A simple decentralized voting system
I’ve been thinking about voting systems and how they could be improved. One aspect of our voting system that has always made me uncomfortable is the opacity of the process: once I submit a ballot, I’m putting my trust in an agency contracted by the government to correctly tally my vote.
Instead, I think there are four high level principles that could be followed to develop a more ideal system:
- Every citizen can vote and then see/confirm their individual vote.
- All citizens can see that someone has voted, but not see the contents of that vote.
- Votes are able to be easily tallied by everyone in the network.
- The voting system is distributed (i.e. not controlled by any single agency).
A decentralized blockchain solution could be revolutionary. And I’m certainly not the first person to think that.1 Even without the somewhat philosophical requirement of 4, a centralized implementation is straightforward for points 1 and 2, but much harder to imagine (for me at least) to satisfy 3.
The difficulty lies in the interplay between the three requirements and the connection between identity in the real world and on the voting network. The problem is more easily solved for a public key that is separate from a voter’s identity.2 This also fits with how our current voting system works (i.e. there is no connection between a voter’s identity and the ballot cast). Here’s an proof-of-concept that could quickly be built into an MVP:
Each voter generates a public/private key on the network. The creation of a public/private key is then safeguarded by people running the election booths.3 A vote is sending a digital asset from the voter’s public key to the public key of an official (ex: each presidential candidate has an address on the network, or, to satisfy the electoral college, an address for each state). Any node on the network can see that a vote was cast, and any independent node can verify tallies in real time.
The protocol would be designed such that the generation of any address comes with a certain amount of digital asset. (Think of it as generating a BTC address that automatically comes with 1 BTC). The easiest implementation would be for the amount of the asset you send to be trivial. This could ensure that every voter is given more than enough to participate in all their elections (as some people have more public representatives than others), as well as allow them to change their minds within a given time period (not sure this is necessary, but a nice addition). The network would simply tally the latest vote to a candidate in each race, regardless of the “amount” sent to that candidate’s address.
In this implementation, “public key” is a misnomer, since the identity of the owner of the public key is only known by the owner him/herself. So he/she can record their public key and then check that the vote went through properly at any point. (Or even resubmit his/her vote within an allotted period of time).
This comes full circle to a general argument for a better identity system. The quasi-private single SSN is useless in this scheme, and quite dangerous in many of its current applications. We could do so much more with a typical private/public key pair generated for every citizen.
http://www.bitcongress.org/ is the only real project I’ve seen working on this problem, but I haven’t seen any code yet. It is being built on Ethereum, which is yet to launch. Hopefully they can build something interesting! I also found that a small Danish political party has aspirations of using blockchain voting schemes. ↩
The other option is to link real world identity to the public address of a voter. In this scheme, every citizen has a public/private key pair. The difficulty here is engineering a solution to the traceability of transactions. Using a method such as Coinjoin, or similar methods employed by ZeroCoin and Darkcoin, transactions could be blocked together and “mixed” such that the network would a) know that a given public key made a vote but b) not know who that public key voted for. The difficulty here is that there is not yet any way to make it such that only the sender of the transaction can then check with 100% certainty that his/her transaction went through. ↩
This ensures at least the same level of effectiveness against voter ID fraud as is currently true. ↩
Follow The Vote
TL;DR: Visit followthevote.org and follow your Congressional representatives’ votes on Twitter.
About two years ago I became interested in building tools to improve our democracy. Specifically I started thinking about the poor granularity of information around our Congressional representatives’ voting behavior. There is lots of information and journalism about what Congress as a whole is doing (or mostly not doing), but it’s much harder to get information about how my specific representatives are voting.
For example: I really care about gun control. And when background check legislation was tabled in April 2013 (following the Sandy Hook tragedy, even!), I was quite upset. But I only read that the Senate as whole killed the bill. Had I known that N.H. Senator Kelly Ayotte, one of my representatives at the time, voted against the bill I would have been able to focus that anger. Whether that means influencing my next vote, writing/tweeting to her, or becoming more politically involved, I think the feedback loop “in the moment” can be much more powerful than the status quo of a quick flurry of debates and political advertising right before an election.
So I built FollowTheVote.org.1 It is a collection of Twitter accounts, one for each and every Congressional representative that tweets out their votes on important legislation. This makes it easy for any citizen to only follow the votes of their representatives. So I’ll personally follow the FTV accounts for Schumer, Gillibrand, and Maloney, while someone in St. Paul should follow the FTV accounts for Franken, Klobuchar, and McCollum.
