Startup Mullings

Entrepreneurship is not a job

Sat, 01 Dec 2012 15:33:18 -0500

My thoughts posted on the personal blog

A Week in the Startup Life (Week 10)

Tue, 21 Aug 2012 17:42:16 -0400

Note: The content here is from May but I’m trying to catch up soon. The introductions are retrospective but the daily notes are kept in the present tense. Hopefully it’s not too confusing.

The first lesson this week was that using Open Source libraries may start off being very helpful but as your product gets more complicated you will most likely have to customize it to do what you want or create your own replacement. We used an Infinite Scroll library that we had to modify and we’ve also had to modify the an image gallery plugin to do what we want. It’s definitely a good way to get started and there’s no point in reinventing the wheel but be aware that as your product grows the tools and libraries you use will need to be updated. The second lesson was that it’s amazingly useful having a coding cofounder and not because two can write more code than one. It’s great having a second perspective on the technical details as well as someone to call me out for being lazy and taking shortcuts. It’s also difficult to QA your own code so having someone else look at it is helpful. In general, it’s great to have overlapping set of skills between the cofounders and a goal for me is to improve my UX skills.

Here’s week ten:

5/21 - Added more events to be tracked in MixPanel and cleaned up the timezone issues for the various social networks we were pulling from. Had a chat with MaxPanel about the best way to track different events and got some useful feedback.
5/22 - Gave users ability to delete their own moments. Started caching commonly used templates and HTML.
5/23 - Spent some time working on financials around the stock options I was granted in the prior job. The goal was to discover how much I’d need to pay to exercise them as well estimate the tax liability.
5/24 - We create our own plugin to have Facebook albums load in Galleria rather than rely on the one provided by Galleria. We need to rewrite this one of these days.
5/25 - Fixed some bugs that Sandy discovered around liking/reposting of moments and Facebook albums. Implemented cascading of likes/reposts as well as dealing with the privacy issues that arose from reposting of moments.
5/26 and 5/27 - Took the weekend off

Eating Yourself: Innovation through Cannibalization

Sun, 12 Aug 2012 22:06:06 -0400

I was rereading the HBR paper on Strategies for Two Sided Markets and came across a passage describing Apple’s mistake of trying to monetize both sides of their market, the consumers and the developers, rather than focusing on one like Microsoft did by giving away the SDK for free.

It got me thinking about Apple’s recovery. Many people credit the iPod with revitalizing Apple but I think there’s more than that. I suspect the bigger reason was the decline of desktop software and the ability to be productive on the web. Suddenly the network effects that existed by having software that only worked on Windows no longer existed. Software started migrating to the web and people were more willing to try new operating systems out. In 2006, I switched to Linux without too much trouble. It was also simple to find help online to deal with the various issues I ran into which made the transition easier. In some ways, Google helped Apple recover by speeding up the move to the web with a more accurate search and a good set of productivity apps.

In general, it’s damn difficult to overcome network effects. Google will not be replaced by a search engine. Facebook will not be replaced by a social network. These network effects will be broken by a behavioral change. Instagram rode this wave of behavioral change of the move to mobile and it was a savvy move for Facebook to make the acquisition. It makes you wonder what Instagram could have become had it stayed independent.

Innovation is cannibalization. By pushing the envelope of technology, pioneering companies cause behavioral changes that will give rise to companies that may end up replacing them. As Clay Christensen notes, it’s rare for a mature company to put resources behind a disruptive technology that will cannibalize itself but it’s the only way to stay relevant. Only 13% of the companies in 1955’s Fortune 500 made the list in 2011. It’s amazing to see how quickly things change and the pace is only getting quicker.

In Defense of Yahoo

Sun, 05 Aug 2012 23:54:12 -0400

Reading recent tech coverage makes you think that each newly startup is more valuable than Yahoo. Yahoo is the 4th most visited site in the world with over 300 million users on Yahoo mail. This is a problem every startup should hope to have.

User acquisition is the most difficult task for a consumer startup. User attrition is an easier problem to solve than user acquisition. Yahoo doesn’t need to build a product that’s 10 times better than the competition, they just need to simplify and improve what they already have. Yahoo also has massive usage among the mass market with millions of people having Yahoo as their home page. These are not the same people that sign up for every startup featured on TechCrunch. Yahoo has challenges but worrying about user acquisition is not one of them. Yahoo will need to develop a vision and relentlessly pursue it. The culture will need to change and vested interests will need to be broken.

It’s easy to criticize Yahoo for ignoring Google and Facebook but impossible to say what Yahoo should be doing now. I look forward to seeing what happens to Yahoo with Marissa Mayer at the helm.

