I’ve been programming for mobile for 15 years. “What?! Mobile wasn’t around in 2000,” you say? Well, you’re forgetting about the original mobile, the TI-83 graphing calculator.

BASIC on the TI-83 was my first experience with programming, and it all started because a teacher told me I could use calculator apps on tests if I made them myself. I’ve been programming ever since, and it has shaped who I am as a person. This is just one example of the profound effect teachers can have.

Today, I’m proud to announce I’m joining a company that helps teachers share these transformational experiences: Teachers Pay Teachers. Teachers Pay Teachers is the world’s first and largest open marketplace for educators to buy, sell, and share their original resources. It’s growing like crazy and has already paid out over $100M to teachers.

I’m excited to join a great team including:

• Sha-Mayn Teh (previously an engineering manager at Google)
• Adam Freed (previously COO at Etsy)
• Jared Cohen (previously COO at Kickstarter)
• Kirsten Nevill-Manning (previously director of FB recruiting)
• Scott Carleton and Alexis Tryon (from Artsicle who I met during the good ol’ days at Dogpatch Labs)

If you’re interested in hearing more, ping me at nhurst@teacherspayteachers.com - I’d love to grab coffee, hear what you’ve been doing, and tell you what we’re up to!

Dear friends,

I made something for you this Christmas. I hope you'll enjoy it!

Each year I have a secret I use to find gifts for people: go to Amazon, find two things I know my gift recipient likes, then compare the "similar items" section to find matches. It's kind of tedious to do, but it works like a charm. Enter computers. Exit the tedious.

This year, I made a website that does the same thing without any effort: GiftMating.com

As elections draw near, remember that it's not just about the President. We'll vote on a number of Congressional seats too.

Notes

* Data courtesy of ProPublica
* Some SOPA supporting senators and representatives are not mentioned here because they're not up for re-election or aren't running again.

Please tweet/share with your friends in other locations!

I recently saw Latency Number Every Programmer Should Know and was having trouble getting my head around the scale, so I made this visualization to help me out.

I converted latency times to their equivalent light distances (how far light travels in that much time) and plotted those distances on a map to get a sense of the latency numbers in relation to each other.

Source Data

The smallest few latency times are included here with their respective comparisons.

About me

My company runs "speed interviewing" events over video chat to connect developers and companies for face-to-face conversations. If you're interested in latency and performance, we're hosting a web-event Tuesday and would love to have you participate: Hirelite.com.

I coded all day yesterday, focusing on a new feature for Hirelite: Speed Dating for Software Jobs. I implemented ~300 lines, added tons of functionality, and was feeling crazy productive. After this surge of productivity, though, the familiar pangs of technical debt anxiety came. I prescribed a hearty dose of refactoring.

I looked for commonalities to refactor among the eight paths of the feature. In particular, two paths stuck out as having more in common with the rest of the feature than the other six. Coincidentally, these two paths were far more commonly used than the others. Additionally, these two paths could effectively cover the scenarios implied by the other six.

Interestingly, these commonalities emerged through implementation. As I implemented the eight paths, the data structures backing each path generally fell in to two buckets. Each of the two more commonly used paths took advantage of different data structures, allowing the less common paths to be covered by one or a combination of the two.

These things weren't obvious from the user experience or product perspective. It took a deep, code-level understanding of the details of the problem and multiple potential solutions to surface this insight.

By the time I actually finished the feature, it was down to 50 lines. Though the simplified solution now seems obvious, I'm sure I would not have found this shortcut without working through the initial cases and writing the 300 lines. 

It always astounds me how much work it takes to understand something well enough to truly simplify it. Sometimes I feel bad, like I've wasted a bunch of time building something that should have been easy from the start. Or maybe I shouldn't have done it at all - I should have hired a rockstarguru who doesn't need to write rough drafts of anything, ever. And then I remember those people don't actually exist.

The same seems to be true for art. Picasso said:

... pictures went forward toward completion by stages. Every day brought something new. A picture used to be a sum of additions. In my case a picture is a sum of destructions. I do a picture - then I destroy it. In the end, though, nothing is lost: the red I took away from one place turns up somewhere else.

