Startups and Shit @startupljackson - Tumblr Blog

Request For Hacks

Today Product Hunt launched a hackathon, which I will be a mother fuckin’ judge of. As such, I’m going to suggest a few projects I am highly confident would be judged as awesome.

DeEchoChamber For Twitter - Diversity Index (Crowdsourcing/Mechanical Turk)

Twitter has an interesting social graph and I’ve personally benefited from seeking out a diverse set of perspectives. I think about proactively following more women and minorities in technology and investing, but also try to find more academics, great technology journalists, intellectual conservatives/liberal economists, etc.

I would love to see someone build an app that gives individuals a score, but also that maps out orgs (e.g. executives at company, investors at a venture firm, etc) and scores the diversity of perspectives experienced at the tops of organizations. As they say, you can’t improve what you don’t measure.

This app could then be used to recommend interesting people to someone who might have a diverse perspective that’s relevant to them. So, if you are in tech, we can cluster the tech people and highlight diverse voices on a number of axes.

You occasionally see people do manual lists that are useful, but I think we could of way better both in terms quantifying interestingness, expanding the kinds of lists we see, personalizing recommendations, and making a UI that makes it easy to form your own opinion (e.g. showing a few sample tweets, common connections, top followers, etc).

My suggested implementation for this would be to use a combination of crowdsourcing and mechanical turk.

The crowd could suggest organizations/people to score and bucket them into organizations. I’d start with unicorn tech companies and venture capital firms, but that’s my jam, so…

The app could then pull in their ‘followed’ list, or take a random sample for folks following large numbers of people where that’s impractical.

Mechanical turk could be used to annotate accounts. For example, we might ask MT to suggest a race, gender, sexual orientation, political bent, industry, etc.

Now we can answer questions like, “which venture firms’ partners have the most diverse following set?” or “who are interesting URM founders I might want to follow?”

This is a little bit crowdsourcing, a little bit AWS, a little bit of ML/likeness algorithm work, but nothing too crazy.

I will figure out how to pay for the mechanical turk for this if somebody builds it.

—————————

FreedomHatingTruthMachine - Scoring Journalism (Machine Learning/NLP/Big Data/Web Crawling)

It’s a shame we talk about fake news, which is a frame oriented around the product of media rather than the process, and allows us to argue about a specific controversy within a news cycle. Are these facts real? Did Trump say X? Is Hillary a Y?

I’m much more interested in a conversation around the process of journalism and what constitutes good journalism. Are we reading good journalists? Does this article meet journalistic standards? As technology platforms, are were biasing towards promoting propaganda and suppressing by omission (e.g. feeds) informative journalism covering sensitive topics? Are we amplifying via discovery (e.g. trending topics) 4chan propaganda laundered through irrefutable media masquerading as journalism?

I believe an input to any solution addressing the “fake news” problem will involve using technology to track and score media, giving publications, authors (named and pseudonymous), and individual articles a “journalism score.” We know humans can do this reasonable well. Now we’re just debating whether we can do this with computers, which I suspect we can.

Inputs to this score might be things like:

Using Politifact and other sources that use humans to assess the veracity of specific claims, and then using NLP to understand the extent to which articles discuss false statements critically, or support them narratively.

Using the social graph of reputable journalists to understand what articles, authors, and publications they deem high-value. Note: this might also require NLP to avoid the quote tweeting of morons problem.

Using the providence of claims made, i.e. understanding if a source is doing direct reporting (best), reporting claims from reputable sources (good), or shilling for 4chan (bad).

Scoring articles based on the journalistic code of ethics (https://www.spj.org/ethicscode.asp)

Other more awesome things that come out of thinking about this more deeply, collecting inputs from smart people, and working on the problem.

Collecting this data will involve a combination of analyzing information graphs, and will involve some very tricky decisions around how to match objective data (info graphs, “factual” providence) with human/subjective data (e.g. politifact) in a way that’s easy to explain, understand, and debate. In some sense this is a proof of concept and the methodology is the most important part.

