challenge the assumption that a finite supply is useful in reducing argument surface.
As Bitcoin’s are lost to theft, technical mistakes, and deaths the supply will continue to contract as Bitcoin becomes a deflationary currency.. For existing holders this seems to be a positive thing. The more the supply contracts the greater proportion of the total value their investment represents. However, it may still be a net negative if it places an upper bound on total value of Bitcoin. Furthermore, it is valuable to realize that, due to the emission schedule of Bitcoin, a large number of Bitcoins are held by a small number of people. I will not attempt to estimate exactly how many, because it is beyond the scope of this article, but I would estimate 0.01% of the world’s population possess at least half of the Bitcoins that will ever exist (it is likely much less, for statistics go here: https://bitinfocharts.com/top-100-richest-bitcoin-addresses.html). This is an intense concentration of wealth, and as the price of a Bitcoin measured in fiat goes up you will expect significant wealth to accrue to these holders.
This natural enrichment of early holders could be considered fair for them shouldering the lion’s share of the initial risk, and believing in a nascent technology before there was significant evidence it would survive. However, the truth of the matter is that having such disproportionately large early holders makes it harder to convince people to buy in, because the primary benefit to their investment is enrichment of the early investors. Now, the response here would be that these people are still incentivized to buy in, as they will end up capturing a larger share than the later holders, however, a structure depending on convincing people to enrich early holders at the expense of later investors is a structure that has made many people at the top quite wealthy. Even now while we are still relatively early in the long life of Bitcoin, it’s difficult for me to envision mass usage, as most are unwilling to enrich a few solely to gain censorship resistant transactions. However, they may purchase Bitcoin as a speculative asset, but my only response to that is I do not see it as a path to adoption.
Furthermore, with Bitcoin (or any other deflationary currency) widescale adoption would provide the largest holders with an entrenched power base. If it were to become globally accepted in the manner described by the proponents of hyperbitcoinization, then early adopters will obtain incredible wealth, and from that, shocking power. Since they are incentivized to hold that wealth and not to spend or deploy it, the wealth changes hands infrequently. This appears to predispose Bitcoin to create an entrenched oligarchic system.
Next, it’s pertinent to consider the value of inflation. Important to this conception is the idea of a risk curve. The risk curve, which can be gracelessly summed up as a comparison between two assets showing how the change in risk affects the expected return, is important to understanding the said value of inflation: For example, you may choose to switch your excess money from USD (low risk, negative expected return) to equities (high risk, high positive expected return). The value of maintaining the negative expected return for USD is that it incentivizes greater deployment of capital up the risk curve. Investors are willing to take on risks in order to protect their wealth and ensure returns. This capital allows for the expansion of the total economic pie as businesses grow and create new products, new efficiencies, and new markets. However, deflationary money can seriously mess with this contention. If you have a well-established deflationary money then your money will have (low risk, positive expected return), and as such you have little incentive to deploy it up the risk curve. This may seem to be a relatively small and technical matter but it is a significant matter. Hyperbitcoinization would be destructive for society and would result in a regression of economic games to zero-sum along with establishment of an entrenched oligarchy. This may not prevent adoption, but it may affect the argument surface.
My argument rests on, “a relationship or shared endeavor, in which multiple people repeatedly participate, and featuring customs, rules, or other features which constrain or motivate participants’ behaviors — to overcome shortcomings in human minds and in the motivating or constraining aspects of said institution that limit who or how many can successfully participate.” The hard cap on Bitcoin has created disincentives to cooperative behavior. The reduction to zero-sum or net-negative games makes it such that the nature of every interaction becomes competitive instead of cooperative.
There are a couple potential counter-arguments to my points here.
The first many Bitcoiners/Austrians (big overlap there) will turn to is an effect referred to as the Cantillon effect or the injection effect. I am not a true economist, but it can be summarized as the place where money enters a system, has a significant effect, and is likely to enrich those closest to the injection point. There is little, but not zero, empirical evidence for this in traditional central banking systems, but even if we accept that it is a real effect other features of Bitcoin help minimize it. Consider who is closest to the injection point in Bitcoin: the miners. The miners are required to either exchange it for fiat to pay power bills, or purchase power directly using Bitcoin. This cost to produce helps eliminate the disproportionate wealth effect (if it exists) from monetary injection.
