This interview is ~30min. If you just want the audio, use the mp3 version. You can also get the video on your iPod/iPhone via iTunes.
For more interviews, visit the Traction Book site.
February 2010 Archives
Every estimate involves uncertainty, which creates error. When you present estimates, at least some mention of that error should be presented along with the estimates themselves.
On graphs, you often represent error with error bars. When written out, you usually add a +/-. Yet pretty much every estimate I come across does not represent the error at all!
Case in point, the Congressional Budget Office (CBO), a bi-partisan agency chartered by Congress to provide economic data to them.
CBO's cost estimates were center-stage at the health summit today. That fact in and of itself is fine. A big part of the health care debate should center around cost.
However, without knowing the error in their estimates, their predictions are useless. For example, suppose plan A is $500 million +/- $300 million and plan B is $400 million +/- $100 million. Can you really say plan A is more costly than plan B?
No, you can't. In some scenarios, plan A will be more costly, but in others it will not. That basic level of scrutiny is seemingly not being applied by Congress or the media.
I went to the CBO site to double check the possibility that they were accounting for error, but it just wasn't being picked up by the Congress or the media. I couldn't find any error bars or +/-s anywhere. Feel free to correct me if I missed them :).
I'm really proud to be in the Philly Inquirer today in both the offline (business, pg3) and online versions (Philly Deal$ blog).Thank you to Joeseph DiStefano at the Inquirer for the nice write up and to Charles Knight of The Next Web, Search for mentioning us to him.
Here's the excerpt:
Knight is also attracted to simplified search engines like Gabriel Weinberg's Valley Forge-based DuckDuckGo.com. "It's Google Light," says Weinberg. "They strip out all the garbage - video, ads. And it's intelligent. You search for 'wolf,' it'll ask, 'What wolf do you mean?' and list some choices."DuckDuckGo.com is the brainchild of Weinberg, a twentysomething graduate of MIT who sold his Web site, NamesDatabase, to Classmates Online Inc. in 2006, and retired to raise his child and invest in new companies with his wife, a GlaxoSmithKline P.L.C. statistician, in 2006.
"Around M.I.T., we had a lot of people starting companies," he said. "We started this group, Hackathon."
His Philly chapter "is growing slowly over time," with help from people at the LiquidHub consulting group, among others. They meet every month, sometimes in an office at Cira Centre, sometimes at the Bear Rock Cafe in King of Prussia. "There's random people making sites," he explained. "We try to put them together."
- I can't take credit for the "Google Light" quote, which must have been from Charles. I usually say "better results and less garbage."
- I'm thirty and I'm not retired; in fact, quite the opposite :)
- The Hackathon group is unique to Philly. I started it upon moving to the area trying to reconstruct some of the entrepreneurial spirit I felt around MIT. I'm actually at our monthly hackathon right now!
Many entrepreneurs think they are building an empire or sparking a powder keg, when they are really trying to start a movement.
I think beginning with the wrong expectation will greatly increase your chances of failure. You need to be prepared for the movement lifestyle. It's struggling to get customers, fighting to get people to care, or even to get people to give your idea a ten-second glance.
My last company was a powder keg. My current one is a movement. The differences in day to day life are staggering.
Containment vs convincing. Short term success vs the long haul. Be prepared.
Empires are rare. They're rare because they take a lot of money, time, and conquest to build.
Canonical examples are Amazon and Zoho. They start with a city (books, network management). Then they take go out and conquer neighboring markets, with no end in sight.
Eventually empires have to start building all sorts of infrastructure. It gets complicated. You're probably not building an empire.
Powder kegs are also rare. They're rare because they're all about the perfect idea, which involves doing the right thing at the right time.
Canonical examples are Skype, I Can Has Cheezburger and now Chatroulette. Once released they immediately take off virally.
You spark that powder keg, and it explodes. Then it's all about containment. If you can keep it under control, you can make an awesome exit, e.g. Hotmail, but you can just as easily implode, e.g. Friendster.
In any case, you probably don't have a powder keg to spark.
Attempted movements are a dime a dozen. If you're in a startup, you're probably trying to start a movement.