Bold claim: With an intellectually honest approach, it should be possible to build a system that roughly rates publications, journalists, and maybe articles in terms of their journalistic quality without biasing against particular ideologies (unless your ideology is lies).

Bold claim #2: This kind of system is inevitable and will be a critical input when Facebook (et al) decide eventually that “real journalism” is critical to vibrant democracy and that “fake news” is an existential threat that requires they (transparently, carefully, etc) build an immune system for political facts into their currently defenseless platforms.

It would be best if this system were build to be open both in terms of code and data.

———————

MagicCoin - (blockchain/games)

We have collectable card apps like PepeCoin, but in addition to being racist they’re also boring. You can’t do anything with them. I’ve long thought that one of the coolest token apps you could build with bockchain would be a collectible card game similar to Magic the Gathering, Hearthstone, etc.

You could just even clone the MTG rules and build a simple starter set, and then fork Cocatrice and use it as an initial client for the token network.

This is not just your average blockchain project though. It has some really hard problems and will require some interesting design decisions in the governance model. These games involve printing new sets of cards (inflationary?), reprinting cards (inflationary!) to manage the price on a per-card basis (fed!), errata, and other economic factors that require careful thought.

There is likely to be some centralization in game mechanic design that would be taken on by the “foundation” supporting the game, or maybe not. This is your call. Maybe you can make an entirely decentralized rules engine off blockchain and standards will emerge.

But, given the scarcity properties of the blockchain, this seems like a perfect and fun application. I would love to be able to buy a set of awesome/fun tokens, pick a client that works for me, then take them to a game server that plays by the rules I like (Magic supports ~5 official formats & infinite kitchen table versions), and just have fun.

I am available for beta testing and game design input. Please also airdrop me some bomb-ass tokens if you make this.

How to get rich in tech, guaranteed.

Today’s NYT article on how employees sometimes lose out is a great read. Employees who like that, might also like Hunter’s article from last week on (not) getting rich at startups. This is the follow up I intended to write for the latter.

I talk to people looking for their next gig on a regular basis. It’s fun to match awesome people who don’t want to found companies up with companies who need awesome people. One question I get asked over and over again in various forms is, do you think I’ll get rich from startup X?

My first piece of advice for startup job seekers is that equity, all things equal, you can’t pick ‘em. On a risk-adjusted basis, startups are likely to be about the same. If there is information that a company has significantly de-risked, it will be priced in. Despite the market often being very wrong, you are unlikely to outsmart it.

If you want to get rich, your best bet on a risk-adjusted basis is to join a profitable and growing public company. Google for short. Make $200-500k all-in a year, work hard and move up a level every 3-5 years, sell options as they vest (in case you joined Enron), and retire at 60, rich. This plan works every time.

Beyond the sub-market salary you’ll receive for joining a startup, there are no financial guarantees. Your equity is probably worthless. The whole damn thing might fall apart any time. The hours are long. A lot of shit won’t work right. Etc.

But, as far as I know, startups are the only way to get 20 years of experience in five. The reason to join a startup is because you are awesome, you’re willing to work hard, and you don’t want to wait 20 years to be making decisions that impact the business.

And if you go in with this mentality, even when startups fail, you succeed. If you put five years into building a company and team, you will end up with a great network of talented and motivated people, lots of first-hand experience, and often some management experience as well.

Worst case, your next step could be going into Google at the VP level it would’ve taken you 15 years to get to joining out of college to “inject some startup DNA,” and catch up on salary within a few years. Unless this internet thing is a fad, that job will always be there for you.

But for all that is good and holy, don’t join a startup for the fucking money.

Sundry advice on picking a startup:

Be clear on what you want. Do you want to join pre product-market fit, or post? Do you need a salary, if so, how much? Do you care about vertical or role? Location? Travel? Etc. Most people that end up in the wrong job didn’t think through what their ideal job was before looking.