Picasso simplified masterfully. In 45 days, he drew 11 bulls, eventually boiling a bull down to just 9 lines.

Despite all this, I still have this gut feeling that I should have known.

• Do you run in to situations like this?
• Any tips on how to shortcut early iterations that take so much time?
• Is this the type of thing that more experience helps you get around or is it purely a creative process?

After we launch the feature, I'll try to follow up with a deeper dive in to the specifics of the implementation.

One of the main reasons all of development isn't outsourced is that it's hard to replace an in-house developer's personal investment in the business, cultural understanding, and market understanding. Additionally, communication is typically orders of magnitude more efficient.

Since starting Hirelite: Speed Dating for Software Jobs, I've been taking a lot of meetings around New York City. After each meeting, I like to find a place nearby to sit, plug in my laptop, and hack for a few hours. In New York, it's hard to find a place you can sit with some elbow room unless you know where to look (Starbucks is usually packed). It turns out, due to zoning regulations, NYC abounds with free public space, often indoors and often with open power outlets where they don't mind you sitting for a few hours.

[Base height illustrated source]

Example

Why is this?

[The Equitable Building source]

Early attempts to regulate building height, base height, and setbacks (how high building walls may be before they must be set back in relation to the building footprint), including the 1916 Zoning Resolutions, were too restrictive and were plagued with other problems inherent to a growing city.

I'm embarrassed to admit that lately, I've been watching a lot of HGTV. Whenever I have programming to do that is easy but time consuming, I watch House Hunters or Designed to Sell in the background. On both shows, buyers and sellers use comparable houses to help them put a value on a house. Similarly, investors value companies based largely on how the company compares to other companies they've seen before. Wouldn't it be nice if there was some way for founders to see the valuations of other startups too?

Finding and Valuing Startup Comparables

You can start valuing your company more like an investor by seeking out comparable startups and working backward from the funding they've raised.

1. Go to the funding section of CrunchBase and look for companies who have received funding recently that are similar to you: similar stage (angel, seed, series X), similar progress (demo, launch, 10k users, paying customers), similar team size/composition.
2. Take the funding amount, divide it by 0.2 and 0.3 to get a range for the post-money valuation. 0.2 and 0.3 correspond to how much dilution you expect from the funding. Example: GroupMe raised $850k. The post-money (high end of the range) is $850k / 0.2 = $4.25m.
3. Subtract the funding amount from each side of the range to get the pre-money valuation. Example: GroupMe's pre-money (high end of the range) is $4.25m - $850k = $3.4m.

Here are a few example pre-money valuations of companies on CrunchBase who have received seed or angel funding within a month. Ideally, you'd look closer at the composition of the team and the progress of company; however, I've only included the team size and how long the company has been operating here.

Through Hirelite, cofounders often ask me how much equity a technical cofounder should get. The graphic below balances the risks cofounders take with their relative contributions to help answer this question. All assumptions and clarifications are noted after the graphic.

This covers one of the most common situations I encounter: For a pre-funding web startup whose team includes only a non-technical cofounder, how much equity should an incoming technical cofounder get?

Technical cofounders, remember that the number of shares or options doesn't matter, just your percentage ownership (what the chart shows).

Assumptions

For this case, I assume the non-technical cofounder has already contributed significantly to the business and will likely get more equity (in this chart their minimum ownership is 50%). This doesn't have to be the case. Technically inclined people can definitely build something on their own, and then seek a non-technical cofounder, retaining more equity for themselves. Even if the technical cofounder doesn't prototype something first, they can still contribute more to the business and receive more equity (ex: someone who did marketing for 2 years out of undergrad manages to recruit a top 10th percentile programmer with a successful app in the iPhone app store).

I assume both the non-technical and the technical cofounders are compensated equally. I also assume they are either working on their company full time now or will be soon. Compensation is probably between $0 and a ramen salary. This is an area that varies widely and can significantly impact how much equity a technical cofounder receives (if they take more salary). It's beyond the scope of this post, but I'll try to cover it in the future.

