Fermi.VC

Union Square Ventures @AlexanderPease GitHub/AlexanderPease alexanderpease.com

[fermivc]

Adding unique field to preexisting MongoEngine collection

I am building a Tornado/MongoEngine app. I wanted to add a field to a large collection of preexisting documents, set that field to a unique string for each document, and enforce the field as 'unique' going forward.

This is surprisingly annoying to do because once you set StringField(unique=True), MongoEngine assumes that only one of your documents can have a null value for that field.

Best workaround:

Add field to document class. Do not set 'unique=True'

class YourClass(Document): newField = StringField()

Add the following code to the class. This sets the field to be both unique and sparse. Sparse means that you can have multiple null values. This is important because we haven't yet run the function that will populate all of the documents with this new field.

meta = { 'indexes': [ {'fields': ['newField'], 'sparse': True, 'unique': True}, ] }

Note that setting 'unique': True does the same thing as if you were to put

newField = StringField(unique=True)

except MongoEngine requires sparse to be handled by the meta dictionary, and if an index is set for sparse this way you must also use the meta index to set the unique part of the index.

Run the function that populates your newField for all documents. A good way to check if you've succeed in making each value unique:

len(YourClass.objects) == len(YourClass.objects.distinct("newField"))

This will return True if each document in has a distinct value for newField. If not, then debug your function that populates newField for all documents.

Once you've succeeded, you can either do nothing (which keeps the sparse index) or delete the meta dictionary from step 2 and set

class YourClass(Document): newField = StringField(unique=True)

With this, all new documents created will require newField as the init method is called. (This can either be a good thing or a headache).

[fermivc]

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. ↩︎

[fermivc]

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

Help!

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.

I built this super, super sporadically and never really got around to launching it. Oops. ↩︎

Fine tuning and adding information at the content level to specific readers is a super interesting and much more broad question. ↩︎

[fermivc]

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.

[1]: 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.

[fermivc]

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).

Variable definitions

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[0] stores which of three transaction types to execute. 0 for ‘vote,’ 1 for ‘register new code’, and 2 for ‘finalize code change’

tx.data[1] is the index of the proposal

tx.datan is number of lines of code to be added

Code markup

Collisions

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

You know it’s cool when Wikipedia doesn’t have an entry for it yet. ↩︎

[fermivc]

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.

Market map

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.

Finnicky consumers

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 ↩︎

http://blogs-images.forbes.com/matthewcarroll/files/2012/10/main-qimg-b8b8f8a8e88b88888ba88ea989daaeb88 ↩︎

http://blogs.marketwatch.com/behindthestorefront/2013/07/16/food-retailers-start-to-take-online-groceries-seriously/ ↩︎

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). ↩︎

[fermivc]

IntroBot

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).

[fermivc]

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

[fermivc]

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.

Market map

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.

P2P/social networks.

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.

Interesting themes

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.

[fermivc]

Alternative Blockchains

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

NameCoin

Mastercoin

Ethereum

Ripple

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.

[fermivc]

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^[1] 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.

[1]:Commodities are particularly interesting, as they remove all human preference and are thus best served by a function.

[fermivc]

Disaggregation of a Bank

There are lots of cool charts thrown around the startup/VC world. One of the best is the "Disaggregation of Craiglist," originally created by Andrew Parker in 2010 and updated by David Haber in 2012 (shown below):

USV has recently invested in a host of internet-enabled financial services, which prompted the question: what would a "Disaggregation of a Bank" look like?

I've started putting together a list of startups providing services that would traditionally be offered by a bank (broadly defined). Please contribute if you have any startups or categories that I'm missing!

I took a first stab at visualizing this in the following deck. Past the Craigslist slide, I tried to diagram out the functions of a bank. Most everyone at USV found this a bit confusing, but, on the next slide, it allowed me to show where companies are innovating. I recommend going full screen to take a look (or feel free to download the PDF):

View Disaggregation of a Bank v1 and other presentations by alexander33be.

Any thoughts? Ways to improve? My hope is that this leads to larger questions: Which banking functions benefit from economy of scale and remain defensible vs. which areas are ready for disruption? Should startups be focused on consumer, business or both at once?