Run a process. Most people fail at this. Startups are interviewing you, but you’re interviewing them too. Write down your criteria. Look at 200 startups, contact 50, do first interviews with 20, move forward with 5, pick between 2-3 good options. If you’re passive, or only talk to a few companies, you’ll be choosing between limited options. You can get 200 plausible startups in a couple hours.

Focus on good people/culture. Above all else, my observation is that when you find good people (high-integrity, smart, hard working, etc) and a compatible culture, you end up happy, even if the company fails. When you ignore suboptimal people fit because you think the product is sexy or you’ll make money, you end up sad.

Accept fair comp. Many people are unrealistic on comp. They want an early-stage experience with Google salaries. It doesn’t work that way, and startups often suck at explaining that. Talk openly with startups about what they can pay you, what they’ve raised, what your needs are, and what milestones will lead to higher salary. Keep in mind, you might be the first person to take a salary and the founders might have been working for two years. Sure, it’s a 50% haircut for you, but for them it’s a big chunk of their first round.

Expect to earn it. Some startups will hire a dev and call them CTO. That doesn’t make them a CTO. If you want to be a CXO, say so up front, but expect to earn it. Entering a startup is a lateral move with unlimited upside. Tell the startup where you want to be, and set milestones that give you what you want that make the company successful. Now you’re a CXO.

Discount the vertical. With a few role exceptions, what you’re actually making isn’t that important. Assuming the company is making something people want and you’re delivering a ton of value, most people can be as happy making enterprise cloud infrastructure as social networking tools. Business development is business development. Do it with/for people who do not suck.

Understand the basics of the business. You shouldn’t try to become an expert, but you can ask some basic questions. What is the valuation of the company, and what’s the valuation for your purposes? How much has the company raised to date? How much money does the company have in the bank, and what is the net burn rate? What milestones does the company need to hit to get to the next round?

This is good to know so you can roughly assess the business. But it’s also a great way to understand how you fit in. Can you help with the core problems the company needs to solve?

Bias towards transparency. Companies can’t be expected to share every single detail with employees, especially potential hires. But, in general bias towards companies that give employees info to make informed decisions You should, for example, know what percent of the company you own. Trust is a two way street, and if the company lies to it’s employees, it’s hard to maintain that trust. Life is too short to watch your back inside the building.

Thinking Outside of Pandora’s Box

“And you ain't get a coin, [Nas], you was getting fucked then I know who I paid, God - Serchlite publishing Use your (brain)” - Jay-Z, Takeover

Pandora, it turns out, is an awful business. While it’s unclear what Spotify’s unit economics look like and it may be better position, it’s currently losing a lot of cash as it grows and admits that its margins are lower than they need to be. The core issue for these companies, as well as anyone else who wants to create a mass-market music service, is that you you need the labels to get scale, and they take all the profits.

While I generally think this is a shitty business, today I’m going to play devil’s advocate and suggest there may be hope for them yet.

Consider that more music is recorded today annually than ever in the past. If you looked at a graph of music industry profits vs recordings over the last 20 years, you’d conclude that they were inversely related. As it turns out, they’re completely uncorrelated, but you get the idea. Great music is being recorded that is making no one money today, because no one is hearing it.

But we still all listen to the same stuff from the same record labels. Why?

As it turns out, music has a network effect. Check out this amazing article from the NYT, which describes an attempt to test how hits become hits. You may also find this Planet Money piece on hit making interesting. From the NYT piece:

In all the social-influence worlds, the most popular songs were much more popular (and the least popular songs were less popular) than in the independent condition. At the same time, however, the particular songs that became hits were different in different worlds, just as cumulative-advantage theory would predict.

Put differently, some artists are better than others, but within the realm of good music, the hits that you need to have to make a platform work are somewhat arbitrary. We like music our friends like. We want to go to the club and hear a song we know.