Some, Hasu comes to mind, have advocated that instead of removing the hard cap there could be a requirement to move your coins regularly or they will be ‘reclaimed’. I have always considered this idea seriously problematic because of the implications it has for some of the fundamental tenets of Bitcoin. One of the primary tenets of Bitcoin is that your key gives you, and solely you, control of your Bitcoin, and this invalidates that assumption. For those who keep their coins in cold storage it also represents a (slight) security risk to have to access the coins and move them to a new wallet. Additionally, this could destroy the predictability of mining rewards which may change the incentive structure. It seems to me this would more fundamentally change the protocol and argument structure than simply continual issuance.
The argument that creating a hard cap and creating such a cult around the inflation schedule has reduced the argument surface surrounding Bitcoin and in so doing improved its social scalability. This would fit neatly with Szabo’s definition, as it basically limits the participant’s ability to influence the inflation rate. It also helps with the argument that a cap was necessary in order to achieve any social scaling of Bitcoin, because the early adopters would not have been motivated to use it without that cap. This theory does have significant merit, and is even somewhat compelling to me. However, the fact that we are already having regular conversations about the cap suggests to me that the argument surface has not been maximally minimized.
The final argument I’ll address is that modification of the inflation schedule begets greater modification of the inflation schedule. I may have to concede this argument. It is possible that by deviating from the cap we have created a scenario where people will continually advocate for changes to the inflation schedule, but Bitcoin governance is helpful here. Bitcoin relies on what can be termed fork-based governance in which people have the freedom to run exactly what node implementation they choose, miners choose which chain to mine, and exchanges choose which versions to trade. This means that the only way for this inflation schedule to change is with a very difficult consensus making process, which reduces the likelihood of more than one switch (and makes the one switch I want incredibly difficult).
Fundamentally, Bitcoin does solve several important scaling issues by creating irreversible, censorship resistant transactions without a central party. However, the economic model of Bitcoin limits social scalability and mass adoption. It may also be important for Bitcoiners to realize that they may be potentially limited their returns and adoption due to devotion to this hard cap.
Update: I have softened my stance and am now more okay with Medium as a syndication platform. I just encourage everyone to control their content and to the best of their ability control their own traffic.
So I have previously published pieces about the advantages of Medium (here) and later published a piece about the advantages of using it as a syndication platform, which mostly still stands. However, for me personally I will no longer be using it any capacity.
I wrote an article (note this article is not up for reasons) criticizing a problematic cryptocurrency, this article was distributed by Medium curators, and was up for five months before I finally received this email:
You can review the article yourself but I do not think it was gratuitously abusive, and I do not think harmful is a meaningful metric here. If any post that could be harmful to a business is taken down then you are effectively eliminating the ability for people to criticize or review. Once I realized it had been taken down I decided to review their Terms of Service (something I should have done before ever promoting them). They include in their terms this line right here:
This blanket statement made me realize that it was no longer going to be a fortuitous relationship for me to allow them to use my content to advance their platform. I encourage all content creators to control your own domain, control your own hosting if you can (or use one who has committed to being anti-censorship like WordPress), control your own email list, and keep backups of your content. I did all of these things so this takedown has little effect on me, but I worry about the effect it could have on someone else.
Note: This is an old article being moved over from Medium
After taking a broad look at stablecoins as a whole and then Dai in specific, I decided to keep the momentum going and take a deep look at the Basis Protocol. Same rules as I did for Dai, you guys are coming with me as we go through the whitepaper (I may skip things that are super boring, or I think are meaningless marketing speak). Also, always remember I do not edit, I write in one straight run through, and it is often late at night when I’m writing so my thoughts tend to…meander.
First we’re skipping all the stuff about why they think stability is important, because that is not what we are discussing today. Plus, most of it sounds like marketing speak to my ear. I am pulling out this one phrase though, just because it made me incredibly nervous when I read it for some reason. Mainly because whenever I see algorithm here, I imagine the potential for nefarious actors to manipulate it by exploiting weaknesses in the algorithm, however, I need to judge them on their implementation and not my biases.
This phrase is going to piss off the libertarians and “sound money” fans who love that Bitcoin is fixed, but I have always been nervous that Bitcoin’s deflationary nature represented a potential risk for a small number of oligarchs to end up with a disproportionate amount of wealth. (Apparently I’m okay with pissing people off tonight, sweet this is going to get interesting)
For the record it took until page 8 for them to start discussing how this works, and that upsets me. Way too much marketing speak before this. The next section is all about the quantity theory of money which is basically the ability to inflate or deflate.
Biggest problem I see here is the potential for aggregate demand to be mismeasured by one of the Oracles, or however they determine this.
