Appetite for Advertising

The chart below is based on a global survey by Statista (n=31,726) of internet users on their attitudes towards …

  1. Acceptance of advertising
  2. Acceptance of using personal data in making advertising more relevant

 A couple of very interesting things to note

  1. Advertising is mostly acceptable in major economies. It’s the use of personal data that is at question.
  2. The biggest economies are more relaxed vs the smaller ones.
  3. China is the most open.


Is regulation coming to a Technology near you?

This blog first appeared on the, here.
Are today’s technology titans exploiting their market position without any attention to broader stakeholder welfare? Essentially, this is the implicit question being discussed in most media today.
Given the recent news flow, it certainly feels like technology has had a rough start to the year. In the past few weeks, they have seen a marked underperformance given negative news flow: Zuckerberg in Washington, Uber‘s self-driving car crash, Tesla recall and President Trump’s tweets against Amazon. Although Google have managed to avoid this recent limelight, they were (not so long ago) slammed a EUR 2.4bn fine for allegedly breaching anti-trust rules.
In my view, these allegations fall in three categories: 1) anti-competitive practices; 2) ownership of content; 3) compromised user privacy. I will tackle the latter two in this post.
Teething problems or poor incentives?
Before looking at the specific situation today, it’s worth taking a step back. This is not something that is happening for the first time.
Technology, very broadly defined, is a tool that enhances productivity. Its development is often undertaken with that sole criterion in mind. This isolation from “real-life” helps innovation and progress. But when used broadly and introduced into the real world, this highly efficient tool faces two major issues a) the tragedy of the commons / prisoner’s dilemma[i] and b) availability to good and bad actors.
This is where we, as a society, agree to ‘norms of usage’. Nuclear technology is great for producing cleaner energy, but it can equally be used as nuclear warfare. Internal combustion engines are great to improve our mobility. As individuals, we have every incentive to make the most use of this technology to improve our mobility/productivity, but society, over time, agrees to norms of usage (such as CO2 emissions).
The internet is no different. It has given us a lot of value (in terms of content and tools) for free, in exchange for advertisements. At the same time, this has been abused by several bad actors – fake content and compromised user privacy are two issues arising from that abuse. Does that mean that the internet and today’s technology titans, riding its coat-tails are all bad? Or does that mean it may be high time to agree to ‘norms of usage’ of this technology? We can also call it regulation.
Part of the reason the bad actors got to where they are today is because of a policy of neutrality practised by these platforms. Why did they do so? A law from 1996 (Section 230, 1996 Communications Decency Act – “No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”) that shelters internet intermediaries from liability for the content their users post. While this made sense in the early days of the internet to ensure innovation and growth, it makes little sense today for platforms, such as Facebook, to shirk responsibility for user content that they store.
technology and society
Will technology come to its own rescue?
Once the poor incentives (for platform neutrality) are adjusted, most likely, yes.
Facebook has repeatedly said that 99% of terrorist content is taken down (supported by Artificial Intelligence tools[ii]) before they are flagged by users. Other platforms are equally capable of using AI tools to flag objectionable content. Such automated response to moderating or taking down ‘bad’ quality content should surely help rebuild trust in these platforms again. This type of regulation (ownership of content) is increasingly being accepted by the likes of Facebook – with Mark Zuckerberg remarking it is ultimately responsible for content. Facebook plan to double the number of people working on cybersecurity and content moderation to 20,000 employees in 2018.
With respect to user privacy and identity, I believe there are two fixes. The first fix is behavioural. Increased scrutiny will make them more responsible. If in the past, Facebook did not restrict access to user data by rogue apps, after the Cambridge Analytica scandal, it will need to be more cautious of such lapses. This means giving users more transparency and control about how, and with whom, their information gets shared. The second fix, in my view, will come from technology itself – Blockchains[iii], and more specifically zero-knowledge proof Blockchains. In layman terms, this cryptography technology enables proving something (in this case user identity) without revealing any information that goes into the proof. This ensures full anonymization of user data, with no link to sensitive or identifiable user data. Combining them with smart contracts (a feature in Ethereum Blockchain) could give users full control over who can access what information. A central authority, such as a government, could issue these. For example, the World Food Program’s (WFP) Building blocks project already uses zero-knowledge Blockchain to dispense aid to Syrian refugees.
Regulation is a double-edged sword
At the end of the day, the regulators will have a difficult job. On the one hand, they want to hold platforms responsible for the content and be responsible with user data thereby creating norms of usage. On the other hand, regulation typically raises barriers to entry. It will make it difficult for smaller firms or new entrants to satisfy those regulatory requirements and possibly restrict their access to data that the behemoths, such as Facebook, already have. This could entrench the market position of the technology titans that it is trying to regulate. If data portability neutralizes platform power, it also exposes the data to abuse by bad actors.
There will be a tricky balance to strike between: safeguarding user abuse and limiting the platform’s market power.
Disclaimer: This is a discussion of broad technology trends and not investment advice. Any investment decisions made are your own and at your own risk. All views, opinions, and statements are my own.
[i] Prisoner’s dilemma is a paradox in decision analysis in which two individuals acting in their own self-interest pursue a course of action that does not result in the ideal outcome. The typical set up is where both parties choose to protect themselves at the expense of the other participant. As a result of following a purely logical thought process, both participants find themselves in a worse state than if they had cooperated with each other in the decision-making process
[ii] If I haven’t already driven home the point about technology’s dual nature, it is worth re-noting that these AI tools that can flag objectionable content can also be used (by bad actors) to create more fake content – text as well as videos.
[iii] Blockchain is a distributed ledger where transactions/activity/information can be recorded chronologically and publicly. Given the use of cryptography to encode the information and “chaining”, it is almost impossible to alter the data retroactively. Interestingly, Blockchain technology is also the biggest threat to established digital platforms due to its ability to democratize “trust” (through decentralizing the record).