I assume the technical cofounder has a reasonably general set of technical skills and can pickup new technical skills quickly. Ex: they're primarily a back-end programmer, who can code basic front-end specs, and who can administer cloud hosting.

I assume both cofounders will be diluted equally as more employees and investors get involved.

Finally, I assume this chart may be a little surprising to non-technical cofounders. Why should a technical cofounder get so much of the company? Especially a company based on your vision. Basically, it's because ideas aren't worth much. Execution is what matters, and in most web startups that falls on the technical founder. Software developers are a hot commodity right now, and many of them know it, so they're on the lookout for really stellar business people to partner with, not some run-of-the-mill, me-too idea having, 10k foot synergizer without any concrete sales prospects, marketing ability, or product experience. Sorry for the rant, but in all seriousness, I've seen too many good non-technical founders delay building a product for months because they're too stingy with equity.

Other Reading

Here are a few other approaches by others in the venture and tech community. There are no hard rules for this kind of stuff, so I'd recommend reading them and synthesizing with this:

• How To Allocate Founder and Employee Equity by Fred Wilson
• Fair Equity Allocation by Joel Spolsky
• The only wrong answer is 50/50: Calculating the co-founder equity split by Dan Shapiro (in response to Joel Spolsky's post)

Notes

[1] Technical cofounder starts with 50%: start by assuming the technical and non-technical cofounders will contribute similarly to the business and are taking similar risks for the business. Give the non-technical cofounder extra equity for anything "above and beyond" (see final assumption above for more). Also, here's an example calculation: 50 (base equity) - 10 (for working prototype) - 5 (has over 10k users) - 10 (has raised VC) = 25. The technical cofounder gets 25% of the company.

[2] Working prototype (not just wireframes) -10%: If a non-technical cofounder has a working prototype, they've likely assumed some risk already to build the prototype (perhaps by contracting it out). Creating wireframes doesn't require much risk taking or even really help de-risk much of the busines.

[2a] Has paying customers -10%: If the company already has paying customers, the non-technical cofounder has already eliminated a huge risk for the business; someone wants the product.

[2b] Has over 10k users -5%: If the company doesn't have paying customers, but does have some reasonable user traction (with an nice trajectory), the non-technical cofounder has de-risked the business some, but not as much as having paying customers.

[3] Non-technical cofounder has significant connections or experience -10%: Connections include: relationships with key people in the company's target market, social network connections, blog readership, etc. Domain specific experience within the target market are helpful as are more broad experiences in sales, marketing, product development, or business development (probably on the order of 5+ years outside of undergrad, grad, or MBA school).

[4] Non-technical cofounder has raised venture capital before -10%: The non-technical cofounder has done a startup before, and someone has trusted them with a lot of money. It will probably be easier to get money again. If your business doesn't need external capital, tone this number down, but it's still important (maybe -5%) because it conveys a degree of startup savvy.

[5] Non-technical cofounder has had a successful exit before -10%: The biggest predictor of future success is past performance. This could also apply to non-exit situations where the non-technical cofounder started a company that is operating successfully and the non-technical cofounder has chosen to move on, etc.

[6a] Salary upon funding -0%: No extra equity for getting a salary upon funding. It may not be a market salary, but a technical founder will likely get something if they elect to. Why should a technical cofounder take less equity upfront because they may or may not receive a salary later? In this situation a technical cofounder and a non-technical cofounder are taking similar risks; therefore, neither get any extra equity.

[6b] Non-technical cofounder has idea and vision -0%: This is just part of the job description and included in the non-technical cofounder's initial 50%. Execution is what matters anyway (see final assumption above for more).

[6c] Non-technical cofounder has MBA -0%: (this is my opinion) In the early days of a startup, an MBA doesn't help much. It certainly doesn't hurt, but it shouldn't affect equity.