The core idea here is that representative democracy is best served by a continuous, unfiltered dialogue between citizens and their representatives. Social networks such as Twitter are ideally constructed to facilitate this discussion on a daily basis. But there is no current infrastructure to get information at the representative level, and as of yet no societal pressure for the representatives themselves to blast out this information of their own accord.
There are many other ways this should happen. I find it a bit crazy, for example, that a NYTimes article on a bill doesn’t geolocate me to insert a sentence about how my representatives voted on the issue.2
FollowTheVote’s administrative backend grants programmatic access to all 500-something accounts, but I still want to curate which votes are sent out to followers (I’ve learned that Congress votes on a lot of either procedural or just plain useless things). So I’m looking for a technologically savvy member of the government who is aware of day-to-day legislation and interested in social media. The ideal candidate is a congressional aide or someone who works for a Senator or Representative.
Pricing Purely Wasteful Negative Externalities
My roommate leaves her A/C on 24/7. Despite being away for weeks at a time this summer. And while I sometimes forget to turn the lights off, or take needlessly long showers, this bugs me.
We don’t pay for A/C, all of our utilities are built into our rental price. My concern stems from what it does to the environment. And that concern is based in the fact that it’s purely wasteful.
There is a price to any amount of energy that we use. And I’m (mostly) fine with the capitalistic structure of society that says if you can pay for that energy you can use it.[^1]
But our prices are completely binary. Technology should step in to help make prices dependent on how wasteful something is. A/C should cost more if no one is in the room. A smart A/C could fairly easily detect movement/infrared heat to estimate if a human is in the room or not. What if we tripled that cost? Or 10x?!
For the latter half of the 20th century there was a struggle between the central planning of economies (great success!) and free markets to determine efficient pricing. My first reaction to my own anecdote was an uneasiness that such pricing structures would have to be centrally determined. Unfortunately, I don’t think we’ve come up with a better way for societies to price externalities (which is why cap and trade, carbon taxes, etc. are so difficult). What makes me feel better about this is the potential for governments to use factors rather than fixed prices as the tool for behavioral change. As in my example, I feel comfortable with a society determining that turning on the A/C when no one is home for 24 hours is 3x more expensive than normal. “Normal” is still set by the markets, the technology required to get energy, etc., all the things that make our markets efficient and foster hard work and technological progress.
I like to think of things that future societies will look back on and say, “That was ridiculous!” In this case, “How could something that hurts others be priced the same whether or not you’re even using it?” The answer could be, “Because binary pricing models were the only option before goods were aware of their surroundings and method of use.”
If anyone knows some good reading on this subject, I am completely ignorant. I’ve thought about this for 2 hours, total. It might fall under a different title, but I’m sure people at the intersection of economics and technology has written about this.
: There’s the separate question of whether the price of energy/waste is accurately taking into account the harm it causes to the environment, and thus to all of humanity. This is a huge question, but one that I’m going to ignore because much smarter people than I have written extensively about it.
Ethereum DAO code markup
One of the most exciting concepts in recent memory is a “decentralized autonomous organization.” I’m going to go through an early example of potential code for such a contract. For a primer on what is a decentralized autonomous organization (DAO, or DAC for “corporation” as the third word), I recommend reading the original white paper by David Johnston or Ethereum conception of a DAO.1
Ethereum, as the only Turing-complete protocol in the works, provides the best view of how this could work. I’ve taken code from their homepage (ethereum.org), which is bare and unexplained, and marked it up to make sense of it. It is written in ECLL (Ethereum C-Like Language) and offers an interesting view as to how they think a simple DAO might operate. There are two levels to the DAO code. The immutable code is the contract itself: this is what you see marked up below. The contract also contains mutable code, which is what the DAO does (stored in contract.storage). This code executes when interacted with and, here’s the key thing, can be changed by voting from the members of the DAO. Think of the immutable code as the structure of our government (how judicial, legislative, and executive branches work together) and the mutable code as our laws (what our government actually does, which changes over time as laws change).
- contract.storage is an array that holds memory for the DAO.
- Every member of the DAO is marked as stored in this array. Every entity on Ethereum has an address k, and are thus stored as a member of the DAO in the corresponding array entry contract.storage[k].
- contract.storage[2^255] holds the number of members of the DAO.