bitcoin – Bubble or Beginning? Both! Part II

When I last wrote on bitcoin, the price was somewhere in the vicinity of $4,500. Bitcoin’s price continues to defy gravity. Not just defy gravity, the price has gone up vertically! It has more than tripled in three months since I last wrote. I don’t know of any other asset/currency that has moved up in value so fast….ever.
Is this a bubble? Even if one considers the possibility that this is not a bubble, one is left with the unanswered question what could have possibly changed over the last year (1500%+), month (100%+), or week (40%+) to have warranted such drastic changes in value.
There can be a handful of potential reasons:

  1. there has been a rapid adoption of the bitcoin “currency”
  2. it is in the “design” of Bitcoin system
  3. we are witnessing mass speculation on the bitcoin “asset”

I think we can safely agree the explanation cannot be rapidly growing adoption. To justify a $15,000 value for a bitcoin, assuming it has the same velocity as a US dollar, it would need to represent 6% of all transactions. That is far from true today: even the lead users might not use bitcoin that frequently.
Of course, as discussed in a previous blog, the velocity of bitcoin is far slower and it only represents a much smaller share of transactions. This “slow velocity” or constrained supply of bitcoin is what I believe is responsible for the rapid price increases. Moreover, demand from ICOs and speculators is squeezing that supply.
To cross check this understanding and the design of the network, I researched bitcoin mining economics, and here is what I found. Most of this data comes from Blockchain.infoand is readily available to anyone to replicate these calculations.