[7] Non-technical cofounder invests money -5% per $10k: This is an approximation based on a very early stage valuation of $200k. The situation will vary from company to company based on who is involved with the company and what they've accomplished so far. When I think about this, I only include money still unspent when the technical cofounder joins (usually the money was used to build the initial prototype. Sometimes I adjust the equity for the "working prototype" step above).

[8] Lower limit 10%: A non-technical cofounder wants to ensure the technical cofounder has compelling incentives. I consider this the minimum if the technical cofounder is not taking a salary.

In both Git and Mercurial, there's a very easy way to browse your local repository in a web browser that I didn't know about until recently. You should really try these right now. They've already made a huge difference for me.

Browse your local Git repository

git instaweb --httpd webrick # view at http://localhost:1234

If you prefer not to use webrick, you could install lighttpd and then just run git instaweb.

Browse your local Mercurial repository

hg serve # view at http://localhost:8000

Both of these tips came from this forrst post and discussion.

It's hard to determine if a tech job board is worth watching (if you're a job seeker) or worth posting to (if you're hiring), so I made a quick visual comparison of job boards in New York City.

What each metric means

• Cost - the cost of a single post. The life time of the post varies per site.
• Headhunter posts - the number of posts originating from recruitment agencies as opposed to from companies that are hiring.
• Typical company sizes - an estimate of the typical size of companies posting to a job board based on the company name, funding stage, salary/equity balance, and other information contained in the post.
• 20 most frequent words - the words most often used in job posts at a particular job board. Technical note: I used Lucene (and the StandardAnalyzer) to help with text processing and frequency calculations, so very common words (a, the, ...) are excluded. Additionally, some special characters were omitted from words (see observations for effects).

Observations

First, I noticed that "c" appeared much more than I expected. Companies can't be requesting C skills enough to put it in the 20 most frequent words used Craigslist and Stack Overflow! Well... maybe Stack Overflow. It turns out, the tool I used process the text (Lucene) cut out special characters, normalizing C++, C#, Objective-C, and C all to C, thus inflating the frequency of "c".

These tricks have saved me a lot of time. Many of them I started using after reading this Definitive Guide to Bash Command Line History by fellow Hacker News reader, pkrumins. It includes a much deeper look at history than the quick examples I cover here. 

Search Your History Quickly

No more history | grep ... or hitting the up button 20 times. Just open a command prompt, press Ctrl-R, and begin typing a word in your command. As you type, the most recent command matching what you type will appear. To continue searching backwards in the history, hit Ctrl-R again. Then hit the left or right key to edit the command or hit enter to run it. 

Increase Your History Size

Once you know how to search your history, make sure your commands stick around for a while. By default, the history size is pretty low, usually only 500. To increase your history size, add the following to either ~/.bashrc, ~/.bash_profile, or /etc/profile:

export HISTFILESIZE=1000000000
export HISTSIZE=1000000

Analyze Your History

Once you've built up a sizable history, analyze it to determine possible aliases that will reduce typing time. To see the top 30 most used commands, run:

cut -f1 -d" " .bash_history | sort | uniq -c | sort -nr | head -n 30

To see the top 30 most used commands including arguments, run:

sort .bash_history | uniq -c | sort -nr | head -n 30

I used to get so mad when I forgot to sudo a command, especially a long one. Not any more. To repeat a command using sudo, run:

sudo !!

Reuse Arguments

Say you want to backup a file then edit that file. Here's how you can reuse an argument from the most recent command:

cp a-very-long-file-name.txt a-very-long-file-name.txt.bak
vi !^

Also, you can use !!:N for the Nth argument, !!:N-M for the Nth to Mth arguments,  !!:$ for the last argument, or !!:* for all arguments.

Self Promotion

If you're a software engineer with Linux experience in NYC, consider coming to Hirelite: Speed Dating for the Hiring Process tomorrow evening (4/27).

I realized I made my first Delicious bookmark just over four years ago. Here's a visualization of my links. It's a decent approximation of what I know.

Update: I switched to pinboard.in after all the Delicious shutdown controversy.

