The Gong Show

I love HackerNews (HN).  The quality of comments there are fantastic. The behavior that emerges from the community is facinating (most recently, the OfferHN meme… or back when the entire front page was nothing but Erlang posts). The peer network support this community provides to entrepreneurs seems to really help people. Kudos to Paul Graham for fostering such an excellent service.

If I had any say in how HN evolved (and I don’t… probably for good reason), I’d do the following:

• I’d remove the New Comment link in the header bar. I don’t understand the point of it and the content hiding behind it seems pretty low-value, especially out of context. Perhaps back when HN was smaller the New Comment page had a better signal-to-noise ratio.
• I’d create a way to follow a few users I really respect. I want to see all the comments and submissions from the people I consider most insightful.  I know I can see them all on their user pages, but I don’t want to click through 10-20 user pages for all the people I find insightful… I’d like to see a single feed that threads all these people together.  
• Similarly, if I’m on a comment thread for an article in HN, I’d love to see some visual cue that highlights the comments from people I’m following.
• I want to create a personal domain blacklist.  It doesn’t have to be HN-wide, I just want to edit out a few domains for myself. The Internet reads Techcrunch to me often enough through other sources like Twitter and Techmeme… I don’t need to see it on HN too. In fact, one of the things I like best about HN is the random lesser-known sources it takes me too.
• I’d modify the bookmarklet so that anytime I click on it when I’m on a random page, if there’s an HN thread for that page it will populate right away. Right now, if you try to submit a page that’s already in HN, you’ll be taken to the previously submitted page’s convesation, which is great, but I think that should happen immediately upon clicking the bookmarklet, not upon submission of the link.  Then I’d have a quick and easy way to figure out “what does HN think about this?”

I’m sure some of these ideas could be accomplished with Greasemonkey. Maybe I’ll tackle one or two of them if I pick up some free time (ed. note — hahahaha).

Count how many computers are around your house.  Single-purpose computing is everywhere: in your car, alarm clock, set top boxes, microwave, digital watch, music players, gaming devices, phone (cordless and mobile), thermostat, etc… Not to mention the multiple old laptops and desktops people have lying around collecting dust.

It’s remarkable how many computers surround you at any given time, especially in such a wide variety of form factors and inputs.

I’m certain that communications across all these single-purpose computers will be possible someday.  Specification Standards efforts like ZigBee are certainly pushing the envelope in that frontier.  

I think any successful effort for communication across lightweight single-purpose computers will have to have the following characteristics:

1. Zero user setup. People can’t be expected to pair up devices (See: slow bluetooth adoption due to user experience issues).
2. Low power consumption. ZigBee is already designed with low-power consumption in mind, but the lower the better, especially if these communications standards will be embedded in anything powered by a battery.
3. Read/Write APIs (not Read-Only). If devices like a microwave or thermostat are communicating in such a way where they only allow other devices to read from their data, but not write or control anything, then I think the applications that will emerge will be somewhat uninteresting or underpowered.  All devices are going to need to expose themselves in a read/write fashion for the emerging applications to really be disruptive.

If you (or someone you know) is working on a startup in this space, I’d love to chat about your vision.

In my little bubble of web services fixation, it often feels like trends are bigger than they actually are.  For example, it feels like every job posting I see is hiring for an objective-c, Ruby or Python programmer.  PHP feels dead, and C++ feels deader than dead.  Yet, the truth (according to Indeed Trends) couldn’t be further from my subjective impression of the market.

This chart shows which programming languages are most commonly referenced in job postings over time since 2005. It’s a strong counterpoint to my assumptions:

I’ve seen a number of startups recently that use the term “check-in” to describe behavior on their site (even if the check-in has nothing to do with Foursquare). Similarly, there are a number of startups that use “tweet” directly in their name (even if they do much more than just add value on top of Twitter). In both cases these startups are using language that was popularized or invented by Foursquare and Twitter, companies that are real leaders in their category.

Using the language of another startup is a disadvantage because you are less likely to be seen as a thought leader in your sector. I’m not advocating that every startup go out and invent a bunch of new words: that would not be productive and would confuse your users needlessly. But, I think an exemplary story from Dick Costolo (while at FeedBurner) can help clearly convey the value of owning your own language.