- tx is the transaction input that is sent as an argument to execute the DAO
- tx.data is an array that stores the two explicit inputs that determine how the DAO will execute.
- tx.data stores which of three transaction types to execute. 0 for ‘vote,’ 1 for ‘register new code’, and 2 for ‘finalize code change’
- tx.data is the index of the proposal
- tx.datan is number of lines of code to be added
More astute readers will note that having only a single array contract.storage to hold both membership as well as code and code changes will create collisions. Consider two members of addresses x and y. A proposed change of L lines of code has a hash offset of k. Collisions occur if x+k < y < x+k+L+1. Given the implied length of contract.storage (2^255 + 1), this may never be a realistic concern. But a DAO that say, included all citizens of a country, might have issues.
You know it’s cool when Wikipedia doesn’t have an entry for it yet. ↩
Agricultural Supply Chain
One of the most interesting markets I’ve looked at is the agricultural supply chain. This came out of our investment in Kitchensurfing this time last year, and subsequently trying to figure out if there were marketplace opportunities within the food supply chain itself.
The current state of the supply chain is best shown by a simple graphic:
At USV we look for markets in which the internet can act as a signalling network for parties that previously could not be efficiently connected at scale. This perfectly describes the journey most of our food makes from farm to end-user. At each stage, inventory is bought and sold, often after holding produce in market timing strategies. Middlemen shouldn’t be exercising this power, they should be fulfilling a logistics need. An online marketplace could disintermediate these players by allowing farmers and end-users to transact.[^1] While farmers only earn 19% margins from selling their crops wholesale, they can earn up to 90% from selling direct. (There are also, of course, potential non-economic benefits from eating local, organic food. Many of the companies I’ve talked to consider that core part of their mission.)
There are quite a few companies approaching these problems from different vectors. As an introduction to some of the more interesting players, I’ll slice by consumer vs. restaurant as the end-user. Keep in mind that many of these companies will (hopefully) quickly expand to serve both types of end-user.
Following that, I have a couple of investment theses which is (hopefully) the interesting part of this post. These reference some of the companies as examples, which is why the market map comes first.
At the end, I’ve written up a brief primer on food distributors, aggregators, and hubs, in case anyone else is interested in the current state of agricultural supply chain logistics.
Good Eggs is probably the market leader for farm-to-consumer. It is an online marketplace to order directly from farmers. Food can be delivered to your home for free, or picked up at a local drop-off point. The company is leveraging existing businesses to act as these drop-off points. The company is also the farthest along from a geographical perspective, serving 4 metropolitan areas.
Wholeshare is group buying from local farms. Anyone can appoint themselves a group coordinator or join a nearby group. Once that group has collectively ordered a minimum shipment, food is delivered to the coordinator. This group purchasing power creates a bigger incentive for sellers, which range from individual farms and fisheries to regional food distributors. The “last mile” problem falls to the group coordinator, who is compensated with 5% of the transaction. The group coordinator may deliver food to his/her neighbors or have them come to their home/business to pick them up.
Fresh Nation is an early stage team in Connecticut building off of existing farmers markets. Farmers list what they bring to those markets online and buyers can pre-order. This is great for farmers, who are typically only able to guess how much of each item to bring each day. They’ve also taken a page from Instacart and TaskRabbit, crowdsourcing a group of personal shoppers who will pick up your order at the market and deliver it to your door.
There is also, of course, the recurring box strategy. Blue Apron has had remarkable success shipping weekly dinner boxes, each containing recipes and the exact amount of ingredients. Quinciple will deliver a box of fresh, locally grown produce each week. The box features a diverse selection, but subscribers cannot choose any of the ingredients. I consider this a nice edge case, but to make a meaningful impact on our food networks we need to have selection.
Turning to companies starting with restaurants as the end-user, Provender enables farmers to sell directly to restaurants and just use food hubs as the delivery infrastructure (rather than a marketplace participant). They have created an online marketplace for farmers and restaurants.
Foodem is a marketplace for wholesale food distribution. Based in Maryland, they’re trying to help smaller distributors get market vision and eliminate the need for them to have a sales team. Farmplicity, in St. Louis, has a similar model.
It’s also worth mentioning that the original entrant in the space, Local Dirt, was founded in 2005 and started by focusing on B2B. The company raised venture capital from OATV as well.