Who knew threads and romantic relationships had so much in common? For those of you new to threads, threads (and processes) allow computers to seemingly do multiple things at once, where each thing is a separate "thread" of execution. On a computer with a single processor, the processor spends short amounts of time executing each thread before switching to another thread to execute. On a multiprocessor system, processors execute threads simultaneously, switching between threads when there are more threads than processors.

Romantic relationship example: A couple tries to decide whether or not to get a pet. The argument gets heated. They keep bringing up unrelated topics. Each time a new topic comes up, they spend five minutes on it.

Romantic relationship example: A couple is getting dressed for a party. The man is dressed and ready to go. The woman is nowhere near done. The man keeps interrupting the woman to ask her if she's ready yet. The man is busy waiting.

Since starting Hirelite, where we get companies and software people talking directly, I've heard a lot of horror stories about working with recruiting agencies. When I hear these stories, I can't help but think of black hat vs. white hat hacking and SEO, so I call recruiters who engage in unethical practices "black hat recruiters". Black hat recruiters resort to the tactics below because they're too lazy to confront the real challenges involved in finding and matching good people and good companies. Please note that not all recruiters are bad, and some provide a lot of value, but this post is not about those recruiters. This post is about black hat recruiting where tactics range from lies to ethically gray practices to illegal activity (in approximate order of how common they are):

On Tuesday, I ran Hirelite's first event: Speed Dating for the Hiring Process for software engineers. Read about how it went here: Results of a Speed Dating Event for Hiring Software Engineers.

There are so many NoSQL systems these days that it's hard to get a quick overview of the major trade-offs involved when evaluating relational and non-relational systems in non-single-server environments. I've developed this visual primer with quite a lot of help (see credits at the end), and it's still a work in progress, so let me know if you see anything misplaced or missing, and I'll fix it.

Note: RDBMSs (MySQL, Postgres, etc) are only featured here for comparison purposes. Also, some of these systems can vary their features by configuration (I use the default configuration here, but will try to delve into others later).

• Consistency means that each client always has the same view of the data.
• Availability means that all clients can always read and write.
• Partition tolerance means that the system works well across physical network partitions.

According to the CAP Theorem, you can only pick two. So how does this all relate to NoSQL systems?

• Relational systems are the databases we've been using for a while now. RDBMSs and systems that support ACIDity and joins are considered relational.
• Key-value systems basically support get, put, and delete operations based on a primary key.
• Column-oriented systems still use tables but have no joins (joins must be handled within your application). Obviously, they store data by column as opposed to traditional row-oriented databases. This makes aggregations much easier.
• Document-oriented systems store structured "documents" such as JSON or XML but have no joins (joins must be handled within your application). It's very easy to map data from object-oriented software to these systems.

Now for the particulars of each CAP configuration and the systems that use each configuration:

• Traditional RDBMSs like Postgres, MySQL, etc (relational)
• Vertica (column-oriented)
• Aster Data (relational)
• Greenplum (relational)

Consistent, Partition-Tolerant (CP) Systems have trouble with availability while keeping data consistent across partitioned nodes. Examples of CP systems include:

• BigTable (column-oriented/tabular)
• Hypertable (column-oriented/tabular)
• HBase (column-oriented/tabular)
• MongoDB (document-oriented)
• Terrastore (document-oriented)
• Redis (key-value)
• Scalaris (key-value)
• MemcacheDB (key-value)
• Berkeley DB (key-value)

Available, Partition-Tolerant (AP) Systems achieve "eventual consistency" through replication and verification. Examples of AP systems include:

• Dynamo (key-value)
• Voldemort (key-value)
• Tokyo Cabinet (key-value)
• KAI (key-value)
• Cassandra (column-oriented/tabular)
• CouchDB (document-oriented)
• SimpleDB (document-oriented)
• Riak (document-oriented)

Self promotion and Credits

• If you're a developer and looking for a job or if you're hiring developers and these data systems are important to you, consider coming to Hirelite: Speed Dating for the Hiring Process on Tuesday.
• This guide draws heavily from a recent Ruby meetup (by Matthew Jording and Michael Bryzek) and a recent MongoDB presentation (given by Dwight Merriman).
• Thanks to DBNess and ansonism for their help with validating system categorizations.
• Thanks to those who helped shape the post after it was written: Stan, Dwight, and others who commented here and on this Hacker News thread.