[fermivc]

Best Google Search Ever

This is an anecdote that got me thinking about how the UI/UX of search will continue to evolve.

I recently was in London for Techstars and ended up eating at a small cafe. I had an excellent eggplant/couscous dish (I'm clearly not a chef). A couple weeks later, a few friends were coming over for dinner and I wanted to recreate that dish.

I couldn't remember the name of the restaurant, anything resembling a street address, or even what the dish would be called.

But I still had a mental image of how I got to the restaurant. So my search was: - Start Google Street View at the St. James Theatre, where Techstars Demo Day was held. - Walk myself through the streets of London using Google Street View towards the restaurant. It's kind of amazing to explore a city this way. - Found the restaurant! (I was actually disappointed that I couldn't Street View my way inside). Now I knew the name to search for. - Foursquare brought it up, and I found a link to their menu. Wasn't the exact recipe, but I was able to make a passable replica knowing the major ingredients.

The point is, we still use search in mostly two capacities: type in a phrase or look at a bird's eye view map. Yes, Google Street View has been around for a while, but I'd never used it in this way before. But it was perfect for my brain: I don't remember names, but I have an excellent visual memory. Everyones memory works differently, and I think search will grow to be able to handle this. Someone else might have remembered the decor in the restaurant, or the color of the sign outside. Or the music that was playing. Or the fact that the bathroom was downstairs. I think all these little things, the oddball remnants in our memory, will help us complete searches like this in the future.

[fermivc]

Thoughts on Quantified Self

USV has been interested in the quantified self space for over a year now. As investors in "networks enabled by the internet," we've seen the majority of the innovation so far as being outside our wheelhouse: hardware products and apps in specific verticals. And so while we have not made any investments yet, we've met with many entrepreneurs and had enough internal discussion to share some of our thoughts.

QS 101

Quantified self = personal data collection Wearable = device that captures data (except your smartphone) Quantified applications = apps that either collect (with or without a wearable) quantified self data from a user and/or sit on top of the quantified self data from other sources

The form factor of the mobile phone will determine the wearables market

The past two years has seen an explosion in fitness-based quantified self devices such as the Fitbit, UP and Nike Fuelband. This mass adoption is even more remarkable given that these devices all track the same few metrics around activity: steps taken, calories burned, hours and quality of sleep.1

Yet this data can be captured by a mobile phone utilizing the same array of sensors2 found in a wearable device. Moves and Runkeeper are currently leading the pack in app-only products. Battery life and developers’ utilization of that battery life will continue to improve. And Apple’s recent announcement of the M7 chip, a processor dedicated towards making quantified self data available to app developers, shows that mobile phones are making the current class of wearables redundant. Current mainstream devices will have to innovate to remain relevant.

The quantified self device industry must shift towards data that is impossible for a mobile phone to capture. At the bottom of this post is a partial breakdown of the quantified self data that is and will be captured by devices and apps. It’s unlikely that mobile phones will be the form factor to, say, capture brain waves while you sleep, so there is opportunity for wearable products in that specific vertical.

Data tracking will become more medically relevant

While the current mainstream devices are focused on fitness, next-generation quantified self devices will enable medically relevant data. The list of next generation wearables (at the bottom of this post) shows potential for data in bloodwork. And these companies are mostly in the Kickstarter/IndieGoGo phase of development, potentially only a year or two away from being available. Similar advances in mainstream, big-brand wearables won't be far behind.

Hopefully, this data can be shared with doctors to improve preventative health.3 It’s possible that this will be a lightweight interaction, such as showing your doctor your accumulated physical activity as part of an annual physical. Or, it could become an integral part of the medical ecosystem, our vitals constantly being monitored and our doctors alerted automatically when necessary.

APIs and data aggregation

The biggest winner in this space will be the API-driven platform that creates a stable ecosystem to aggregate and interface between data sources. Developers will have access to larger pools of data. There may also be powerful social networks built on top of this data. Barriers to entry will be particularly high if this platform player figures out how to be HIPAA compliant and become the point of contact between patients and medical professionals, a particularly difficult barrier for fledgling teams. It's hard to overestimate the value of medically relevant data coming from user engagement in a network lightweight consumer apps.

Signals in the Quantified Self Industry

The original list is here. It includes links to companies and will evolve over time. Feel free to contribute!