  1. Bitcoin is a very poor payment system. The cost of running the Bitcoin network is ~$6.5M/day and the transaction fee covers only $3M/day. The fee doesn’t cover the cost of running the network despite the fee per transaction already being an insane $50+/transaction. Per some observers, the cost per transaction on the bitcoin network is 1000x more expensive vs Visa/Mastercard network.
  2. Mining today (at $15,000 per coin) is very profitable, but only because of the mining reward. The reward for confirming a block for the network is 12.5 bitcoins. This represents today 80% of the revenue that the network makes. With this reward there is a strong incentive to mine coins (75%+ gross margin), without this reward, there is no incentive to mine.
  3. And in case you wonder who pays for this mining reward, it is everyone who owns a bitcoin. Mining reward (today) corresponds to an annual 3.5% tax on all bitcoin owners.
  4. But the long term supply of bitcoin is limited to 21M coins, and the system is designed such that the mining reward for solving the block should half every four years.
  5. Here in lies the the circularity. There are only two ways to keep the mining network running economically
  • Raising the transaction fee further from $50/transaction: That will kill any hope for wider adoption of the bitcoin. An even higher transaction fee per block, which may or may not get spread over more than the 3000 transactions that are within a block today (depends on real world adoption). Energy (or silicon costs) are unlikely to half every four years, as the cryptography problem keeps getting harder over time in order to keep the time to solve a block around 10mins (to offset more and more hash power coming from more miners on the network).
  • An increasing bitcoin price to pay for the networks energy/silicon bill: If energy costs were to stay where they are (which they wont given the every increasing complexity of the cryptography problem to keep time to solve a block at 10 mins) and halving of mining reward every 4 years, that would require the bitcoin price (in steady state) to double every four years. That is almost 20% p.a. increase designed in….. forever!

While trying to further cross check the order of magnitude of these numbers, I came across a paper by Harald Vranken (where he looks at environmental sustainability of bitcoins) which came to similar conclusions but in Feb 2017… bitcoin price has of course changed quite a bit (10x) since then.
Having discussed 1 and 2… we are left with 3. And while it is not a conclusion supported with unquestionable evidence, it is a conclusion from deduction (and a couple of anecdotes). The design of bitcoin network would support “only” a 20% per annum increase in its value. With bitcoins having gained 15x value over the course of 2017, it seems likely the reason for this rapid rise is the result of mass speculation. While I don’t have evidence, I have anecdotes of people either inquiring about investing in bitcoins or being already invested in it. These acquaintances have no background in finance or in investing. Typically, that is a sign of an end of a bull run!
I stand by the not-so-headline making conclusion I drew last time – it is both a beginning and a bubble! While there is value in bitcoin, I don’t know why it should be $15,000! With bitcoin now trading on future contracts, I don’t know which way we are headed next.
What do you think? Please do share your thoughts.
Disclaimer: This is a discussion of broad technology trends and not investment advice. Any investment decisions made are your own and at your own risk. All views, opinions, and statements are my own.
Exhibit 1: The Bitcoin network’s Daily P&L (as of Dec 9, 2017) 

Exhibit 2: Harald’s view of Daily Revenue/Costs of Bitcoin mining (as of Feb 2017). The price of bitcoin has gone up >10x since then…. 

Exhibit 3: Detailed Calculations and sources for Daily Revenues and Costs for the Bitcoin network

bitcoin: Bubble or Beginning? Both!

This blog first appeared on the here.
There is no denying that the recent cryptocurrency boom resembles the one we saw among dotcom companies in the late 1990s. Celebrities then were taking equity stakes in start-ups in exchange for promoting them, and we are seeing a similar trend today[i].
In the late 1990s, over a period of just four months, the Nasdaq Index almost doubled, while over a similar period this year, the value of bitcoin shot up fivefold.
Exhibit 1: Larger than dotcom boom?

How has this happened?

Andy Kessler[ii] at the Wall Street Journal recently poured cold water on bitcoin’s appreciation by suggesting that the currency is possibly ten times overpriced. While I appreciate the simplicity of the analysis and the quick conclusion that it brings us to, I think it is also worth questioning the underlying assumption. That assumption is that bitcoin’s worth is in its utility as a payment service, like Visa or Mastercard (or ‘software as a service’ as the author calls it). Of course, with that logic, the author’s conclusion of its value around $300 is fair (in fact bitcoin’s not that good as a payment service…it takes 10 minutes to complete a block transaction).
But I disagree that that is all there is to bitcoin. It is not just an application, it is an alternative currency. Its value, thus, is in its usage. The more people use it, the more real and valuable it becomes.
I have written previously on what is special about bitcoin. To recap the most important (soft) feature of bitcoin:

Finally, a currency must have some value associated with it. In the case of normal bank notes, the central bank underwrites that value and users trust the central bank (or the government). Somehow bitcoin has gained trust of its users as a store of value and as more people use it, the more that trust grows. In a way, a currency system is the world’s oldest network effects model – a currency has value, because its users agree there is value in it . From that perspective, bitcoin isn’t a fad anymore; its users are willing to trade one bitcoin for almost USD 735.

If bitcoin is a replacement for fiat currency, then the comparison to a Visa/MasterCard payment system is not relevant. Instead, bitcoin should be compared to a fiat currency, where one could argue, trust is in short supply these days. In contrast, the supply of bitcoins is slowing and will saturate at 21 million coins. Then, the value of a bitcoin is determined by the demand for bitcoin-financed transactions. (In comparison, the monetary base for US dollar has grown about 4x over the past 10 years. See here.)

Initial coin offerings (ICOs)

The proliferation of ICOs as a funding mechanism for new ideas created immense demand for bitcoin/ethereum/other cryptocurrencies. ICOs raise money in the form of bitcoin/ethereum tokens (hence generating demand) and in turn issue their own tokens, representing digital equity in the particular ICO.
This is what has boosted the values of cryptocurrencies. If usage or adoption drives the value of a currency, the cryptocurrency has invented its own economy to drive that usage.
Exhibit 2: Illustrative diagram: ICO’s creating demand for cryptocurrencies and boosting their price.

Looking at the rapid rise in the price of cryptocurrencies and the rapid rise in ICO offerings, I would assume causation here. By the end of August 2017, 89 ICO coin sales worth $1.1 billion had been conducted year-to-date, ten times as much as in all of 2016[iii],[iv]. Of this amount, about $500 million was raised in June alone[v]. Participation by “investors” in these ICOs is funded by bitcoin/ether/other cryptocurrencies.
If that is the incremental demand for these currencies, which led to the price rises, it would suggest that the “velocity” of cryptocurrencies is very poor compared to a US dollar (about 25-60 times slower). As such, in a very perverse way, the poor transactional characteristics of cryptocurrencies are most likely[vi] responsible for the rise in their value. By creating a very tight supply of cryptocurrency, even a mere $0.5bn/month demand (for ICO) has been enough to create sky rocketing valuations.
Exhibit 3: Why is bitcoin trading so high?

Note: I have assumed the transactions supported by USD are roughly the size of US GDP ($20T). 

Is it different this time?

To be fair, the demand for an ICO (and hence bitcoin/ethereum) should reflect the quality of the businesses they are funding. I am sure there are some gems that will come out of this mania, just like during the internet boom, but unfortunately most businesses appear to be just riding a euphoric wave. Another mind boggling statistic is that ICO funding took over traditional venture capital funding in mid-2017.
The conclusion is a mixed one: While this is unlikely to end well, bitcoin has established itself as an alternative currency.
Disclaimer: This is a discussion of broad technology trends and not investment advice. Any investment decisions made are your own and at your own risk.
All views, opinions, and statements are my own.
[ii] I am by no means undermining his analysis or the person. While I agree with the conclusion (bitcoins are overpriced, most likely), I just do not agree with the comparison with payment networks.
[vi] I welcome other explanations but I haven’t come across a better one (yet). This is the best I have (for now) come up with.

Artificial Intelligence: “the best, or the worst thing, ever to happen to humanity!”

