HNHacker News
TopNewBestAskShowJobs

emmett

4,858 karma · joined February 19, 2007

Founder and CEO of Softmax. Softmax is building massively multiplayer benchmarks for coding agents to train social intelligence.

Previously very interim CEO of OpenAI, YC Partner, and founder and CEO of Twitch.

Email me at emmett@softmax.com.

submissionscomments
emmett··on Twitch nabs exclusive streaming deal with Blizzard for e-sports events
Golden loot boxes are basically guaranteed to have good loot in them...try it, you might like it !
emmett··on Twitch nabs exclusive streaming deal with Blizzard for e-sports events
We care a lot about stream quality at Twitch. Can I ask where you're watching from that you get so many stutters? That's a key metric we measure ("buffer empties per hour"), so I know for a fact it's not high everywhere.
emmett··on Options vs. Cash
If you join a startup immediately following a Series A and you only get 0.01%, you almost certainly got screwed. We were fairly stingy with equity, and an engineer joining then would have gotten around 0.5%, 50x what you're basing your math on. And over a 4 year vesting period, not a 6 year one.

$2mm ($40k x 50) for your 4 years of work is substantially less bad. And employees who stayed with us tended to get new equity grants over time as well.

So yeah, don't take a startup job for tiny amounts of equity. You should get significant equity for joining that early with that much risk.

emmett··on Ask HN: How much did you pay for your brand identity?
The first design was expedient: we needed something in order to launch the website. So we did a really minimal amount of work. It was never great -- nothing done in four hours can be great -- but it sufficed to be able to launch a product.

After about a year, we had clear product momentum, so it was worth investing in a better brand design since we understood our direction and customers at that point well enough to do a good job that we wouldn't have to change later.

emmett··on Ask HN: How much did you pay for your brand identity?
The first version of the Twitch brand, space-themed, was created by Jacob Woodsey (our only designer at the time and now owner of all social features at Twitch), in approximately 4 hours spread over the course of a single day. So probably equivalent to $1500 to get a high quality designer for a single day.

For the second version of the Twitch brand (Purple!), one year out and post-series-B: we had 3 designers on the Twitch brand, not quite full-time, for about 6-8 weeks. Plus a lot of Jacob's time. I'd estimate the total cost around $50-$100k to hire a high end design firm to do the same work.

In both cases, it was the right level of investment for Twitch at the time. Something quick and dirty to get us going in the first year, and something with some depth and quality once we had enough momentum to commit to a brand direction.

emmett··on Uber C.E.O. To Leave Trump Advisory Council After Criticism
Trump's senior strategist, his appointee to the Security Council and political advisor is Steve Bannon.

  “Isn’t the beating heart of this problem, the real beating  
  heart of it, of what we gotta get sorted here, not illegal 
  immigration?” Bannon asked Miller. “As horrific as that 
  is, and it’s horrific, don’t we have a problem? We’ve 
  looked the other way on this legal immigration that’s 
  kinda overwhelmed the country?”

  Bannon goes on to decry the “oligarchs” of Silicon Valley   
  and Washington and call the number of immigrants in the 
  United States “scary.”
It's very hard to say Trump's given "no sign" of being against legal immigration. He has yet to actually put forward an anti-legal immigration Bill in the first two weeks of his presidency, but it seems likely he will given his choice of advisors.
emmett··on John Carmack on expert witnesses and 'non literal' copying
My father was an expert witness in electronic discovery cases for many years in the 90s and early 00s.

I remember so many stories growing up about similar situations. The main lesson was that it was actually way easier to be an expert witness when your side was totally in the wrong, because you could just make stuff up. Explaining the truth about how computers work is hard, compared to making up simpler explanations that don't happen to actually reflect reality.

emmett··on Daniel Gross of Apple leaves to become Y Combinator’s newest partner
You don't need partners great at scaling profitable companies from 100 to 500 employees.

You need partners who know how to help a company go from two to ten employees and raise a first round of funding.