Update: Here's a print version of the Visual Guide To NoSQL Systems if you need one quickly (warning: it's not all that pretty and I may not keep it updated, but as of 3/17/2010, it's current).

Networking can really suck for introverts. I know because I'm one of them. You're probably thinking, "Of course! Introverts are shy and have trouble with social interactions." However, introversion is much more complex and encompasses an overlapping spectrum of feelings. Here's my take on it:

About once a year, I accidentally type my password into the user name field on a login form and click submit. Immediately, I realize what I did as I see my password sitting there, mocking me in plain text from the user name field. I did that today, and it's time to go reset my passwords. For those of you who aren't involved in web development, let me explain why.

Do you know the high pitched sound you hear when a TV is on but muted? Imagine you heard that noise constantly... and really loudly. That is tinnitus: chronic ringing in the ears. Currently, 12 million Americans suffer enough from their tinnitus to require medical attention, and 2 million patients report being unable to function "normally" day-to-day. Tinnitus can even be so extreme that it drives some sufferers to suicide.

Setting up an email server and making it work with anti-spam protocols is one of those things that's 100x easier to do the 2nd time around. I follow these steps whenever I need to do it.

• Apache
• Ubuntu 9.04 (these instructions should work for other versions, but that's what my target currently is)

Ingredients

• postfix
• dkim-filter - DKIM (DomainKeys Identified Mail - new standard)
  
• dk-filter - DomainKeys (legacy protocol - needed for Yahoo, etc)
  
• SPF records

Postfix Setup

sudo aptitude install postfix

and configuration of myhostname, mydomain, and mynetworks (which determines which hosts may send mail through the server). Here's the section from a sample configuration:

myhostname = yourdomain.comalias_maps = hash:/etc/aliasesalias_database = hash:/etc/aliasesmyorigin = /etc/mailnamemydestination = yourdomain.com, localhostrelayhost =mynetworks = 127.0.0.0/8 [::ffff:127.0.0.0]/104 [::1]/128mailbox_size_limit = 0recipient_delimiter = + inet_interfaces = allmydomain = yourdomain.com

DKIM setup and configuration for Postfix and your DNS

Next we need to install and configure DKIM. This will make the server sign email headers so that users can assert, through DNS, that an email's origin has not been forged. It's safe to directly follow the instructions here: DKIM setup and configuration for Postfix and your DNS. The only catch may be with your DNS; as some DNS's (particularly GoDaddy) will not display the fully qualified TXT record name. If you enter "mail._domainkey.mydomain.com" your DNS may show the record name as "mail._domainkey". If it does, don't worry about it. It's just following an internal convention. When the DNS server receives queries for "mail._domainkey.mydomain.com", it will respond correctly.

milter_default_action = acceptmilter_protocol = 2smtpd_milters = inet:localhost:8891,inet:localhost:8892non_smtpd_milters = inet:localhost:8891,inet:localhost:8892

SPF setup and configuration for Postfix and your DNS

SPF (Sender Policy Framework) allows mail recipients to reject mail received by senders who DNS does not recognize as authorized senders for a domain. It's safe to directly follow the instructions here: SPF record setup and configuration for Postfix and your DNS, though you may need to create the "smtpd_recipient_restrictions" section in your postfix configuration.

• compatibility with how I write - The easiest way for me to sync post writing across my desktop, laptop, and phone is to save drafts in gmail. It seems like the Posterous team had this specific use case in mind for their product.
  
• orientation toward longer-form posts - I need a place to ramble, and to me, Posterous has a look and feel that lends itself to longer posts.
  
• focus on content creation - Tumblr seems like it's based on sharing other content, which is fine, but not what I'm going for with this blog.