The advantage the record labels have had even to date, is that they have controlled discovery in the form of radio (yes, it still matters), and therefor been able to monopolize as a group — if not strictly control — what becomes a hit. Internet nerds forget the scale of terrestrial radio and its importance even today. Ask Macklemore.

But the demographics of radio aside, tech music companies have achieved amazing scale today. Spotify and Pandora have on the order of 80M users each. Apple will likely catch up in the next 1-2 years. These platforms are arguably much better positioned than record labels + radio to make the hits.

You can see this in the power of Spotify’s and Apple’s Discover features. These are compelling features for users and have the ability to put unknown music and artists in front of millions of users. And given their current scale, it’s almost certainly the case that digital music companies at scale can make their own hits.

If I were one of these platforms, I would make the following deal with unsigned artists. Put your music on our platform and sign a non-exclusive, perpetual, royalty-free deal with us for all your future music. In return, we will pay you 100% of all profits for the first 10M plays with no middle man.

I would then start testing tracks from my stable of Macklemores, trying to find those that work for certain segments. With that as a baseline, I would have my editors attempt to create hits/stars via Discover (et al). And while my streams of Taylor help her grow her empire at near zero profit, I would be free to take 100% of the profit on these tracks, increasing my blended margin while continuing to keep the label’s back catalogs and paying them the profits on those.

And if that worked, I’d start helping my new stars complimentary products. Find service providers. Sell your merch. Book a show. Market to fans. You’re a (small) business, man. Let us help. Who needs a label? I bet there’s great margin in helping artists monetize their brands.

Is this a win for the artist? It depends whether you’re Tay Tay, or Tay-Tay-living-in-obscurity.

Is it a win for art? Probably. You make a lot more filling a bar with people who have heard of you than streaming online. No bar is too small for most artists. And the less power incumbents sitting on back catalogs have, the more innovation we’re likely to see in tech. That’s a good thing. It’s not clear more profit for labels has any impact on the quality or quantity of music.

Is it a win for music platforms? Obviously so. If current market dynamics continue, music will become a loss leader. You see it with YouTube and Apple, and it’s likely Pandora and Spotify will end up acquired if they can’t find a business with real margins. You can’t give 100% of your profits to other companies forever, regardless of scale.

The only losers here are the record label. And fuck them.

I’ll close with your obligatory reminder that artists don’t make money on recorded music, never have, and never will.

The Solution to the Problem With the Solution to Twitter’s Problems

Like most nerds, I’ve played a bit with Moments this week to see what Twitter has been up to. At the risk of being the one millionth person to give Twitter product advice, here are a few thoughts I haven’t heard discussed much in the analyses I’ve seen this week.

But first, let me say that we should keep in mind that it is a v1.0 and I should be looking for potential, not perfection. We should also keep in mind — as many have rightly pointed out — that people who already like Twitter aren’t the target audience, so it doesn’t matter if we love it.

The Problems

Twitter’s fundamental problem is that it is not big enough. They have a few hundred million users, which while impressive, is not big enough to be highly profitable given their business model. Advertisers need them to have more users, which means Wall St needs to see them grow.

Why Moments? Because they can’t grow Core Twitter. They have been trying for years and it has not worked. In part, this is a limitation of the product. Frankly, I thought it would never work outside of SoMa, so 300M users is pretty fucking good. They’ve now saturated a very large niche.

A second related problem, is the Twitter brand. There are no new users to go find. After almost 10 years, a billion users have tried Twitter and most of them had a bad experience. There are 700M people who think “Twitter” is not for them. Twitter has been trying to reacquire these customers for a while. Moments is an acknowledgement that this will not work.

The Solution

If Core Twitter is too complicated for the next 300M users, Twitter needs an app or apps that appeal to these users. They have begun branching out with Vine and Periscope already, and they’ve been pretty successful with these efforts. But Moments is the big bet.

Moments is an attempt to take the content that makes twitter appeal to addicts, and package it up in a way that appeals to Buzzfeed users. It’s something that if it works could make your mom finally say, “Wow, Twitter is pretty cool. I love the neat stuff I find there.”