Lastly, Plovgh is taking the most head-on logistical approach. The have built an online platform for farmers, logistics providers, retailers, and individuals to coordinate trades without intermediaries. This disintermediates brokers and merchants from the process. Unlike any of the other companies mentioned above, Plovgh isn’t trying to define relationships that should exist in the supply chain: it is enabling all actors to buy and sell from one another, setting up lightweight commercial exchanges for buying and selling. As the market “figures itself out” Plovgh can react to the submarkets that show promise.
Selling to end-consumers scares me. We are very finnicky, particularly around the last mile problem (see below). In tech-historical terms, it’s interesting to think about online grocery shopping vs. online apparel shopping.
At the dawn of online retail, i.e. the beginning of time, i.e. ~15 years ago, there were many arguments against buying your spring wardrobe from Gap.com: 1) How do we know if the clothes will fit? If this color will look good on me? 2) How will retailers deal with customer returns? What will the rates of return look like? Will this be too high for consumers to deal with / too expensive for retailers? 3) Lots of people enjoy shopping for clothes in person, and 4) Questions/concerns about credit card payments, online fraud, interstate taxation, etc.
Fast forward a few years, and more than half of all Americans have shopped for clothes online in the past year.1 This is in addition to the higher gross margins and increasing reliance of the industry on apparel e-retail.2
A priori, the arguments against buying your clothes online are stronger than the arguments against buying your groceries online. Much of the food you buy is a commodity, and if it’s not a commodity, you have a 100% expectation of what the product will be like (all Honey Bunches of Oats boxes are fairly identical). Food, especially staples, can be a recurring purchase. There is little potential for having to return delivered groceries. And, purely anecdotally, I think more people enjoy shopping for clothing than they do going to Stop & Shop every week.
But buying food online is a much, much smaller part of our society than buying clothing online. It is currently less than 1% of the food retail market.3 And there have been major endeavors, from early ventures like Peapod to recent major entrants like Amazon Fresh.4 I would argue that the societal impact of companies like Peapod and Fresh Direct (not necessarily their investment returns) have been a surprising failure.5
The reason for the lack of market penetration mostly boils down to consumer behavior. The business model has been tried enough that, except for the key lack of market desire, it can work from a logistical and economic perspective. And it’s an open topic for discussion as to why consumers haven’t been more excited by getting food delivered to their home. But as an investor, the takeaway is a natural hesitance towards investing in any consumer-focused business that has not already shown significant user traction. Put another way, I think seed investors should be extra discerning here.
An interesting case study here is Instacart. Of anyone so far, consumers have responded the most favorably to their product experience. (I really haven’t mentioned Instacart in this post because they are not fundamentally disruptive to the food supply chain. They add on an extra step, rather than taking out any of the existing steps).
Contrast this to restaurants, which as a business will be much more willing to jump through hoops for either cost savings or “localness”. While it’s niche, the highest end restaurants actually employ people full time as “food sourcers.” Their entire job is to talk to farmers, using spreadsheets to track their procurement of the best local ingredients. There are sections of the pyramid immediately below which can’t go to this trouble/cost, but may be interested in an online platform that enables this kind of sourcing at a much lower economic and logistical cost.
Another way to explain it is that companies tackling both the supply chain and selling directly to consumers are tackling two difficult problems at once. I am much more comfortable with a company focusing on one of these two areas. Success in either of the two can be a massive return, I happen to be more interested in the former problem as an opportunity with a wealth of nascent technology-driven approaches. I am bullish on companies developing disruptive supply chain models by initially catering to business end-users before entering the consumer market.
Logistics and the Last Mile Problem
Ultimately this is a massive logistical problem. Some companies are taking on the entire supply chain, others just part of it. It is particularly tricky to figure out the last mile problem, i.e. how to deliver to end-users. Good Eggs has a few options. Whole Share is betting that each community will have a single “coordinator” who is willing, whether for free or for a fee, to act as a mini-aggregator between the farmers and the community. Some are outsourcing to existing delivery networks. Others are seeing if the farmers can just take all the logistics into their own hands.
How far down the chain?
A big question is how far down the chain a startup should go. Should they be interfacing with farmers directly? Or is that too big leap, and should instead be innovating on only part of the supply chain.
Most of the companies on this list, including Good Eggs and Fresh Nation, are connecting directly to the farmers. But Wholeshare, for example, purchases from wholesale distribution networks. These distributors normally sell to restaurants and grocers, the latter of which adds a 25-55% markup. So clearly there is still room to innovate and pass savings on to the consumer without dealing with food producers themselves.