In March 2016, an AI-based computer program, AlphaGo, beat a world champion, Lee Sedol, in the ancient game of Go. This was a milestone of sorts, and more eventful than the defeat of Gary Kasporov at the hands of IBM’s DeepBlue in 1997. Go is a lot more complex than Chess, with 10130 more possible moves vs chess[i]. This means that a rule-based system would have difficulty scanning all possible responses for an optimal one. AlphaGo was based on a form of “Artificial Intelligence”[ii].
Since then, businesses have rushed to claim their expertise in artificial intelligence (AI) or the advances they expect to see in the near future. In fact, the mention of the phrase “artificial intelligence” during quarterly earnings calls has grown three times[iii] over the last year. A number of businesses connected to AI have seen their stocks dramatically increase in value (Nvidia and Advanced Micro Devices are a case in point).
Is this the fourth industrial revolution?
Deep learning is the ability of a computer program to gain an abstract level of knowledge from churning through a lot of data and eventually apply it in new situations. The computers are not explicitly programmed to follow a rule – the program can use its abstract understanding to take decisions when faced with new situations. While AI concepts have been around since the middle of last century, the easy availability of massive amounts of data and computing power has enabled resurgence in this field again. Andrew Ng’s famous paper in deep learning describes the setup that puts some of these numbers in perspective. To be able to correctly identify a cat in YouTube videos, it took 10million images, 1000 machines (16,000 cores) and 3 days to train the algorithm.
Exhibit 1: Timeline of Artificial Intelligence Technologies (Source: NVIDIA)

While the first two industrial revolutions enhanced our physical prowess (through mechanical energy (water/steam) or electrical energy), the third industrial revolution was already about enhancing our cognitive abilities (IT/Computers: by programming rules-based logic into machinery). This fourth industrial revolution further enhances our cognitive abilities (AI: by removing the rules based logic to enable a certain level of understanding into machinery[iv]).
The shared benefit of each of the industrial revolutions is significantly improved productivity. The idea that a computer program does not need explicit programming and can learn by itself is non-trivial, when it comes to scaling the productivity benefits.
Exhibit 2: AI: The fourth Industrial Revolution

What could artificial intelligence accomplish?
There is a long list of areas where AI can prove useful, no wonder it appears every CEO wants to be AI-ready.
However, I believe we can categorise that potential into the following broad use-cases:

  • Speed: Productivity would be a first order benefit of AI. Because artificial intelligence-based algorithms self-learn, less human effort is needed for automating many tasks leading to significant productivity gains in speeding up routine tasks, such as customer support, transliteration, driving, reading medical reports/ scans.
  • Specificity: The self-learning aspect would mean the algorithm can provide individually relevant experiences to each user. Imagine a time Amazon could predict (based on our shopping behaviour) what and when we need to buy, even before we realise we want to buy it.
  • Serendipity: Finally, given the very scalable computer capacity now available to AI algorithms, a large number of simulations can be run to model very complex systems and finally understand them in ways that we haven’t explored, leading to better understanding. A real example of that was found in AlphaGo’s gameplay that defied 1000s of years of wisdom[v] and made moves that eventually led to it winning against a champion Go player. Another example of such a complex system where our understanding is limited is short term price movements in the financial markets, where multiple participants of varying risk appetites and behavioural biases influence the price on a daily basis.

Exhibit 3: The three types of benefits of AI

Along with the promise come some risks. These risks have been discussed by various observers.

  • The GIGO problem (Garbage in Garbage Out). An AI system would be only as good as the data on which it is trained. For example, an algorithm used for shortlisting candidates for an interview will pick up on any historical bias and apply it going forward.
  • The stupidity argument: Some may not agree with the “intelligence” part in the term AI, given it requires tons of data before it can start producing useful results.
  • Black box problem: Deep neural networks can have 100s if not 1000s of layers. The greater the layers, the greater the levels of abstraction that it can pick on, leading to opacity of is operation. Because they are learning networks and not explicitly programmed, with their own internal representation of the problem they are trying to solve, how they make decisions is not known. Would you be willing to let an autonomous car drive you from Los Angeles to San Francisco, when you (or the experts) have very little clue into how the car makes the thousands of decisions it needs to make over the journey?
  • Extreme productivity = massive job losses: When a tool enhances productivity, we can do more with less human involvement. AI is unique in that it will be the first time we can (almost) do away with any human involvement, and as such it would be hugely scalable as well as impactful. The magnitude and the speed of job losses created by this fourth industrial revolution could be unprecedented.
  • Risk of general intelligence: Lastly, there is the threat of artificial general intelligence, or “self-aware” artificial intelligence. Most AI agents today work in narrow domains and have no agenda but to maximise the utility function. What if an AI programmed for increasing farm output starts to operate outside of the farm and starts taking over the planet to grow crops? What if it does everything it has to in order to achieve that goal? Elon Musk is worried that this is where the current efforts of AI could lead us to, so he is working on OpenAI, a non-profit research company that wants to make safe general intelligence.  Stephen Hawking has warned that the creation of powerful artificial intelligence will be “either the best, or the worst thing, ever to happen to humanity”.