FeedBurner did a great job of creating a lexicon to describe the value they created for users. For example running a feed through FeedBurner was called “burning” your feed. When industry blogs like Mashable would write about the category of companies that enabled you to get analytics and insights for your RSS feed activity, they would say something like “You could burn your feed with XYZ company, but really FeedBurner is the best way to burn your feed.”

You see what happened there… Even if the review of XYZ co is extremely positive, if you own the language that third-parties use to describe the service you provide, it’s game over. You are the de facto standard, the bar by which everyone else is measured. And that’s a great situation to be in.

So, as you think about the copy on your site, look for ways you can own your own language. If you can’t find any, that might be a sign that your product is not differentiated enough from the other services available in your market.

Naval Ravikant wrote a great (and concise!) post on a trend in popular web services of taking data that was previously private and making it public.  I totally agree with this observation, and want to add to his post. From Naval’s post:

All sorts of businesses are being built by violating assumptions about the privacy of data.

Flickr violated the assumption that you wanted your photos private by default. Before Flickr came along, the default photo sharing model, espoused by Shutterfly, Snapfish etc., was that of private photo sharing.

Naval then continues the post by counting a bunch of different services as examples (foursquare, Twitter, Blippy, Instagram, etc…). He then ends the post by asking which data set will next be given the “public by default” treatment, thus creating the next hot startup along the way?

My comments on Naval’s thoughts:

Public by Default is Not New

This trend of public by default has been happening for awhile (at least 5 years), and yet every time it is applied to a new data set, the result seems surprising and new.  

Ari Paparo wrote back in 2005 a reflection on the Social Bookmarking space after Joshua Schacter sold del.icio.us to Yahoo.  Ari wrote that del.icio.us succeeded where many other bookmarking services failed before because (in large part) del.icio.us was the first social bookmarking service to make all bookmarks public by default.

Ari’s post I’m sure is not the first to expound on the value of public by default in web services, but it’s the oldest example I can think of off the top of my head. The whole Web 2.0 boom has been predicated in part on this public by default approach, of which del.icio.us and Flicker were pioneers. And yet, even today, making user data public by default in areas where traditionally the data has been private feels like radically new innovation each time it happens. 

Why Is Public By Default So Effective?

I think there are a few reasons why public by default keeps working over and over again in different data sets:

1. Public by default means more useful data exposed to every user. (A simple point, but more data is better than less data… duh).

2. Public data allows you to start the typical “social validation -> voyeurism” cycle that drives the popularity of nearly all social sites. If a data set is public, you can put it in a reverse-chronologically sorted feed to keep things current for users.

3. More public data allows you to show interesting metadata that emerges from the underlying data (what’s historically popular, what’s trending, what’s popular in your social graph, etc…). If you try to do this with private (anonymized) data, you’ll likely end with with an accidental privacy breach (think Google Buzz and the Contacts leaks).

4. Public by default means that unintended use cases are more likely to emerge.  For example, one of the best blogs in the music industry is actually a feed published on del.icio.us by mediaeater. I don’t think Joshua ever intended people to “blog” on del.icio.us, but it works great! The same unintended use cases definitely emerged in Twitter that never emerged in the previous incarnation of status updates (AIM, Y!Messenger, etc…).

So, what’s next?

Here are some data sets that are ripe for startups to change from private over to public by default. Some (probably all) of these are already happening, but have yet to reach mainstream adoption.  I think they all will eventually hit a broad audience as we continue the trend towards living in public:

• Health (I know HIPAA adds huge friction here, but there will definitely be public patients in the future… we’re already seeing this in health coping forums and disease-specific blogs).
• Finance (kaChing, StockTwits, Covestor are the beginnings of a larger movement that I think will gain broad adoption on Wall Street eventually).
• Dating (Online dating today is a largely private process, which I think will open up over time as the stigma of having an e-girlfriend wears away).
• Consulting (Today, billions of dollars in cognitive load are hidden behind private documents and consulting relationships. This data will move into the open via services like StackOverflow and Quora, where consulting-level of quality data is available for free).
• Education (OpenCourseware is just the tip of the iceberg… Many classes will be widely syndicated through public channels.  All the economics are tied up in accreditation anyway, so there’s no reason not to open up class video and document feeds, especially at public govt-funded universities).