Mainstream (i.e. in Fitbit, Fuelband, UP, Moves, Runkeeper, etc.)

Calories/Distance: tracking how many steps you take and how active you are

Sleep: tracks amount and quality of sleep, some products also time alarms to your sleep cycle

“Active” tracking: requires you to manually input data. A suite of apps have been built for fitness regimes and dietary tracking.

Available, but not mainstream

Heart rate: measuring your pulse

Withings Activity Tracker is a little clip-on that fulfills that mainstream categories but also has a built in optical pulse tracker. Requires the user to hold their finger up to the device. They also make a blood pressure monitor.

Azumio has an app for measuring pulse using a phones native camera and flash.

Posture

LUMOBack monitors lower back posture and pelvic alignment. They’ve built a waistband with inertial sensors.

Breathing

Asthmapolis has built a bluetooth sensor to monitor the environment, alerting the user of conditions that could trigger an asthma attack

Spire has a yet-to-launch clip-on device to monitor breathing, showing people their respiratory patterns and helping monitor stress and emotions.

Tinke tracks and helps with breathing exercises.

Advanced Fitness

Nike Hyperdunk is the next generation of in-shoe sensors. Designed specifically to track and analyze your basketball game.

Active Mind Technology is soon to launch golf specific sensors and analysis after a successful fundraise on AngelList.

Fertility

Ovuline uses machine learning to help couples conceive faster. Works with some of the mainstream wearables technology.

Genetics

23andMe is the leader in affordable DNA sequencing and analysis.

Microbiome: the ecological community of microogranisms that live inside of you

uBiome provides a $89 kit to “quantify your gut” and compare it to other user data sets. Dozens of health conditions are correlated with the microbiome -- from asthma to diabetes, autism to depression, irritable bowel, Crohn’s, chronic sinusitis, heart disease, and more.

Next generation

Blood

Sano Intelligence is building the “API for the bloodstream,” made possible by their proprietary transdermal blood chemistry sensor patch

Child/infant

Sproutling is creating a baby monitor for the quantified self age. Will track temperature, heart beat and sleeping position from an ankle band that wirelessly connects to the parents mobile phone.

Brain/Focus

Melon is a headband to measure and improve focus. Completed a Kickstarter fundraise for their initial product.

Muse is another brain sensing headband, available for preorder now. It picks up brainwaves in real-time and sends them to your phone, tracking this data and even turning them into commands!

Ballistocardiography: measures mechanical forces of the heart, gives data on heartbeat (cardiac contractions), respiration (chest wall movements) and movement

Beddit is building a sensor that lies on your bed and picks up on these forces while you sleep. Available for preorder.

Future

In vivo sensors

Nutritional analysis (without having to manually input everything you eat)

Mainstream blood glucose monitoring (as is currently the norm for diabetics)

Integrated into clothing

And much, much more...

That there is so much consumer interest in a product with so little power is indicative of how exciting the space will become. ↩︎

An accelerometer, gyroscope and compass are industry standard in smart phones. ↩︎

My personal realization came when my grandfather was recently admitted to the hospital for the second stroke in two weeks. Doctors were adjusting his blood pressure medication, causing it to fluctuate. But the only data points they get are when it’s too late, once he’s already had the stroke. It’s insane to me that our current medical system is not using a wearable device to monitor his blood pressure in real-time, monitoring the effects of medication outside of the hospital as well. A simple algorithm could easily have provided early warning signs as his condition worsened. ↩︎

[fermivc]

Ridiculously Fresh

The best ingredients are fresh. So fresh that they’ve just been picked. How do we get to a world in which most of our food can taste this good? There are two obvious possibilities:

Grow food everywhere. While there are plenty of small scale efforts to grow crops in urban areas, this is inherently unscalable.

Speed up the supply chain. Most crops are held by wholesale buyers for long periods of time at warehouses, sacrificing quality for market opportunity as prices fluctuate. Even with vast improvements in the system, it’s not physically possible to transport produce from farm to table on the time scale I’m envisioning.

The answer is a combination of the two, the product of thinking so far out of the box that you end up in one: grow food as it moves through the supply chain.