Movements are all about getting followers. You simplify your core idea and construct a message. You court evangelists, one by one.
If you can do this, eventually your movement may grow on its own. If it's a mainstream enough idea, it could grow into a revolution. Canonical examples are Wikipedia and Twitter.
More likely though, if all goes well you'll end up with a good business with happy customers who tell other people in similar situations about your products.
When you have your idea, you like to think of it as a "game-changer," i.e. an empire or a powder keg. With high probability, it is neither. It's a movement. Of course, it could eventually turn into an empire (Google) or have its powder keg moment (Twitter). Bu it won't start out like that.
So before you start focusing on getting traction, you need to know what kind of company are you starting. It will mold your focus and expectations correctly. And while your at it, also decide whether you are selling vitamins or painkillers.
Eli is 11mo old. Here's the semi-unintuitive stuff we've learned with regards to clothing him.
- Zippers. Love 'em. Baby clothes come in all sorts of different connecting pieces, e.g. snapping down the middle, snapping down the legs, and all sorts of other snapping. The problem is snapping sucks, especially when the baby is wriggly. With the zipper, you just shove in their legs, zip up and you're done.
- Footsies. Babies aren't supposed to sleep with anything in their cribs, but they can get cold. So what we've converged to doing is dressing him in a fleece with footsies with a onesie underneath. If the pajamas don't have footsies, then their feet get really cold, and so you end up having to put socks on them. Not only are socks annoying but they have a tendency to come off in the middle of the night. And even if they don't, without the footsies the leg can ride up and expose their calves.
- Snurgle. For the first few months we swaddled Eli, in what we came to call the Snurgle. Swaddling really worked for us.
- Onesies. They're awesome. There is a tendency to buy lots of different types of clothes, i.e. overalls, shirts, pullovers, etc. because they're all cute. And that of course is true. However, the onesie has a special appeal because they can just wear it as is and it classifies an outfit. Additionally, you can layer over it (pants/shirts) and if they get that dirty (food/whatever) you can take it off but don't have to take off everything. Finally, it keeps their belly covered, whereas if they just wore a shit it would not.
- More onesies. We've like the Gerber and Carter ones the best. You can buy the white ones in the multi-packs, which we routinely use as a base layer. Note, however, that the Gerber ones run small. You also have to be careful about the head openings being too big (with any brand). Finally, we've found that if you take the head size into account, you can get bigger sizes and they work just fine.
- Changing with the bottle. Changing clothes can be a pain. I've taken to making Eli a bottle first, and letting him eat while I change him. He is then calm throughout. This technique only started working after he could hold his bottle though, at around 6mo or so (I forget!).
- Sizing. The sizes vary so much from brand to brand it is somewhat ridiculous. It's to the point where you can't really trust them at all. Additionally, the width of the clothes varies a ton. Eli's really skinny, so that has a big impact. Bottom line is if you have to really look at the item beyond the size.
- More sizing. We bought a ton of 6-12mo clothes, which is the standard size. But really at 6mo they didn't fit Eli because they seem to be for more on the 12mo side even though they say 6-12mo on them. So we had to go back out and get a bunch of 9mo stuff, which is harder to come by.
- Stores. We found a lot of good stuff and good prices at Target, Carters, and Costco.
- Outside fleece. We bought a really thick outside fleece at Target that was proved invaluable. We put him in it in the late fall and early winter when we went outside for walks. He was warm and sometimes it lulled him to sleep. It covered his hands and feet and had a hood.
- Hooded sweatshirts. Sweatshirts are cute, but we were too nervous to put him to sleep in ones that had hoods.
- Shoes. Haven't had much of a use for them :)
- Laundry. We initially underestimated how many clothes he would go through. Between peeing, eating, and generally getting into things, we're changing him often a few times a day.
I was listening to C-SPAN radio this morning, and the governor of Montana was on. He said that governors wake up each morning thinking mainly of three things: education, medication, and incarceration. Overall, depending on how you count, these three things add up to about 75% of state budget expenditures.