Conceptually, I think most agree this is the right kind of solution to the problem. Now the open question is, will it work? Will users come back to Twitter for Moments?

The Problem With the Solution

My belief is that Moments is destined to fail as implemented today for one simple reason. The users Twitter expects to use Moments don’t like Core Twitter. How do you market a new feature in the app to these users? “Hey, I know you think you know what Twitter is and have already tried it, but we added a feature that makes it awesome and if you just give us another shot you’ll love it.” That is a hard sell.

Someone in my twitter feed suggested Twitter go all out advertising on Moments. I’m not sure that’s prudent, but again, what do you say to people who’ve already made up their minds about your product? How many of them can you really acquire?

Meanwhile, you do have 300M users to not alienate. This means you’re constrained in terms of changes you can make to the core interface. Secret could scrap their whole app and replace it because they had nothing to lose. Twitter doesn’t need 300M different users. They need 300M more users.

So today you have an app with a tab for novice users. It’s not bad for a v1 and I could see it becoming amazing. But it doesn’t solve the acquisition problem.

I lack the imagination to see any solution to this and have not heard one articulated. Believing people will just show back up now is magical thinking.

The Solution to the Problem With the Solution to Twitter’s Problems

Moments should be a standalone app. Period. It can still use the Twitter namespace and shared login. Be made by Twitter, etc. But Twitter needs something it can market to users that is not Twitter.

By making it a standalone, Twitter can avoid all of the problems that exist with implementing it as a tab. There is no dual onboarding experience, no @mentions, no main feed to educate the user on, no DMs, etc. Just great content and an occasional invitation to “join the conversation” or “follow all of Bieber’s posts” at just the right time. See graphic representation below.

“Have you tried Moments? It’s like Snapchat stories meets Buzzfeed. It’s my new go to on the train.”

Moments is in a sense a big bet that’s not big enough. They’ve added a tab that doesn’t offend the core user without removing the pieces that alienate the new users.

Furthermore, Moments is fundamentally different than Twitter in some ways that create friction. Moments has editors, and editors create bias (or minimally the perception). Editorial is slow, twitter is real time. Twitter is a tech platform, editors can piss off media subjects who use Twitter. Twitter is an uncurated social content network. Moments is a curated content consumption product.

Arguably, to make the best media product, Twitter needs to create a clear separation between these teams. You can see them struggling with this already, as genius Jay Yarrow points out in his pretty decent weekly podcast. This will almost certainly require them to firewall off the editorial team or to eliminate Twitter-employed human editors. Treating Moments as a separate property has the nice effect of solving this problem as well.

Then again, what do I know? I just tweet shit.

defaulting to positive

“Don’t care what they may say We got that attitude.” - Attitude, Kool AD

I got this uplifting note in my inboxes this morning. No doubt you received it as well.

What I love about Silicon Valley—what I believe makes it a special place—is that the default is to give encouragement. Whether you’re starting company to share what you had for lunch, or your jam is to make art from garbage, the default response you get from friends is some version of “I am into the idea of you being into your idea. Go do it!” Encouragement is the default.

This, in my mind, is Silicon Valley’s single most important cultural asset, accepting failure being the more often discussed compliment. It’s why we have more startups than anywhere in the country, more not-for-profits, and why this town ended up with a weeklong naked group vacation in the desert (hey, some things are the cost of doing business).

The cost of this optimism is relatively small. By definition, the people who “don't have what it takes” rarely get past bootstrapping, bad ideas rarely make it past seed funding, and in the end the experience and perspective gained by these founders makes them great contributors in other startups. Nobody has empathy for a founder’s struggles like someone who’s been there.

Most people have a chief hater inside their heads telling them they’re doomed. It costs you exactly zero to tell them how you think they could get their crazy idea to succeed, instead of the ways they will probably fail. If you’ve ever quit a job to work on a startup and had your mom tell you it was a bad idea, you know how meaningful a bit of encouragement can be.