Freight Farms has figured out how to grow crops inside a shipping container in any external environment. Each 40 foot by 8 foot container is packed with produce growing under energy-efficient LEDs regardless of whether the container is held in a lot, is traveling on a flatbed, or has arrived at its ultimate destination. And while their focus is not on consumer yet, it’s possible to imagine a futuristic Whole Foods made up of a series of these containers. Need carrots? Walk into the appropriate container and pick a few out of the hydroponically nourished soil. Suburban homes could have a container in their backyard. Freight Farms could even periodically airdrop containers onto the roofs of urban apartment buildings.1

I’ve spent the past month doing a deep dive into the agricultural supply chain. There are a couple of talented teams creating online marketplaces to disrupt the middleman-heavy status quo. And while Freight Farms is outside of this model, and not a USV company, it has been one of the most fun to think through the implications of the world that they’re envisioning.

The future is full of awesome helicopters/drones making stuff like this a reality. And most urban rooftops are totally wasted space anyway. ↩︎

[fermivc]

Connecting on LinkedIn is pointless gamification

LinkedIn is an network of interconnected professional personas. I’ve already argued that the current connection model makes for a useless professional address book. This post is about why connections themselves are an arbitrary hindrance on the value (to consumers!) of the LinkedIn graph.

Part of the idea behind being on LinkedIn is the possibility that someone interesting will see my profile and reach out to me. It might be someone I met briefly at a conference, someone who just Googled my name, or, hopefully, a recruiter with my potential dream job that I didn’t even know existed.

And my profile is exactly “who” I want them to find: my resume, education, skills, accomplishment and a picture of me wearing a tie.1 This is my professional persona. So the question is: who do I NOT want to be able to access this information?

The answer is no one. The expectation value of benefit to me increases the more open to the world that profile is.2 Yet for anyone to see my profile, they have to be a “second-degree” connection to me. For me, and most other people I’ve talked to, that interaction only occurs with people I’ve met in real life.

On the flip side, there are plenty of people’s resumes I’d be interested in seeing, but who I either am not personally acquainted with or don’t want to bother with a pointless LinkedIn connection. For example, I’d love to see Reid Hoffman’s career track, to understand how he got to where he is today, without bugging him about it.

Why, then, the connection model? It’s a classic case of a network constraining information to earn money. There are no connections on the backend: so long as you pay LinkedIn enough money you get free rein to troll your way through the entire platform. This is only feasible as a business expense, of course, so just recruiters and companies get to see the whole lay-of-the-talent-land.

Constraining information that does not want to be constrained is an inherently unsustainable business model for an internet company.3 The ease of creating a LinkedIn profile used to be a major technical advantage. But it’s increasingly easy to put up your resume online. There are alternative services specializing in specific verticals. Most members of the tech community just throw up their own website. And Google works just as well, possibly even better, than LinkedIn’s internal search.

As an aside, a byproduct of the connection model is a brilliant reinforcement of LinkedIn’s network effects. Connecting keeps us coming back to LinkedIn, remind us of it’s value and reinforcing the idea that LinkedIn is “the” platform for our professional personas. All while providing us with only a tiny fraction of its value of the platform. So the next time you make a connection on LinkedIn, remember why you even had to do it in the first place.

This is a generic statement. I’m not wearing a tie in my profile. Here’s proof. ↩︎

So do the number of useless connections, but those are already pretty easy to ignore. ↩︎

Contrast this to Facebook, in which we all definitely do not want those pictures to be viewed by our future employers. ↩︎

[fermivc]

Citi Bike's Failure of Information

Citi Bike in NYC is great, when it works. Plenty has been written about the programs engineering struggles. But even more astounding, and ultimately more detrimental to the program, is the shoddy quality of information in the program.

Above is a picture from my phone of the Citi Bike app on Saturday night. I had ridden a Citi Bike back to the East Village from Midtown. All of the Citi Bike docks shown are supposed to have many free spots, represented by being only "partially full" in dark blue. All of the ones I've circled in red were, in fact, completely full. It ended up taking me longer to ride around searching for a spot in vain than it would have taken for me to have just walked.

Besides being extremely frustrating for a consumer, it's much more dangerous to Citi Bike itself. For example, the program is spending significant money and effort trying to dynamically balance the allocation of bikes. How can they do this when all their data is plain wrong?