I'm a part-time partner, and I run a company with >500 employees, but I basically draw entirely on my first 2-3 years of experience when giving advice.

emmett··on Waymo: Google's self-driving car company
Do you really think it hasn't occurred to WayMo/Google that maybe they should test their cars in snow/rain/sleet/fog/confetti?
emmett··on Buttery Smooth Emacs
A famous aphorism of David Wheeler goes: "All problems in computer science can be solved by another level of indirection" (the "fundamental theorem of software engineering"). This is often deliberately mis-quoted with "abstraction layer" substituted for "level of indirection". Kevlin Henney's corollary to this is, "...except for the problem of too many layers of indirection."

https://en.wikipedia.org/wiki/Indirection

emmett··on Where San Francisco Wants New Subway Lines
I went to a private school. It was pretty non-traditional in many respects, and it wouldn't surprise me if writing a thesis wasn't common at other schools.
emmett··on Where San Francisco Wants New Subway Lines
Chicken and egg. Subways enable density. Right now if you greatly increased the density of many SF neighborhoods, you'd overwhelm the transportation infrastructure. (Obviously there are neighborhoods with transit that could already support increased density).

Hilariously, I wrote my highschool senior thesis on this exact topic. It was a case study on Seattle, Houston, and New York and how density changed before/after transit was added. You can see the pattern pretty clearly, transit leads to density and not the other way around. Which makes sense, right? Because density can be increased one building at a time, but generally subways need a huge investment to put in an entire line at once and benefit from having an extensive network.

emmett··on Elon Musk on How to Build the Future
I'm not super familiar with facelift costs over time, but a quick google gives me this link: http://www.drhodgkinson.com.au/news-resource/different-types...

Quoth the article: "Facelifts have come a long way in the last 20 years, not only in terms of technique, but also in terms of accessibility to both women and men. In the 1970s and even into the 1980s, the facelift was a luxury reserved for the rich and famous."

This leads me to believe facelifts have greatly declined in cost over the past 30 years. The number of such surgeries has greatly increased as well ("Since 2000, overall procedures have risen 115 percent, but the types of procedures patients are choosing are changing." -- http://www.plasticsurgery.org/news/2016/new-statistics-refle... )

Do you have any other examples? Because the one you gave doesn't appear to support the argument that prices will remain high for long periods of time. Especially given that facelifts are a cosmetic surgery and thus there's relatively little drive to give them to everyone.

emmett··on Elon Musk on How to Build the Future
Historically, expensive treatments for the rich have become universal treatments for everyone over time.

What makes you believe that life extension will be any different? Or, if you disagree that expensive treatments generally stay expensive, what are your examples?

I agree we could have a temporary awkward period in the middle, say 20 years, where it's not cheap yet. But on the scale of history that's a short period of time...I'll admit that's cold comfort to those who die in the meantime.

(See: antibiotics, insulin, appendectomies, lasik, ...)

emmett··on The "Acid King," Serving Life Without Parole
You are exactly correct. Nothing is off-topic on Hacker News; the rule is "anything that gratifies one's intellectual curiosity", so that's a pretty wide mandate.

I know you're saying you're not complaining, but you clearly are. Complaining things are off-topic isn't really helpful.

I don't think the article is very good, but the problem isn't that LSD usage is somehow off-topic.

emmett··on The Dreaded Weekly Status Email
Status emails are not a "Potemkin supervisory framework". They're a scalable way of communicating progress broadly in a way that can be referenced later, rather than an ephemeral way to communicate to one person.
emmett··on YC’s Summer Reading
I don't think my anecdotal evidence is particularly stronger than the author's. My point is just to take it all with the knowledge that what worked at google in an experiment there won't necessarily work for you in the same way. The world is complex and these things are contingent.
emmett··on YC’s Summer Reading
I love scifi, and but I totally agree. I thought Ready Player One was terrible. My only explanation is that it's a huge nostalgia rush for a certain set of Gen X readers.

I really liked Mr Penumbra's 24 Hour Book Store, but I think for the same sort of nostalgic reasons as people love Ready Player One.

To each their own though...

emmett··on Stealing Facebook access_tokens using CSRF in device login flow
I'm pretty sure a first year security engineer at Facebook gets paid a lot more than $5k too. They're not your employer, they don't owe you money, if you want to mess around looking for security vulnerabilities in your free time it's good of them to pay you at all.

I fully support Facebook paying bug bounties, but let's compare apples to apples here.

emmett··on Investor Day
I still don't see how on average it's worse for students. It might be worse for student A who would have applied to MS knowing a higher chance of getting in, but that's then probably worse for student B who genuinely wants to work at MS and doesn't get the slot that A took.