Can there be a winner at scale?
A big question with any geographically constrained marketplace is if there can be a major venture capital return. The biggest internet businesses are essentially re-aggregators (Google for finding pages, Amazon for finding products) made possible by the internet’s ubiquitous reach, which is naturally at odds with real-world geographical constraint.
Most of the companies mentioned in this post are starting off locally, so it’s hard to see much competition between them yet. Provender is starting in Montreal, FoodEm in Maryland, Good Eggs is in SF, LA, NY and NOLA, etc. Unfortunately, there is a real concern that this sector will always be fragmented, and it is way, way too earlier to tell.
Examples of successes in other geographically constrained verticals have lock-in that food may not have. While you take most of your Ubers in your hometown, it’s has homescreen lock-in so that you use it whenever you travel somewhere else. AirBnB needs local supply to be relevant, but by definition people use it when they are traveling.
Brief primer on wholesale food distribution
The food distribution market is highly fragmented. There are about 50 big national distributors. Each one works via regional branches with separate P&Ls. The regional market is even more fragmented. For example, Maryland alone is served by 1,600 regional distributors.
Food aggregators/hubs are the physical locations out of which food distributors store produce (i.e. a staging ground). The distributor moves food from the farmer to be stored at an aggregator before being shipped to the end-customer.
The aggregator/hub is it’s own corporate entity, renting out space and materials to distributors. This is not a fast moving market; the aggregators I looked at have rented out the same parts of their warehouse to a particular distributor for decades.
For example, the Hunt Point Terminal Market in the Bronx, the largest aggregator in the world, serves the New York metropolitan area. It has acres of food storage, serviceable by rail and trucks. It houses $2.5B of fresh produce in climate controlled environments. 62 market merchants operate out of Hunt point. For example, A&J produce sells wholesale food in three categories: vegetables from the Americas; mushrooms from upstate NY, Pennsylvania, and Canada; and fruit from all around the world. The is the scale of industrial agriculture.
Food hubs are smaller versions of aggregators. They tend to serve local farms. Some are non-profits backed by government initiative.
This stat is from 2012, so it’s certainly higher this year. https://www.internetretailer.com/2012/01/24/books-clothes-and-electronics-are-most-shopped-categories ↩
A secondary (but important point) is that most of these models have only innovated on the “last mile” problem. Meaning that they are really just taking groceries from a grocery store/warehouse to the consumer’s house. I find companies that are trying to shake up the actual supply chain itself, not just adding a final step, to be much more interesting as well as potentially disruptive. But these companies are still illustrative of shaky consumer behavior. ↩
That all these companies are privately owned is frustrating, as I can’t give much supporting evidence (but also indicative of lack of success). ↩
A couple months ago Brittany, our portfolio manager, asked me to help write a simple app to automate two-way introductions. She spends a lot of her time connecting people within the portfolio, so it seemed a worthwhile hack. I’ve open sourced the code here if anyone else has similar needs.
Create a personalized message to Person 1 asking if they would like to meet Person 2. In that message is a link. If Person 1 clicks on the link, it will automatically send the actual intro email between Person 1 and Person 2 (so you don’t have to worry about it). The app also includes a database for keeping track of when people open the emails and make the intros. So you can always see if someone says no to/forgets about the intro.
Here’s a screenshot of the app. It requires no coding to get this up and running as a stand-alone Tornado app, just a couple of configuration variables. I’ve included instructions to get it up on Heroku. Will need a MongoDB instance if you’d like to use the database functionality (i.e. track your intros).
Giving patients back their health records
One of the subareas of healthcare we’re interested in are companies that give patients easy access to their healthcare records. I’m tracking 4 companies that fall in this category. They are all mobile-first, very early stage, and so far have minimal product differentiation. I actually really like having this in my life, so figured some of you might too: HelloDoctor (launched), Prime (launched October 2013), Picnic (private beta, just got into YC), and Gliimpse (pre-launch).
Why now? ACA provides now incentives for (and will soon penalize doctors for not using) EMR. One of the conditions for these EMRs is that they be accessible by patients. While few, if any, EMR providers are building APIs, there are a number of companies taking a Mint-like approach (get username and password, log in and scrape XML data) to present and aggregate patients’ medical records in a clear and portable form factor.
