BobbyH

My experience is that if you do not look like a white yuppie, cabs won't even stop for you to refuse service, especially at night. I once grew a goatee that made me look Hispanic, and I couldn't hail a cab at night to save my life.

I was upset about this and even once filed a complaint with the Taxi and Limousine Commission (TLC) after a cab slowed down, looked at me, then sped away to pick up a white guy down the street. However, for the complaint to result in a sanction, you have to show up at the TLC court to testify, or the complaint is tossed. I didn't want to spend a day in TLC court so I didn't follow up.

When it got warmer, I shaved my goatee and stopped wearing my leather jacket. Magically, I then got picked up every time. I was disgusted by this and now I take cabs only in extremis.

Basecamp lets you link to messages, comments on messages, to-do items, etc. It has a robust per-project auth/permissions system. Other collaboration sites offer something similar.

In general, when dealing with work that occurs via emails, it seems like it's easier to build a separate system that has email notifications, versus trying to glue features onto email.

> With inflation, he was actually being paid less 10 years later.

I looked this up and Neil Alan Smith's wages grew by 41% from 2001 to 2010, outpacing inflation of 23%.

Inflation from Aug. 2001 to Aug. 2010 increased prices by 23% (http://inflationdata.com/inflation/inflation_Rate/historical...).

Neil Alan Smith's wages started in 2001 at the federal minimum wage of $5.15. The federal minimum wage was constant from 1997 to 2007, but Florida passed a minimum wage law increasing its state minimum wage to $6.15 in 2005. Other adjustments to the Florida minimum wage increased his wage to $7.25 by 2010. (http://wiki.answers.com/Q/What_is_Florida_minimum_wage_histo... and http://www.floridajobs.org/minimumwage/index.htm)

One thing the article does not mention is "breakage": "Breakage is a term used in accounting to indicate gift cards that have been sold but never redeemed. Revenue from breakage is almost entirely profit, since companies need not provide any goods or services for unredeemed gift cards." http://en.wikipedia.org/wiki/Breakage

This article (http://www.journalofaccountancy.com/issues/2007/nov/accounti...) says that the average breakage is 10-19%. Let's say it's 10% to be conservative. Posie's Cafe said that over 1,000 customers bought the promotion. A 10% breakage implies that at least 100 customers bought but never used their Groupon coupon. Posie's Cafe only gets half of this, but it does give them a free $300. I guess that's not that much, but if breakage is 19%, 190 customers would have given them a free $570.

EDIT: using kareemm's datapoint of 30-40% below, the breakage would be $900-$1,200.

Yeah, in fact, Steven Jay Gould wrote a great article on how "As Mickey became increasingly well behaved over the years, his appearance became more youthful. Measurements of three stages in his development revealed a larger relative head size, larger eyes, and an enlarged cranium--all traits of juvenility." As he puts it, "Humans feel affection for animals with juvenile features."

full article: http://www.monmsci.net/~kbaldwin/mickey.pdf

The post makes a good point, but it does overlook the fact that the "Godfather of Anime", Osamu Tezuka (http://en.wikipedia.org/wiki/Osamu_Tezuka) was heavily influenced by Walt Disney and American animation.

For instance, the Wikipedia article notes: "The distinctive "large eyes" style of Japanese animation was invented by Tezuka, drawing inspirations on cartoons of the time such as Betty Boop and Walt Disney's Bambi and Mickey Mouse."

You can see how Bambi was drawn with large eyes here: http://www.google.com/images?hl=en&q=bambi

You can see Betty Boop here: http://www.google.com/images?hl=en&q=betty+boop

You can see Mickey Mouse here: http://www.google.com/images?hl=en&q=mickey+mouse+classi...

You can compare the style of those eyes to Tezuka's Astro Boy, first published in 1952: http://en.wikipedia.org/wiki/File:AstroBoyVolume1.jpg

Wordpress does this, but rather than serve the images from s3, it uses s3 to store the images and populate a self-hosted varnish cache: http://blog.apokalyptik.com/2007/10/10/so-you-wanna-see-an-i... This reduces the s3 bill by an order of magnitude (http://ma.tt/2007/10/s3-news/), so you may want to consider this approach.

In fact, using this approach, you could use s3 (just storage) to undercut s3 (storage+bandwidth) on cost and get lots of customers! I'd be in! :-)

He applied this CSS rule to all elements on the page: "letter-spacing:-0.1em !important;". This is why the page looks so cramped.

He also applied this to the body text: "font:140% 'Inconsolata', monospace;" I presume the site looks good in Inconsolata for those who have it installed.

This site is a good reminder to check your site on another computer. Also, make common font choices, or at least default to something specific that looks good.

Craigslist has manually screened every adult services post since May 2009. In a blog post last month, they noted: "before being posted each individual ad is reviewed by an attorney licensed to practice law in the US, trained to enforce craigslist’s posting guidelines, which are stricter than those typically used by yellow pages, newspapers, or any other company that we are aware of. More than 700,000 ads were rejected by those attorneys in the year following implementation of manual screening, for falling short of our guidelines."

http://blog.craigslist.org/2010/08/manual-screening-matters/

Craigslist also does other stuff, including phone verification for every adult services ad: http://blog.craigslist.org/2010/05/an-open-invitation-to-rac...