Health care is very understated because it encompasses a lot more than Medicaid alone, including health care benefits and pensions of current and former state employees. The note at the bottom bears this out a bit. It should be really about 25%.
Corrections is also quite understated, because it doesn't include law enforcement and the judicial system, both of which of course are involved in the entire process of incarceration. It should be really about 10%.
As the site where this came from points out, these are averages, and the %s can vary widely from state to state. From the site:
"For example, Alaska and West Virginia each spend 11 percent of their budgets on K-12 education, while Michigan, at the other end of the spectrum, spends 31 percent. Similarly, Medicaid makes up 10 percent or less of state budgets in Alabama, Hawaii, and Wyoming but more than 30 percent of the budgets in Maine, Missouri, and Pennsylvania."
Can you name the top five line items of the US federal budget? I think knowing this answer is a pre-requisite to seriously participating in the ongoing discussions/debates over what should be done about the US deficit/debt. That's because cuts will have to come from these items to be significant in the long term.
I've been asking this question of people for years because I personally think it is important for a wide range of issues, including thinking about who to elect for President. I think people are most shocked at how small education is (2%), and after that science/medical R&D (3%).
Last night I went searching for the best info-graphic explanation. The following is the best one I found, which I got from here.
Depending on what you count in"Safety Net Programs", interest payments can be 4th or 5th. I think that understates its importance, though, because of how fast it is currently growing. If unchecked, it will be 10-12% in only a few years.
This interview is ~30min. At Jimmy's request, we did not record video, so I put his picture up throughout the audio. If you just want the audio, use the mp3 version. You can also get the video on your iPod/iPhone via iTunes.
For more interviews, visit the Traction Book site.
Continue reading Jimmy Wales on Getting Traction.
When I sold my last company, we had about twenty million cumulative user registrations, were signing up about one million new users a month and were sending out about one million emails a day.
When you get those kinds of numbers, the one in a million occurrence happens quite frequently.
- The craziest person you've ever met. An all caps rambling email that makes little to no sense.
- The angriest person you've ever met. The accusatory email claiming something ridiculous like you deliberately hacked into their computer or caused it to go haywire and they are reporting you to the Attorney General's office as consequence.
- The strangest piece of mail you've ever received. The African woman who insists your site is a dating service where you personally match people up and so physically mails to your PO Box a profile with picture.
- The code path you never knew was possible. How did that get in my server logs / in my database / on that screen shot?
- The check in the mail. We didn't take checks. Yet that didn't stop people from sending them.
The first times these happened, I was pretty amazed/scared/intrigued, or whatever the case may be. But what was really amazing to me, was how repetitive these events were.
These cases above happened on the order of one in a million, which means I was getting a few a month.
The same you-hacked-my-computer emails. The checks in the mail. And yes, I would receive about one letter a month from some African person who misunderstood my site and sent me their life story with picture. It was always from Africa. And I never quite figured that one out.
Happy Valentine's day. Here's something I hate: tl;dr comments. tl;dr stands for "too long;didn't read." I've seen these types of comments increasingly on Hacker News lately, the only online forum I routinely visit. Searchyc (a site that searches the forum) confirms the increasing frequency of these comments over the past few months.
People who comment tl;dr are pretty much confirming they didn't read the article they are commenting on, which means the chances of their comment being useful is slim.
The worst is just 'tl;dr' and nothing else, or perhaps even worse, a tl;dr followed by an emoticon. It adds nothing except that the article seemed long in the commenter's opinion. And it's really smug, which annoys me in and of itself.
Slightly longer tl;dr comments try to summarize the article, which can actually be useful. But I'd strongly prefer if they didn't use that abbreviation; just say summary or abstract. Saying 'tl;dr' just brings that smug tone into it.
Finally some people say they serve a purpose to inform writers their posts are too long. I don't buy that, especially on a forum like Hacker News that works on a voting system. Just don't upvote it.
I think pg (the forum creator) agrees, though I sent him an email about it and never got a reply back.
Here's my crazy social network setup:
- I submit status updates to Twitter, which auto-posts to Facebook and LinkedIn (and now Buzz).