The final obvious question then is what can we do to make sure that AI is the “best thing ever to happen to humanity” and not the worst?

  • Data is key: It goes without saying that the quality of data as well as the bias of the data needs to be actively managed in order to make sure we don’t unintentionally program biases.
  • Stupid but simple: I wouldn’t dismiss the neural networks of today that use a lot of data to learn as stupid. Of course their representation of knowledge is not optimal (in the form of strength of branches connecting an artificial neural network) and their understanding is slow (takes time to iterate), but for now, this approach works as we have no scarcity of data. However, the next generation of algorithms will surely improve efficiency regarding size of the data set. Many teams are working on these approaches including: Numenta, Geometric Intelligence, Klmera and Vicarious.
  • A matter of learning a new language: The current lack of insight into how an AI makes decisions is due to its different representation of knowledge. But could a matrix of weights be translatable to language as we understand it? I don’t know. Maybe we can train simple AI (where we understand the logic and can supervise learning) to start translating their rationality to us. If we extend this logic enough, could we teach the AI to translate its representation into how we represent knowledge?
  • An AI tax: Bill Gates recently made this seemingly controversial suggestion[vi], but is there merit in it? I think it’s a brilliant suggestion, at least for a transition period until the workforce can be retrained. The issue with productivity gains is that they accrue to capital owners. Taxes on AI could be the source of funds to retrain the displaced workforce.
  • Threat of general intelligence: I’m a bit relaxed about this as most AI we know today is some form of complex correlation exercise, applicable in very narrow domains. Is there a risk that all these narrow AI collaborate to form a super AI that would destroy humanity? I don’t know, but I do know that we are far away from such a point (or indeed doubt if we will ever get to that point). There are groups lobbying for a “master off” button of sorts into the AI if such a day were to arrive. I believe some sort of a standard body should emerge over time which could enable “good practices”, just as exist in coding!

I remain optimistic and look forward to the coming fourth industrial revolution!
In the long term, from an investment point of view, the resurgence of Artificial intelligence will lead to growing need for computing power, which benefits semiconductor and equipment makers. The productivity benefits resulting from use of AI would get split between the end user (consumers and businesses) and the software vendors.
Disclaimer: This is a discussion of broad technology trends and not investment advice. Any investment decisions made are your own and at your own risk.
All views, opinions, and statements are my own.
[i] Go has 250150 moves and Chess has 3580 moves. To put the scale of these numbers in perspective, the entire universe (as much as we know) has ~1080 atoms
[ii] In particular, a combination of Monte Carlo (game) tree search and deep (convolutional) neural networks for those interested
[iii] Document Search on Sentieo Plotter for Transcripts for “AI or Artificial Intelligence or machine learning or Deep learning”
[iv] There are some extending the AI progress towards Cybernetics – potentially interfacing a human mind with an artificial intelligent brain that one could use for certain kinds of compute / memory / search intensive tasks – but I am not ready to make that leap of faith yet, given our limited understanding of the human brain.
[v] “Google’s AI won the game Go by defying millennia of basic human instinct”, Quartz March 25 2016
[vi] “Robots that take people’s jobs should pay taxes, says Bill Gates”, The Telegraph February 20 2017