Besides, maybe you’re just fucking wrong.

Twitter’s Product is Fucking Fine

It seems all the rage to write thought pieces on Twitter’s product problems. Hardcore users think they know the perfect ways to make it finally work. The point they miss is that there is no way to make Twitter (the product) a public market success. Twitter (the company) likely knows this, and is presumably working on a much more intelligent strategy.

Let’s review Twitter facts and core problems. Twitter had approximately 300M users, though as some have suggested that number is likely fairly inflated by bots/automatic integration, non-human account, etc. I would presume the real number of engaged humans is 150-200M (pulled out of ass). This is not a large enough social network for advertisers. They need more aggregate eyeballs.

Problem two, over 1B people have tried Twitter, formed an opinion, and left. The suggestion that increasing the amount of text you can plug in, or “open” is going to get these users to un-form their opinions is magical thinking. A strategy of re-activation has been tried (if you have an inactive account, you’ve seen the emails) and has not worked. Twitter’s brand is something very different to your cousin who set up an account in 2012 for a shot at free concert tickets than it is to power users. These users are never coming back to (product) Twitter.

Problem three, Twitter is worth too much damn money. It’s valued at $20B on ~$2B in revenue. An acquisition would likely need to be at $25-30B. Or, the stock price would have to drop to a point where it became rational to buy the thing as-is. People suggest Google and/or Facebook as buyers. While either could monetize Twitter better than Twitter, neither needs it strategically, and they likely couldn’t monetize it well enough to justify a purchase near this price.

The scenario in which Twitter becomes cheap enough to be bought given it’s current team/product is a scenario that plays out over too long a period to retain talent, in which case it arguably stays undesirable even at a lower price. Translation: Yahoo! buys them. Twitter needs to fix itself, in public. And they need to do it on vesting schedules while managing internal morale and improving external perception.

I’m pretty sure Twitter knows all this. My observation is that they are focused on improving engagement in Core Twitter (still important), while building/acquiring a suite of next-gen social products (Vine, Periscope, etc) that can be bootstrapped with Core Twitter. Don’t be surprised if you see some of the better Saccagestions spun out into standalone apps. Twitter’s core user base and existing content are sufficient to bootstrap more or less any kind of social product, and they have proprietary access to this asset. If I were them, I’d build, buy, and potentially invest in promising companies that could benefit from proprietary access to the graph/content.

You can’t go back in time and capture those 700M users who think Core Twitter is not for them, but many may give Sporter a shot, or TVitter. Again though, Twitter isn’t stupid, they get this. No doubt they’re working on finding the right focus internally and we’ll see new apps over the next year. The but is that most social apps fail (RIP Twitter Music) regardless of talent, and building them in public is a curse. And the ones that win, win slowly. Forced hypergrowth is a cancer that kills social products (RIP Google Plus).

So, as much as you’ve thought about Twitter’s product problems, Twitter employees have thought much more about them. And given the current state of the company, the core question we should be asking ourselves isn’t who the right CEO is to fix the product. The core question should be, who is the right CEO to buy the team enough time to get the strategy working and demonstrate to Wall St that Twitter the company is bigger than Twitter the product.

Now That’s What I Call Tweeting Volume 1

Someone on the twitters ask me where they could find a list of top SLJ tweets. As far as I know, there's not a service that does this. A few hours of coding later (not including rate limiting :/), here you go.

This is an approximate list of the top 100 SLJ tweets by engagement going back to 2011, weighted by date (to approximate follower count). Since the API doesn't allow you to do very much, and I suspect the web search (which you have to scrape for tweets beyond the last 3,000) purposefully returns incomplete results, we'll all have to live with the idea that the best stuff is still out there.

Enjoy...

Blessed is he who, in the name of profit, shepherds the user through the funnel, for he is truly his user's keeper.