If you have the time, for a fun compare and contrast, read the following two articles back to back:

• A feature on Nick Denton in Wired from 2004.
• A feature on Nick Denton in the New Yorker from 2010.

Each article is a good read alone, but they are GREAT reads in combo.

Some pull quotes from the two articles for those with less time:

About the Gawker business model in 2004…

[I]f all the ads for Gawker are sold for the prices on its rate card, the total could be well over $10,000 a month. At the high end, that’s $80K or so net per blog per year - nice pocket change but not yet the stuff of moguls. Even if, as some analysts say, that amount triples or quadruples in a few years, we’re not talking about a bunk bed at Herb Allen’s summer camp. That’s why, Denton says, he’s not hiring full-time employees. Maybe it’s not a real business.

and in contrast, here’s what the business looks like today from the 2010 article:

As it is, given the thin margins of online publishing, Denton’s cultural impact greatly exceeds his revenues, which are somewhere on the order of fifteen to twenty million dollars a year. His ownership stake in the company is around sixty to seventy per cent, and every so often he attempts to consolidate by buying back shares that he has given to current and former employees. The rate he offered earlier this year would have put the company’s value at only thirty million dollars, or a fraction of what most analysts have estimated.  

And also this great quote from the 2004 article… What would the world be like today if Denton had been successful in buying Pyra Labs from Ev and Meg?

Denton had been speaking with evangelistic fervor about the movement for months. He had his own blog and even tried to buy Pyra Labs, creator of Blogger software, from founders Evan Williams and Meg Hourihan.

And one of my favorite parts… instead of just saying that Nick runs a network of blogs… Wired can’t stop calling Nick’s business “nanopublishing.”  So classically Wired!

In Nick Denton’s nanopublishing empire, that last word is usually accompanied by eye rolling and air quotes.

I don’t know about you but I’m definitely doing some “eye rolling” right now.

If you’re ever wondering the total number of registered users on a specific web service, there is pretty reliable method to figuring it out.

First: find a user in the service that signed up today.  Most services have a public feed, especially when they are still small, and you can find users that registered the same day in that feed.  Or, if that doesn’t work, you can create a new account, and you know you will have an account created today.

Second: go to the profile page of the new user. Sometimes a serial user ID number is exposed in the URL of the profile page ( which would look something like http://domain.com/user/12345 ) in which case, you’re done. You now know the total registered users number. However, most services these days use the usernames in the URL instead of a serial user ID number.  In that case, continue on to the next step.

Third: hit “ctrl+u” to get the source code for the page.  Then look for the URLs associated with either A) the RSS feed related to the user’s profile or B) the profile picture of the url.  You can usually find these by doing a search of the page for either “rss”, “xml”, or “img”. Make sure the user has a unique profile image and not the default image. In the URLs and filenames for the RSS feed or profile pic, you will likely find the serial user ID number.

If you still have not found the serial user ID number at this point, it’s very likely that the web service has intentionally taken precautionary steps to obfuscate their total registered users number, and you can give up the search.  

I don’t mean to be subversive or sneaky in anyway by posting this methodology. In fact, I’m trying to be exactly the opposite… I’m deliberately shining light on a common exposure of total registered user numbers.  There are fairly easy steps that any developer can take to obfuscate the total number of registered users, and by publishing this methodology, I’m highlights the low hanging fruit that developers can start obfuscating.

Some services don’t care about obfuscating the total registered users number, and that’s great.  For example, Tumblr publicly trumpets their total number of publishers (just shy of 9MM). I love that type of transparency; it’s great! But, I encourage any web service to make a very deliberate decision about whether or not to hide your total registered users number, don’t let it leak out because you are ignorant that you’re exposing this number.