- I submit pictures (mainly of my son Eli) to Posterous, which auto-posts to Facebook (and has eliweinberg.com pointing to it).
- I write blog posts, which get auto-posted by feedburner to Twitter.
- I upload videos to YouTube, which I then post to Posterous, and which get auto-posted now to Buzz as well.
- I upload videos directly to Facebook because they come out better that way (in HD).
- I submit stories to Hacker News when appropriate.
- I get auto-emailed comments from basically everything.
- I check twitter and Facebook (and probably now Buzz) regularly.
I think I'm way out of the mainstream on this behavior because I'm (a) highly technical; (b) want the widest possible audience for my blog; and (c) want anyone who desires to be able to get the latest pictures of my son as easily as possible.
The mainstream is never going to do the above. But will they even ever adopt more than one social network in the long run?
That is, in a few years, will you see most people on Facebook now also regularly checking and posting to something else, e.g. Twitter or Buzz?
I don't think so. Friendster died. Yahoo 360 died. Myspace is dying. I'm starting to really believe Twitter and Facebook and now Buzz are in a winner-take-all battle for mainstream social network attention.
Yes, they're all a little different, especially Twitter, which can have its special use cases. However, as time passes and the battle matures, I can only imagine that Facebook and Buzz will try to capture some of these use cases.
I don't think the solution is more personal aggregators. Those are for tech people.
Finally, I don't think the UI experience on any of them are particularly good. They succeed in spite of themselves.
People are mad about the recent Supreme Court decision on the right of corporations to influence political campaigns. Lawrence Lessig is even calling for a convention to restore our democracy, which he insists is broken. The movement has even drafted a proposed constitutional amendment.
This pattern (court decision -> outrage -> proposed amendment) happens every few years. Well, I'm sure it happens more often than that, but it actually gains some traction every few years. The last one I recall was around eminent domain and the Kelo case.
In any case, I agree with Lessig that we should call a convention:
"If 2/3ds of the legislatures of the states demand it, Congress must call a convention. That convention then must meet and deliberate about amendments to the constitution. If it agrees, it then proposes amendments to the states. 3/4ths of the states must then ratify any amendment before becomes law. Thus, 12 states of 50 have the power to veto any change, meaning no change could happen unless it appealed to a solid group of Red States and a solid group of Blue."
However, I think the scope of that convention needs to be much bigger. Yes, we should call a convention, but it just shouldn't be about this one problem, one amendment.
If we're going to call a constitutional convention, we should really consider a host of amendments, i.e. a modern Bill of Rights.
The Bill of Rights was ratified, in part, because it was many amendments, not one. In the same way, proposing many together would actually give the movement legs, which other amendments have failed to capture by themselves.
A more substantive argument is that our Constitution is need of a lot of amendments to clarify things. Corporations' status as people/citizens and eminent domain are two good examples.
Equally good (and popular) are finally spelling out the right to privacy, a right to eduction, and raising the standard of gender and ethnic equality to that of race and religion. Less popular but possibilities are a social security trust fund, no military draft, and direct election of the president. There are of course many other possibilities as well.
I think people dislike the Supreme Court so much because they think they just make stuff up. That's of course not true (read opinions), but there are a whole host of issues that the Constitution really doesn't have much to say about. In those cases, they do try to make strong logical arguments (mostly), but they aren't really backed up the way should be (with the Constitution). A modern Bill of Rights can help us get back that foundation.
The YC/Ron Conway angel investing strategy seems to me to be: focus on finding really smart, determined people, and the returns will follow.
Their strategy makes perfect sense for them, but not necessarily for me.
Here's them:
- Essentially unlimited funds to invest in who they want.
- Incredible deal flow, probably capturing a large % of promising early stage startups in their investing spaces.
- Lots and lots of deals.
Here's me (you probably know what's coming):
- Orders of magnitude less funds.
- Orders of magnitude less deal flow.
- Orders of magnitude less deals.
Paul Graham was just on Mixergy saying potential exits are so unpredictable, he doesn't even think about them when making investment decisions.