As Craigslist notes, this is a lot more screening than any other services does, including classifieds or the Yellow Pages.

Yeah, I agree. What I was trying to articulate is that this is a good example of a principal/agent problem, as the Principal (owner) has interests that differ from those of his Agent. Solutions to this generally either decrease information asymmetry or involve a better incentive system for the Agent. However, as you point out, it's possible that the Agent will act in his interest and still benefit the Principal.

Wow, I just looked and this exact example is actually featured on the Wikipedia page for Principle-agent problem. http://en.wikipedia.org/wiki/Principal-agent_problem#Employm...

If you think about it, it is economically rational behavior for a barista/bartender to do this for regular customers. After all, she gives you a drink that is free to her and gets back real money as a tip. The only loser is the coffee shop or bar, who has trouble tracking the number of drinks served.

My friend runs a bar and this is a huge problem for him that he calls a "bribe back". The only solution he found is hiring good people, as monitoring the bartenders makes him look like he doesn't trust or respect them.

It looks like you were downvoted a lot for this post: http://news.ycombinator.com/item?id=1646903

The karma for that post is actually less than -4, as explained elsewhere, the karma for a post can now be lower than -4.

Hi Ryan, I think that the rule of thumb used in the spreadsheet overstates the average customer tenure (as calculated by the alternate formula I outlined in another comment): https://docs.google.com/drawings/edit?id=1-9z5wczSfbM3KLZavF...

In case you find this worthwhile, I uploaded a spreadsheet showing my calculations which you can see here: https://spreadsheets.google.com/ccc?key=0Ag-jmDLo09WNdHBNY1d...

I use a different method for calculating the "Avg customer lifetime (months)", which is implied by some logic from a great book called The Loyalty Effect: http://books.google.com/books?id=ctAj_SfSrKIC&printsec=f...

Using this formula, I get different results for "Avg customer lifetime (months)" than the OP.

I calculate the average length of the customer tenure based on the formula: 0.5 = (1-churn%)^t, where "churn%" is the monthly churn rate, and t is time passed in months. Basically, this formula says: when will 50% of the customers be left?

You can solve for t:

0.5 = (1-churn%)^t

ln (0.5) =ln(1 - churn%)^t

ln (0.5) =t x ln(1 - churn%)

t = ln (0.5) / ln (1 - churn%)

You can test this math by calculating how long t is for a churn of 50% (it's 1 month).

Using this math, the average tenure for a monthly churn of 1% would be 60 months. The average tenure is useful because you can then do a discounted cash flow analysis on 100% of the cash flows until time t, to calculate the lifetime value of the average customer. So in this case, you would be discounting 100% of 60 months of cash flows.

The average tenure goes down rapidly as you increase the churn rate. At 2% churn, the average tenure is 34 months. At 3%, it's 23 months. At 5%, it's 14 months. And at 10%, it's 7 months.

If you have enough data, you can use a non-constant churn rate as well, as churn rate definitely goes down over time.

> [A peep hole] is [a disruptive product] if you're a door-to-door salesperson

Geoff Moore's theories apply to disruptive or discontinuous products that force customers to make a significant change in behavior. A peep hole may force door-to-door salespeople to make significant changes in behavior, but the customer doesn't make a significant change in behavior. The "disruption" or "discontinuity" refers to the changes in behavior of the consumer.

This sounds like a great strategy of using local customer and authoritative references to increase social proof to close sales.

However, the "bowling pin/alley" strategy mentioned in the articles applies in the slightly different context of a disruptive product. A peep hole is not a disruptive product.

A summary of Crossing the Chasm (a great book):

* All products are first adopted by Innovators, then Early Adopters, the Early Majority, the Late Majority, and finally Laggards. You may be an Early Adopter in one market (e.g. gadgets) but a Laggard in another (e.g. fashion).

* Disruptive products may do well with Innovators and Early Adopters, but the Early Majority is not willing to deal with a product that they don't understand (because it's not like another product they know).

* Happily, the Early Majority finds out about new products from the Early Adopters. So if you're building a disruptive product, you need to first make a critical mass of Early Adopters willing to recommend you to the Early Majority.

* To do this, he suggests that you make a "whole product" that addresses 100% of the needs of a particular niche (a "bowling pin"). The users in that niche will endure the pains of early-adoption in order to have 100% of their needs met (versus a generalized product that only meets, say, 80% of your needs). You will thus get happy Early Adopters in one niche.

* Then, you add the features you need to create a whole product for the next niche. As you succeed in each niche, you are not only improving your product, but you are collecting customer references in each niche. Eventually, if you have great customer references in a lot of niches, you can "cross the chasm" faced by all disruptive products by having enough overall customer references that the Early Majority will hear about and try your product.

* So, using this approach gives you two advantages. First, you have explicitly considered a bunch of niches and added features to make your product satisfy customer needs deeply. Second, you have a critical mass of customer references to "cross the chasm" to the mass market.