A real concern here is that all of these companies are just front-ending patient data. Sadly, the fragmentation of EMR might actually mean there will be some value here. Unless the industry adopts a standard protocol or coalesces into a few players (at least geographically), most patients will have their medical data stored in different silos. Aggregating and normalizing this data so that it can be used all together will be valuable (but not necessarily defensible) and should go hand-in-hand with enabling patient access
Therapy at Web Scale, and other innovation in mental health
I’ve spent quite a bit of time looking at the mental health space. Part of this interest comes from an analysis of the health sector. We look for markets with low barriers to entry, not an easy thing to find in the world of health insurance, EMR, and hospital sales cycles. Mental health is one of the most most attractive subsectors in the health space for internet-focused investors:
One quarter of all American who receive mental health care list themselves as the primary payer. There is already a large customer pool operating outside of health insurance, suggesting that lightweight startups have a better chance of capturing market share.
Similarly, 45% of untreated Americans cited high costs as the primary barrier to care. New models that innovate by lowering price points may be able to tap )nto this unserved market.
Proportionally rising allocation towards outpatient care. Outpatient care rose from 24% of mental health expenditure in 1986 to 33% in 2005. While we don’t see exciting investment opportunities in prescription medications or inpatient care, the most interesting mental health startups are innovating in ways that will revolutionize the outpatient experience.
The Affordable Care Act clearly creates more mental health mandates, requiring all insurers to include mental health benefits. Plus, the act is being implemented right after most states slashed mental health budgets in response to the recent recession.
There are three categories of internet-based mental health startups:
Therapy at web scale.
My personal favorite, this is using technology to bring therapy itself online. While long distance therapy has been existence for quite a while (for instance, therapists using Skype for video sessions), we’re seeing a number of internet-based platforms to facilitate this behavior.
I think there is a massive gap between seeing a therapist and “nothing.” It takes both a huge financial and mental commitment to decide to see a professional therapist on a regular basis. Net-native therapy platforms have the potential to help the millions of people caught in the middle, people who probably should see a therapist but can’t bring themselves to make that leap.
TalkSession is a next-gen teletherapy platform. They are starting small and highly curated (there are 26 psychologists/psychiatrists listed so far), but aspire to grow into make teletherapy both universally accessible and affordable. While TalkSession looks like the most polished entrant, there is a tail of similar models. BreakThrough and Blah Therapy are quite similar. Stillpoint Spaces specializes in video therapy sessions with counselors trained in Depth psychology. Online Therapy is an older competitor.
The above platforms have taken teletherapy quite literally, upgrading from phone calls to vidchat over the internet. One of the most interesting things is that there is a huge market for text-based therapy. TalkTala offers both, but has found that most users prefer the asynchronicity and anonymity of texting. TalkTala also has a hybrid model with combines free, public posting (and response from therapists) with paid, private discussions with a therapist.
7 Cups of Tea is a p2p platform to connect with “listeners.” Founded by psychologists, they give listeners brief training on how to be a passive human presence and not an active therapist.While a true marketplace, the majority of listeners elect to offer their services for free. Blah Therapy, mentioned in category 1, also free non-therapist listeners. It’s an interesting idea that this kind of human interaction may be sufficient for non-serious mental illnesses.
Personality Cafe is an example of an older social network focused on mental health. Started as a social forum dedicated to the Myers Briggs type indicator, it’s a place for many people struggling to mental illnesses to connect with and support one another. In this vein, there are two new exciting apps, Secret and Whisper. While neither is directly related to therapy, I think both are relevant. The internet, in large part due to the anonymity it affords, has always been a place to vent and an outlet for personal struggle. I see both apps as progress towards better understanding what form factors can best facilitate these kinds of emotions and desire for human connection.
Find a therapist, i.e. ZocDoc for therapy.
Perhaps unsurprisingly there are few teams working to build an online marketplace for offline therapy. While I think offline therapy should and will continue to be the predominant form of mental health care, the entrepreneurial/investment opportunity is less exciting. By nature, seeing a therapist should be a long-term transaction. Unlike with ZocDoc, for which you may need to find many different types of doctors in a year, you’re probably sticking with a single therapist for quite a while. From a purely economic standpoint, it is much harder for an online marketplace to capture in an initial connection before the users (on both sides) go off-platform. That’s why there are many more teams working on teletherapy. The best directory by far is still Psychology Today’s Find a Therapist.