Okay so here I am going to copy and paste my own previous explanation of these different coins from my stablecoins article.
The base shares are created in the genesis block and all new base coins go to them as “dividends” (holy shit I hope they spent some of that $100 million on Securities lawyers). The base coins are a separate coin from the shares that is issued to holders of the base shares when the price of the base coin exceeds peg, in theory diluting their value and bringing the value back to peg. Base bonds are issued when the value falls below peg. When this happens you can purchase a Base bond for the current value so say $.80 and then redeem it for $1.
Now where this gets even trickier is when it comes time to issue new coins is that it goes first in, first out. So it goes to the oldest bond holder first, and then continues down the line, and if there are in theory no bond holders then the coins are issued to the share holders.
Now what is the issue with this system? Economics mostly. Let’s start with what is the most obvious to me, my incentive if I am a whale and if this is liquidity is to become a large share holder, and then whenever I get a dividend to immediately sell it, purchase bonds, wait for them to redeem, and then dump again. If the market gets too big for me to do this alone, I do what all the whales are already doing in crypto and organize this process with a few other people.
Okay let’s discuss these mechanisms for a minute. First one is obviously worthless. One glitch in the feed and you could seriously hurt your monetary supply. Toss it. Second one is probably reasonable. However, both the second and the third one are going to run into the problems that the Ethereum DAO and the EOS launch have had. Mainly most people just do not vote. The incentive may help, but it makes this protocol naturally inflationary, which will require the issuing of more bonds, and I worry that the inflation could start to outrun the system.
This claim that it would require 50% of the voting coinbase is an interesting one. Namely because it is important to remember that likely much of the coinbase will not be voting (see Ethereum DAO and EOS). Thus comparing it to Bitcoin where you need objectively 51% of the hashpower to execute a double-spend attack (closer to 33% for a selfish mining attack) is very different than 50% of the voting coinbase which might be closer to 10% of the total supply. Therefore, it is much more vulnerable than they make it seem here. They think that their incentive system will help increase the percentage who are voting, but it is important to remember much of this coin is probably going to be like other stablecoins where it is held at an exchange and used primarily for short-term trading, so I do not buy that
We already discussed this above.
Already discussed this above too.
Well I kinda want that later discussion, this makes me very nervous. Also means there is always going to be a risk that bond holders are taking on, which is different than the risk-less way they pitch it.
Here we go, they are saying basically what I said. Now, there are other existential risks that occur when demand falls like they are worried about here, mainly centered around the “stability fund” they discuss alter in this paper.
I’m just imaging what would happen if the United States Government decided to start just defaulting on their debt. (The answer is the rate at which people would lend them money would skyrocket as their credit rating plummeted, and the USD would likely lose it’s place as global reserve currency, while it inflates in order to try to cover remaining debt) So basically not the analogy you want to evoke.
Intuitively to me it feels like at any point of serious expansion, arising after a long period of stability, it is going to be more beneficial to be a shareholder than a bondholder.
Okay, so contraction is interesting. We’re back to these bonds. My issue with these bonds again, is that you could potentially (though five years is a long time) be left holding a worthless bond. If I am understanding the clearing price part right, they are also basically trying to incentivize bond purchases near the peg, which is actually really clever. I quite like the thought that went into that, because it was initially counterintuitive to me. Basically, you are encouraged to put your order at a high price, because if they need more coins than that, it will actually fill at a lower price than your order, but you are guaranteed to be getting in first. Clever little bit of game theory and incentive setting. Props to the Basis team for that.
Quite frankly, I do not think I am qualified to analyze this part. My one fear is the fitting of GBM parameters to those assets, which generally trend up. I am still worried about potential Black Swan events, but I am not qualified to judge this model setup.
This seems true, if the price is expected to return to peg. Though I am not sure about people taking short positions, because of the cost. There is also the key point here that as long as there is sufficient liquidity, that will come in later when they discuss the “stability fund”.
I just realized they do not discuss their “stability fund” in their whitepaper. In some of their other documents, they say that they will use money raised to provide a stability fund that will provide off chain stability in price. So just like the other stablecoins, they are trading their own coin to maintain the peg. One of the biggest peg breaks for Dai was when their bot doing this trading failed. They also need this to provide the liquidity neccessary for price responsiveness, but in order to maintain the fund they will continually need money coming in. That is….dangerous.
Overall, I would not trust Basis, or any other stablecoin I have analyzed so far. Remember, I wrote this late at night, and in one straight shot without editing, so if you catch any errors please let me know.