June 24, 2017

Bitcoin is unstable without the block reward

With Miles Carlsten, Harry Kalodner, and Matt Weinberg, I have a new paper titled On the instability of Bitcoin without the block reward, which Harry will present at ACM CCS next week. The paper predicts that miner incentives will start to go haywire as Bitcoin rewards shift from block rewards to transaction fees, based on theoretical results that closely match up with findings from our new Bitcoin mining simulator.

Bitcoin provides two incentives for miners: block rewards and transaction fees. Currently the vast majority of miner revenues come from block rewards, but in the long run they will come primarily from transaction fees as block rewards dwindle. This design decision has been discussed a lot, but in terms of monetary policy and hardly ever in terms of security. There has been an implicit belief that the transition to transaction fees will not affect the security and stability of the block chain, and in particular that it is immaterial whether miners receive (say) 25 bitcoins as a fixed reward or 25 bitcoins in expectation via transaction fees.

We reexamine this assumption in our paper, and our findings make disturbing news for the future security of Bitcoin and many other cryptocurrencies. Our key insight is that with only transaction fees, the variance of the miner reward is very high due to the randomness of the block arrival time, and it becomes attractive to fork a “wealthy” block to “steal” the rewards therein. [1]


The figure shows a scenario where forking might be more profitable than extending the longest chain. See the paper for a full explanation.

Here’s how things could go wrong. Due to the possibility of profitable forking, the default strategy is no longer best; we lay out a menagerie of interesting and bizarre strategies in the paper. The most worrisome is “undercutting,” where miners capture as little of the available transaction fees as they can get away with, leaving the rest in the pool as an incentive for the next miner to extend their block rather than a competing block.

We also show rigorously that selfish mining gets worse when block rewards are replaced by transaction fees, motivated by the following intuition: if you happen to mine a new block just seconds after the last one was found, you gain nothing by publishing, so you might as well keep it for selfish mining in case you get lucky. The variance in transaction fees enables strategies like this that simply don’t make sense when the block reward is fixed.

If miners switch to these deviant strategies, the blockchain will be much less secure because of the mining power wasted due to constant forking, undercutting, and withholding of found blocks.

We derive most of our results in two separate ways: analytically, i.e., using game theory, and with a new mining simulator that we created. This gives us added confidence in our findings. For example, in one setting, the theory predicts a rather grotesque equilibrium involving on the Lambert W function, with the proof running to several pages. Sure enough, in our simulations of the same setting, the Lambert miner does best. We hope that our analytical techniques as well as our simulator will be useful to other researchers. We have made the simulator code open-source.

What is the impact of our findings? The Bitcoin community will probably need to respond to this problem in the long run, potentially via a fork, to discourage deviant strategies. We aren’t predicting that deviant strategies will arise in the short term, and there is a long runway for mitigation steps to be rolled out. The fact that blocks have filled up due to their 1MB limit decreases the variance of transaction fees between different blocks, and this mitigates the problem somewhat, although it is far from a complete and satisfactory solution. For example, at the time of writing our paper, the previous 1000 blocks included per-block transaction fees ranging from 0.03 BTC to 4.51, with a mean of 0.49 and standard deviation of 0.25 (over half the mean!). So simply maintaining the block-size limit probably won’t resolve the underlying issues.

At a deeper level, our results suggest a fundamental rethinking of the role of block rewards in cryptocurrency design. The prevailing view is that the block reward is a necessary but temporary evil to achieve an initial allocation of coins in the absence of a central authority. The transaction-fee regime is seen as the ideal steady state of the system. But our work shows that incentivizing compliant miner behavior in the transaction fee regime is a significantly more daunting task than in the block reward regime. So perhaps designers of new cryptocurrencies should make the block reward permanent and accept monetary inflation as inevitable. Transaction fees would still exist, but merely as an incentive for miners to include transactions in their blocks.

One final point: there is a science of designing economic incentives so that rational players will behave in a desired way, and it’s called mechanism design. Creators of cryptocurrencies (as well as creators of applications such as the DAO) are essentially doing mechanism design. But mechanism design is hard, and our paper is the latest among many to point out that the mechanisms embedded in cryptocurrencies have flaws. Yet, sadly, the cryptocurrency community is currently disjoint from the mechanism design community. That is why I’m thrilled that mechanism design expert Matt Weinberg, who’s behind all the sophisticated theory in our paper, is joining Princeton’s faculty next semester. Expect more research from us on the mechanism design of cryptocurrencies!