Many of these companies allow for mental health treatment while remaining anonymous. Of any medical profession, my instinct is that identity and in-person, human interaction is most important for therapy. Yet the market is tending towards the exact opposite: asynchronous messaging. My hope is that this is a good entry for people unfamiliar with or nervous about teletherapy or offline therapy, but that those who need it graduate to more in-depth forms of mental health care. If it becomes an important market, we’ll see more therapists taking advantage of the asynchronicity to use this as a supplemental form of income.
Another question is how far much the market will flatten. It’s possible that the willingness of altruistic non-therapists to act in some mental health care capacity, such as the free listeners on 7 Cups of Tea, will further compress the costs of engaging with a licensed medical professional, whether on- or offline. Is there a significant difference between texting a therapist vs. a “listener?” If there is, which will potential users unfamiliar with mental health care choose? Will there ever be regulation around this issue.
And, finally, an idea I like: technology here can actually lower the switching costs between mental health professionals. Currently, if you decide to leave your therapist and try someone else you must start from scratch. Despite telling your life story to someone, this new therapist knows absolutely nothing about you; it’s like going through a break-up and then starting over with someone else. But most of these online texting platforms save all of your conversations, which can be transferred over to a new therapist if so desired. Rather than go through everything again, a new therapist can read through a new patient’s history be quickly brought up to speed.
There is an overwhelming amount of innovation occurring in the cryptocurrency space. I mean that literally - it’s quite difficult to parse and understand each of the “alternative” projects being built on top of and/or around the Bitcoin blockchain.
To that end, I present a primer on what I believe are the 5 most interesting endeavors:
- Untapped capabilities BTC protocol itself, including smart property, scripting and proposed payments request mechanism
An understanding of these projects allows for a more coherent view on some major questions the cryptocurrency community is addressing. What other problems require a mechanism for distributed consensus? How should identity work in the cryptocurrency world? Should scripting be Turing-complete, or is that overkill? How flawed is BTC’s proof of work, and what other options are there? What does a future of distributed autonomous organizations look like?
I’ll offer my opinions in future posts, but the purpose of this research is to present an unbiased introduction to the technologies and some arguments from people far smarter than I am. To the latter point, the bottom of the document lists some of the best blog posts around these ideas.
In defense of my selection of what to highlight: I am interested in projects that, to an end-user, most differentiate themselves from our current use of Bitcoin as a “simple” currency. So, for example, while LiteCoin and PeerCoin have meaningful technical differences from Bitcoin (mining hardware and proof of stake, respectively), what they enable the end-user to do is not nearly as exciting as the AppCoins of Mastercoin or full Turing-complete scripting capabilities of Ethereum. Similar logic applies in many other cases, but I am happy to persuaded otherwise.
Again, the link is here.
Examining “Software Eating the World”
I recently read a blog post by Ethan Kaplan which argues that
…the world itself is now possible to represent by polymorphic functions with refactorable logic.
This is what we mean by “software eating the world.” Out of personal experience, technologists can often get excited by the application of logic to all of life’s decisions. So, I’d like to consider the two parameters along which this is occurring:
- We are identifying new, previously un-parameterized areas of life to replace with software.
- We are improving on preexisting functions.
The first is intuitive. 10 years ago, only a human could recommend a song to you. Then Pandora began writing algorithms to do it. In the past decade or so we have begun solving this for finding restaurants, movies, taxis, and even love. There are probably still many untouched fields to explore here (if I could think of them I’d build it!). As an investor, this happens to be the most exciting kind of opportunity to come across.
The second parameter begs an interesting question: What are the upper bounds to quantifying every variable? You can often see them directly in the product experience. Searching for a car online (to continue Kaplan’s central example) takes a list of inputs: make, year, price, horsepower, etc. But what if you get inside the car and the driver’s seat doesn’t go back far enough? Or the steering is surprisingly loose for your taste? Or just that, in person, the car doesn’t look quite as sexy as you’d hoped?
I think some of these variables may be inherently unquantifiable. Others disagree. But either way, some of the functions we use may already be approaching the limit^ of an inherent tradeoff given the current computing form factors we use: UX vs. time. Even if it were feasible for Cars.com to ask me all those questions, it’s much more human to just drive the car for 5 min. So, until the next technological breakthrough, how much more effort will we put into improving these functions?
At this point in history “software is eating the world” for only those parts of life that are most easily quantifiable.
:Commodities are particularly interesting, as they remove all human preference and are thus best served by a function.