My premise is I can't do that because I simply don't do enough deals and don't have similar deal flow. Given these facts, I think angels like me probably need a modified strategy. Essentially, we should have a higher bar to invest.
Yes, we should only invest in smart, determined people. Yet I think we also need to think at least a little bit about exit probabilities and magnitudes.
But doesn't that contradict Paul Graham's point that they are inherently unpredictable? I don't think so if you concentrate on more realistic exit possibilities.
That is, I agree that exits are unpredictable, but big exits are way more unpredictable. Googles are pretty much black swans. IPOs are less rare, but still pretty rare (at least in the past decade). And 100M+ exits happen somewhat frequently, but are still rare enough that YC hasn't had one yet in its five year history.
But they will--it's just a matter of time. Ron Conway has had many big exits over his longer time span. Both YC and Ron Conway will make their great returns on these rare, unpredictable events.
And that's the thing. I don't think I can count on being part of any big exit event at my small deal rate. But I think I can ensure more of a chance of decent returns (as opposed to super great returns) by thinking a bit about potential exits.
I should probably look for founders with some idea about customer acquisition and/or market. For example, a founder doing something interesting in the dating space would probably be more interesting for my strategy than a founder doing something in a brand new market with great yet completely unproved potential.
The bottom line is their strategy might get them in on the next Google. However, me following their strategy at my deal rate will most likely not get me anywhere close to the next Google. But with a modified strategy, I can probably still make decent returns and work my way up to their strategy, which I would of course love to do if I could.
I've gotten more serious about angel investing lately, and I'm struggling with how to value prospective angel investments.
Let's say my target is to get 30%/yr on my angel investments. If my investments have 5 year time horizons (on average), and 10% hit the high mark, that 1 out of 10 that hits needs to have a 37x exit event (on my share). If I was more successful at picking winners, say 2 out of 10, that drops in half to 19x exit events.
By the above logic, which I'm starting to believe, I should only consider investments that have a realistic opportunity, albeit unlikely, to return 20x on my investment. Should I have such a cutoff?
As my share in the company goes up, it gets more likely that their high exit outcome will return the required amount. I probably won't put in much more money per deal, so a higher share for me means getting in early and/or being really hands-on.
This is a sysadmin post. The other day I was running a background process on a production machine. I thought it wasn't going to eat up many resources, but I was wrong. It turned out it was doing a lot of random I/O and things ground to a halt.Now often I will have top, vmstat and iostat open and renice annoying background processes to 20 when appropriate (r in top). If you don't already know, 'nix processes (I use FreeBSD) can have priorities, which the scheduler takes into account when giving out resources.
These priorities range from -20 to 20 (on FreeBSD at least), and you can see them in top under the PRI column. 0 is neutral. If you set something to 20, it will be tied for lowest priority process in the system.
nice is another command that messes with process priorities. It starts them out at a particular priority, as opposed to changing a priority via renice. For example 'nice -n 20 ./process' will start process at priority 20.
Then I got to thinking, why don't I do more of this initial nice stuff? Maybe seasoned sysadmins all already do this, but I went back through all my scripts and crontabs and explicitly set nice values. Then I went through my daemontools run scripts and set nice values there as well.
My web server (nginx) already did this via the 'worker_priority' variable, which I had previously set to -6. I set negative values to my most important scripts and relative values between them in order of importance. I set positive values to less important scripts, and then 20 to various background processes kicked off via cron. For example, backups are now niced to 20.
The system was already runnnig pretty smoothly, but now it is even more so. And I think, more importantly, it will react better in times of need.
Final tip. When I want to renice a bunch of stuff, I usually do so on the command line instead of top, e.g.:
ps auxww | grep -i crawl | awk '{print $2}' | xargs renice 10
That will take all the processes that match 'crawl' and renice them to 10.
This interview is ~40min. If you just want the audio, use the mp3 version. You can also get the video on your iPod/iPhone via iTunes.
Thank you to Prakash Swaminathan for introducing me to Alexis for this interview.
For more interviews, visit the Traction Book site.
Get major Traction Book updates, like when it's ready :)