[1] The problems uncover arise not because transaction fees may arrive erratically, but because blocks inevitably arrive unpredictably.  We model transaction fees as arriving at a uniform rate. The rate is non-uniform in practice, which is an additional complication. This is a theme throughout our paper: we show that undesirable behaviors will arise even in simplified, “clean” models. This is bad news both because we think things will probably be worse in practice and because we want cryptocurrency mining games to be analytically tractable. Our work shows that in a transaction-fee regime, predicting behavior will be fiendishly complex.

Update: see also Bryan Ford’s response to this post (and paper).

Language necessarily contains human biases, and so will machines trained on language corpora

I have a new draft paper with Aylin Caliskan-Islam and Joanna Bryson titled Semantics derived automatically from language corpora necessarily contain human biases. We show empirically that natural language necessarily contains human biases, and the paradigm of training machine learning on language corpora means that AI will inevitably imbibe these biases as well.

Specifically, we look at “word embeddings”, a state-of-the-art language representation used in machine learning. Each word is mapped to a point in a 300-dimensional vector space so that semantically similar words map to nearby points.

We show that a wide variety of results from psychology on human bias can be replicated using nothing but these word embeddings. We primarily look at the Implicit Association Test (IAT), a widely used and accepted test of implicit bias. The IAT asks subjects to pair concepts together (e.g., white/black-sounding names with pleasant or unpleasant words) and measures reaction times as an indicator of bias. In place of reaction times, we use the semantic closeness between pairs of words.

In short, we were able to replicate every single result that we tested, with high effect sizes and low p-values.

These include innocuous, universal associations (flowers are associated with pleasantness and insects with unpleasantness), racial prejudice (European-American names are associated with pleasantness and African-American names with unpleasantness), and a variety of gender stereotypes (for example, career words are associated with male names and family words with female names).

But we go further. We show that information about the real world is recoverable from word embeddings to a striking degree. The figure below shows that for 50 occupation words (doctor, engineer, …), we can accurately predict the percentage of U.S. workers in that occupation who are women using nothing but the semantic closeness of the occupation word to feminine words!

These results simultaneously show that the biases in question are embedded in human language, and that word embeddings are picking up the biases.

Our finding of pervasive, human-like bias in AI may be surprising, but we consider it inevitable. We mean “bias” in a morally neutral sense. Some biases are prejudices, which society deems unacceptable. Others are facts about the real world (such as gender gaps in occupations), even if they reflect historical injustices that we wish to mitigate. Yet others are perfectly innocuous.

Algorithms don’t have a good way of telling these apart. If AI learns language sufficiently well, it will also learn cultural associations that are offensive, objectionable, or harmful. At a high level, bias is meaning. “Debiasing” these machine models, while intriguing and technically interesting, necessarily harms meaning.

Instead, we suggest that mitigating prejudice should be a separate component of an AI system. Rather than altering AI’s representation of language, we should alter how or whether it acts on that knowledge, just as humans are able to learn not to act on our implicit biases. This requires a long-term research program that includes ethicists and domain experts, rather than formulating ethics as just another technical constraint in a learning system.

Finally, our results have implications for human prejudice. Given how deeply bias is embedded in language, to what extent does the influence of language explain prejudiced behavior? And could transmission of language explain transmission of prejudices? These explanations are simplistic, but that is precisely our point: in the future, we should treat these as “null hypotheses’’ to be eliminated before we turn to more complex accounts of bias in humans.

Can Facebook really make ads unblockable?

[This is a joint post with Grant Storey, a Princeton undergraduate who is working with me on a tool to help users understand Facebook’s targeted advertising.]

Facebook announced two days ago that it would make its ads indistinguishable from regular posts, and hence impossible to block. But within hours, the developers of Adblock Plus released an update which enabled the tool to continue blocking Facebook ads. The ball is now back in Facebook’s court. So far, all it’s done is issue a rather petulant statement. The burning question is this: can Facebook really make ads indistinguishable from content? Who ultimately has the upper hand in the ad blocking wars?