Why would you want to hide the total number of registered users in a web service? An example from WWII helps illustrate the answer. During the war, the standard (non-mathematical) intelligence efforts estimated German tank production to be ~1200/month. But, when the Germans produced each tank, they stamped each one with a serial number. So, statisticians were able to take a sample of serial numbers from defeated tanks and use inference methods to determine that the Germans were actually producing exactly 246 tanks/month. This key piece of intelligence helped shape the American/UK strategy in the road to Germany. (read a better summary of this WWII story here)

So, by exposing your total registered number of users you are exposing information that could be used by your competition to their advantage.  That said, if you’re spending that much time worrying about your competition, I’d suggest your efforts could be better focused elsewhere.

Free has always been a cornerstone of web services competitive advantage VS offline businesses.  Craigslist free listings VS news classified pay per character model. Wikipedia’s free encyclopedia VS Microsoft Encarta for $25/license. Skype for free VS long distance landline calling charges.

But now, there are a crop of new companies emerging that are buying market share in established online verticals by going more free against competing web services. A few examples:

- Artfire: a competitor to Etsy and other crafting, handmade sites.  They charge no listing fees and take no percentage of the resulting transaction (compared to Etsy’s $0.20 listing fee and 3.5% of the transaction).

- Venmo: a peer-to-peer money transfer network. They charge no fees to transfer money to your friends, even if the transfer is sourced in part by debit/credit cards (compared to PayPal’s $0.30 transaction fee plus 2.9% of the transfer amount).

- Gamervision’s Swap-and-Sell Market: a marketplace to trade or sell videogames. All transactions have no fees (compared to eBay’s complex fee structure).

All of these competitors are in the early days of development. It’s possible that as they acheive scale and develop a network effect, they will switch over to incorporate more fees.  I’m sure some companies will do exactly that. But, I find it unlikely that they all raise fees in the end because fee hikes would undermine the reason why people flocked to them in the first place. And, it would open them up to being undercut by a startup, just like they are doing to established market leaders.

Instead, I see these companies as attempting a scorched earth strategy (best executed to date online by craigslist).  Why would I ever transfer money to my friends via PayPal when I can use Venmo for free? If these competitors do acheive massive scale, I think it will force the established market leaders to either A) lower price or B) construct features and barriers to justify higher pricing. Free is the lowest price possible, so now all these companies have to compete on the basis of A) size of networks and B) user experience.

I don’t know how the movie will end for any of these *more free* companies. But I do know one winner in the story: the end-consumer.

I’m holding office hours tomorrow (Wed 10/13) at the Cambridge Innovation Center tomorrow on the 4th floor from 3pm - 5pm.  Come drop by and lets chat about what you’re working on (or just say hi) if you’re in the area.  You don’t have to be a member of CIC to participate.

Any remember Google Answers? It was a Q&A service where questions were posted with a monetary bounty that would reward the answerer with a financial incentive. The answerers were experts at Googling around the net, and they often posted the Google queries they used along side their answers. It was a precursor to MTurk and all the Q&A sites online today, and in many ways it was a pioneer in this space of human computation.

Yet, Google Answers was considered a failure by Google… in fact, if I recall correctly, it was Google’s first publicly failed and shuttered product. Froogle was also considered a failure at the same time, but they never closed it off the same way they closed Google Answers. It’s remarkable to see in the wake of Google Answers’s failure just how hot the Q&A sector of the internet has become over the last year.

StackOverflow, Quora, and other niche Q&A sites have done something really awesome. They have fostered communities where people contribute answers to each other what would normally be $100/hr consulting advice *for free.*  Google assumed that you had to pay people for high quality answers to domain-specific questions, but it turns out people will willingly give away copious cognitive surplus in exchange for social credibility amongst their peers (points, badges, leaderboards, etc).  In short, paying someone with fungible reputation has proven to be more motivating than paying someone cash, in the Q&A sector.

However, fostering a community that collectively values the leaderboards and point systems in a given web service is very difficult… The tech sector is littered with dead, open-source StackOverflow clones that assumed “if you build it, they will come.”  So, the successful Q&A sites have proven that users can be better motivated without financial incentive, but constructing the incentive mechanics is the (relatively) easy part of the equation… the difficult part is building a community that actually cares about you.

I’m going to need some serious winter boots for the upcoming New England winter. Got any recommendations?  

I’m looking for something that can transition from the street to the office without looking too awkwardly informal.  I cruised around Zappos for awhile, but felt uninspired.  Got a pair you love?