Here is an illustration: http://faculty.msb.edu/homak/homahelpsite/webhelp/Crossing_t...

Here's a link to the Techcrunch comments mentioned in the NYT blog post: http://techcrunch.com/2010/07/06/skype-recruits-yahoo-engine...

No, but they were "fat startups" aka startups that weren't founded on lean startup principles, like customer validation and customer development. Fat startups definitely can be successful, but it's more of a crap shoot. Lean startup advocates would claim that the probability of a lean startup succeeding is much higher.

HN does not use the popularity of previous stories from that domain as a weight. It uses the number of votes, adjusted for the amount of time it's been since the link or comment was added. So if a story has 8 votes after 1 minute, it will rank much higher than a story that has 8 votes after 1 day.

details: http://blog.linkibol.com/2010/05/07/how-to-build-a-popularit...

A Google News search found this tidbit from last month: http://blogs.wsj.com/venturecapital/2010/07/12/the-daily-sta...

"The much-hyped start-up is out looking for [$100 MM] from both strategic and financial investors to increase the manufacturing of its fuel cell-based power generators, VentureWire reports. The financing would come at the same valuation - between $1 billion and $2 billion - as the $100 million round raised last year. It’s already raised more than $400 million in equity so far from Kleiner Perkins Caufield & Byers, Morgan Stanley, New Enterprise Associates and others."

You could also use Hadoop to analyze data from, say, a global network of distributed pollution sensors to find factories that are polluting. Anyway, the point I was trying to make is that Hadoop is a nice building block. You could use Hadoop to mine any large data set, my examples were arbitrary.

Yeah, Hadoop is parallelizable by design, so it's just a matter of renting out EC2 clusters. In any case, I'm not talking about competing with Cloudera, but using Hadoop in some new specific vertical. For instance, an ad network that uses Hadoop could mine user/user-clickstream data better and get more clicks. That would be using Hadoop as a building block. You could also use Hadoop to power awesome analytics on, say, an e-commerce platform.

Other good building blocks to use are Hadoop (map/reduce) and Lucene/Solr (search).

P.S. Great post.

I feel bad for the Etherpad guys, who as I understand it, sold AppJet to Google to work on Google Wave: http://news.ycombinator.com/item?id=977015

http://www.custdev.com/ - Learn how to do Customer Development, an essential component of the Lean Startup philosophy. Learn how to: Find prospects and how to approach them, Develop and iterate your "Minimal Viable Products" (MVPs), Segment markets, position products and when to pivot, Validate before scaling"

http://www.lulu.com/product/paperback/startup-lessons-learne... - "Written by Eric Ries, one of BusinessWeek's Best Young Entrepreneurs of 2007, Startup Lessons Learned is a collection of 128 essays originally published at StartupLessonsLearned.com from 2008-2009. Each essay, or blog post, is unmodified and presented chronologically, rather then thematically, so that the reader can see ideas develop over time. Eric Ries provides actionable insight and advice on how to build successful products by listening to and learning from your customers. Startup Lessons Learned is a must-read for anyone trying to build a company in this new economy."

http://www.startupbook.net/ - "Start Small, Stay Small: A Developer's Guide to Launching a Startup"

Taking a step back, I think programmers should probably avoid working with "less-programmer-ey people" that are inexperienced at making products, especially when they over-emphasize the importance of "my fantastic idea". This is because inexperienced non-programmers tend to assume that programmers are just serfs that make their great idea happen, instead of treating a programmer as an integral partner in executing an idea.

This isn't their fault, per se. It's just that these sorts of people tend to have no idea how hard it is to develop a site/product. I would much rather work with somebody that respects how hard it is to create a site/product with high product-market fit.

If an inexperienced non-programmer demonstrates that s/he is legalistic by insisting upon a signed NDA, I would run for the hills rather than sign that NDA.

I mentioned that depression is possibly an evolved mechanism. The mechanism is: depression makes you ruminate which leads to insights. Meanwhile, you feel bad so you have an incentive to make a positive change in your life as a result of these insights.

Here is a NYT article about this: http://www.nytimes.com/2010/02/28/magazine/28depression-t.ht...

This really changed the way I think about depression. I now think about depression as a signal to ruminate, have insights, and make change.

However, don't try this at home, kids. I mean, it makes a lot of sense to work with a therapist/doctor, if you can find an effective one.

Your comment about "I don't usually have a cheerful face" really rings a bell for me. For years, I tried to force a cheerful face, even though it was like swimming upstream. Now, I simply accept that I have a low positive affect (not bubbly).

Marty Seligman would say we have a low "happiness set point" which is extremely hard to change. I know that Paul Ekman (http://en.wikipedia.org/wiki/Paul_Ekman) says smiling can make you happier. That may be true for others, but it didn't help me. In other words, I think that trying to change your happiness set point is a low ROI investment.

In contrast, for me, sleeping well and on a regular schedule is a very high ROI investment. So is lowering stress and spending lots of time in Flow.

Yeah, that can really help. The worst part is the next day hangover. Dealing with suicidal ideation and a hangover is not fun. Also, if you're like me, you feel bad about yourself when you "waste a day" on a weekend while nursing your hangover.