Isn't it better if everyone expresses their true preferences and then we do the best matching we can on that basis, rather than having students attempt to game the system in response to company ratings? Your description makes the system sound close to optimal to me.

emmett··on Investor Day
Why is sharing the rating with students so critical? I can see why people would be curious, but I'm not sure why it matters so much?
emmett··on Later Stage Advice for Startups
(1) I totally agree that there are people who violate this rule and do great. Hiring really senior early can totally work, especially if you get the product right the first time. In fact, it can be a huge accelerant.

(2) There's a difference between the founder of a startup and early employees. The founder has to be a manager from day 1. I wish I had been more experienced starting Twitch. I generally think more experience for founders is a positive, especially when it comes to management.

emmett··on Later Stage Advice for Startups
There are multiple ways to be a "senior" employee.

The first is to be a very experienced individual contributor. Amazing programmer with deep experience, world class designer, sales person who can get to the right person in any org. These people are gold. Hire the best and most experienced you can.

The second way to be senior is to be great at hiring and building organization. These people are very valuable later, and will save you a lot of heart ache. Hiring them early is a mistake in general, because their instinct for solving problems is to hire great people and develop great process rather than to simply solve the problem themselves. For an early stage startup there are too many unknowns and this amounts to premature optimization. Later on this is the a critical skill set.

Inexperienced managers are nearly universally terrible. Inexperienced individual contributors can often make up for it with raw talent, and you gain experience fast at a startup. Therefore early on experience is not at a premium, you should hire primarily for talent above everything else. Later on experience becomes increasingly valuable.

Short version: Hire the best makers you can early even if they're junior, hire the best managers you can later which means more senior folks.

emmett··on “I was fired from my internship for proposing a more flexible dress code”
From the article:

A note about this post, which is being linked to from all over the internet: This situation is not about “young people today.” The letter-writer’s generation is far from the first to bridle at dress codes or misunderstand office culture or start out with little knowledge of how things work in offices. This is about being young and new to the work world, not about what generation they belong to. Most of us made plenty of mistakes when we first started work — I certainly did. So please go a little easier on this person.

emmett··on Twilio S-1
If you had to guess what the net loss for Facebook at IPO was, what would you say? More or less than Twilio?
emmett··on Flash-Free Video in 2016
We're working on the switch (I announced at TwitchCon). It's particularly hard for live video, but that's just an excuse. Clearly the time has come.
emmett··on When to join a startup
While I think you describe the unfortunate common case, my personal experience has not run this way.

Much more often, there has been agreement that there is a need for a new layer of leadership that we did not have before, and then we have to decide whether we promote or hire on top of someone. When we have had to hire on top of someone, they have taken that option gracefully while gracefully also keeping the package of other options that they were handed when they joined.

"Report to the new person" works surprisingly well if you sell it the right way.

emmett··on Economic Inequality: The Simplified Version
Intrinsic wrongness of inequality is basically argument #4 from your link. It's the idea that the split of wealth is just inherently unfair, not because of any bad consequence, but because of something being wrong with having wealth too unevenly distributed whether or not it actually has any bad effects.

Claims 1-3 from your link are claims of instrumental wrongness of inequality claims 1-3 from your link. They point at bad consequences (rich people will control the media, rich people will corrupt the government, rich people will have unfair advantages for their children).

As the author of the article you links points out: "In principle, these effects could avoided, without reducing economic inequality, through such means as the public financing of political campaigns and making high-quality public education available to all children". It's not that inequality necessarily must cause these consequences, it's that it will tend to under some (most?) given policy regimes.

I think I misunderstood your original comment...when you said that "Inequality is not per se bad" was incorrect, I assumed you were making argument #4, which I did not see debated hardly at all. Most people were making some variant on arguments #1-#3.

In which case, I would respond to you: Paul does talk about the bad consequences and bad causes of inequality, and claims that you should address those causes and consequences directly rather than attempt to limit inequality in itself. The argument against this has to come in some form of "well yes, in theory it's possible to limit the bad consequences of inequality, but in practice it's too hard because X and thus we should directly attack inequality".

I didn't see anyone really saying that, everyone seemed to believe Paul thinks inequality doesn't have bad consequences or that he thinks we shouldn't address those bad consequences.

emmett··on Economic Inequality: The Simplified Version
The previous thread had lots of reasons economic inequality is instrumentally bad. I really saw nothing about how it's intrinsically bad, and I was looking. Can you link to a specific comment maybe?
emmett··on Economic Inequality: The Simplified Version
Great point! That sounds like a genuine disagreement. Can you expand a bit on why economic inequality is intrinsically bad, rather than instrumentally bad?
← PreviousPage 2 of 30Next →