There are two reasons — one technical, one legal — why we don’t think Facebook will succeed in making its ads unblockable, if a user really wants to block them.

The technical reason is that the web is an open platform. When you visit facebook.com, Facebook’s server sends your browser the page content along with instructions on how to render them on the screen, but it is entirely up to your browser to follow those instructions. The browser ultimately acts on behalf of the user, and gives you — through extensions — an extraordinary degree of control over its behavior, and in particular, over what gets displayed on the screen. This is what enables the ecosystem of ad-blocking and tracker-blocking extensions to exist, along with extensions for customizing web pages in various other interesting ways.

Indeed, the change that Adblock Plus made in order to block the new, supposedly unblockable ads is just a single line in the tool’s default blocklist:

facebook.com##div[id^="substream_"] div[id^="hyperfeed_story_id_"][data-xt]

What’s happening here is that Facebook’s HTML code for ads has slight differences from the code for regular posts, so that Facebook can keep things straight for its own internal purposes. But because of the open nature of the web, Facebook is forced to expose these differences to the browser and to extensions such as Adblock Plus. The line of code above allows Adblock Plus to distinguish the two categories by exploiting those differences.

Facebook engineers could try harder to obfuscate the differences. For example, they could use non-human-readable element IDs to make it harder to figure out what’s going on, or even randomize the IDs on every page load. We’re surprised they’re not already doing this, given the grandiose announcement of the company’s intent to bypass ad blockers. But there’s a limit to what Facebook can do. Ultimately, Facebook’s human users have to be able to tell ads apart, because failure to clearly distinguish ads from regular posts would run headlong into the Federal Trade Commission’s rules against misleading advertising — rules that the commission enforces vigorously. [1, 2] And that’s the second reason why we think Facebook is barking up the wrong tree.

Facebook does allow human users to easily recognize ads: currently, ads say “Sponsored” and have a drop-down with various ad-related functions, including a link to the Ad Preferences page. And that means someone could create an ad-blocking tool that looks at exactly the information that a human user would look at. Such a tool would be mostly immune to Facebook’s attempts to make the HTML code of ads and non-ads indistinguishable. Again, the open nature of the web means that blocking tools will always have the ability to scan posts for text suggestive of ads, links to Ad Preferences pages, and other markers.

But don’t take our word for it: take our code for it instead. We’ve created a prototype tool that detects Facebook ads without relying on hidden HTML code to distinguish them. [Update: the source code is here.] The extension examines each post in the user’s news feed and marks those with the “Sponsored” link as ads. This is a simple proof of concept, but the detection method could easily be made much more robust without incurring a performance penalty. Since our tool is for demonstration purposes, it doesn’t block ads but instead marks them as shown in the image below.  

All of this must be utterly obvious to the smart engineers at Facebook, so the whole “unblockable ads” PR push seems likely to be a big bluff. But why? One possibility is that it’s part of a plan to make ad blockers look like the bad guys. Hand in hand, the company seems to be making a good-faith effort to make ads more relevant and give users more control over them. Facebook also points out, correctly, that its ads don’t contain active code and aren’t delivered from third-party servers, and therefore aren’t as susceptible to malware.

Facebook does deserve kudos for trying to clean up and improve the ad experience. If there is any hope for a peaceful resolution to the ad blocking wars, it is that ads won’t be so annoying as to push people to install ad blockers, and will be actually useful at least some of the time. If anyone can pull this off, it is Facebook, with the depth of data it has about its users. But is Facebook’s move too little, too late? On most of the rest of the web, ads continue to be creepy malware-ridden performance hogs, which means people will continue to install ad blockers, and as long as it is technically feasible for ad blockers to block Facebook ads, they’re going to continue to do so. Let’s hope there’s a way out of this spiral.

[1] Obligatory disclaimer: we’re not lawyers.

[2] Facebook claims that Adblock Plus’s updates “don’t just block ads but also posts from friends and Pages”. What they’re most likely referring to that Adblock Plus blocks ads that are triggered by one of your friends Liking the advertiser’s page. But these are still ads: somebody paid for them to appear in your feed. Facebook is trying to blur the distinction in its press statement, but it can’t do that in its user interface, because that is exactly what the FTC prohibits.