The Startup Advantage: Details, Details, Details

Sat, 28 Jul 2012 15:25:24 -0400

A frustration I’ve been experiencing more and more is having to reload a webpage in order to change the date range in the options. If a company expects me to keep a site open for more than a day they should make it easy for me to update the options. The big example is Google Analytics - I open up a page, choose a date range, and get to see my charts. If I keep the tab open and want to want to run the same analysis the next day, I’m forced to reload the page to even be able to include today in the date range. It’s an unnecessary action for the user and it would be easy to correct this behavior with some simple Javascript.

Such small details don’t matter individually but together they reflect a lack of empathy for the user that impacts a company culture. We should always be striving to make a user’s experience better and doubly so whenever it’s actually an easy fix. Other easily fixable examples I’ve seen are clearing entire forms when there’s an error with one field and not highlighting the field that’s giving the error.

I suspect the reason these aren’t fixed is a managerial problem. The application works and there’s no reason to go back when there are all sorts of new shiny things that can be built. No one wants to do a cost vs value analysis for these minor fixes so they stay the way they are. I suppose you need to either build things the right way immediately, fix it without letting anyone know, or resign to leaving it alone.

There’s a reason startups tend to have better products. They don’t go through analyses to determine whether to make minor changes, all it takes is for someone to decide that something needs to be fixed and the next deployment, probably within a few hours, will have it solved. Combined with the massive sense of ownership that comes with working at a startup, that’s a lot of improvements that would be done at a startup but not a larger company.

Race to 0: RIM vs IE

Tue, 10 Jul 2012 21:49:00 -0400

I read an article earlier today about how companies are preparing for a possible demise of RIM and couldn’t help but compare RIM’s decline over only a few years compared to how long it’s taking IE to disappear.

To confirm that there is in fact a difference in behavior, we can compare the RIM share among smartphones and IE share among browsers. Turns out that they are noticeably different: IE is on a linear decline with close to 70% in Q3 2008 but around 36% in Q1 2012 while RIM starts at 16% in Q3 2008, goes up to a high of 21% in 2009 and then drops to 7% in Q1 2012. Plotting their % decline since the data starting point highlights this further. If we calculate the average decline per quarter from their highest levels and try to see how long it will take to hit 0% share, IE will take almost 4 years while RIM will take less than 5 quarters.

Why are they so different? If they’re both in the enterprise why don’t we see a similar decline in both? I was able to think of a few reasons but would love to hear what others think.

RIM’s competition has been much stronger - both Apple and Android have been eating up the share at a massive rate while the browser market has been relatively stable. This is compounded by smartphones being a new, quickly evolving industry where people are upgrading phones as frequently as they can.
Guy Kawasaki says that companies should focus on making their product 10 times better than the existing competition in order to get adoption. This may be a lot easier to accomplish with smartphones than with browsers.
Browsers are an older industry and there’s no point in even doing this comparison. We should do this analysis when the smartphone market is more mature and we can normalize the two time frames.

I tried digging in a bit further but it’s unfortunate how difficult it is to find browser market share data. I’d love to dive in and look at the trend in the browser market since the 1990s and see how that compares to the trend in smartphones. If anyone has this data please let me know.

Here’s the Google spreadsheet if you want to play around with the data.

Need for Speed - A Week in the Startup Life (Week 9)

Mon, 09 Jul 2012 22:10:13 -0400

Note: The content here is from ~2 months ago but I’m hoping to catch up soon. I’m going to start writing the introductions retrospectively but keep the daily notes in the present tense. Hopefully it’s not too confusing but let me know if is.

This was another productive week getting our new profile page ready. Our goal was to make Glossi’s profile pages look better by adding a larger image and an improved layout. We also wanted to increase the speed as much as we could. An approach we ended up implementing is to first load the user’s profile information and then load the user’s social media content asynchronously. This way the top of the page loads immediately and the content appears when it’s ready. Because the page is broken into various components it also became easier to cache certain fragments and further improve the site speed. One issue with this approach is that the Google bot doesn’t crawl the asynchronous content so we’re working on a fix.

Here’s week nine:

5/14 - Made a ton of improvements to Glossi’s infinite scroll to prepare for the new profile page. I’m much more comfortable with Javascript now so I made some changes to the library and submitted a fork request on github. Our goal was to make the infinite scroll library load from a different url based on user actions on the page which the existing library didn’t support.
5/15 - Finally got multiple queues working using Celery and RabbitMQ. This allows us to have high priority and regular tasks treated differently. This also leads to site speed improvement since we can do certain things asynchronously that we weren’t comfortable with doing before since we weren’t confident that they would get done in time. The reason I didn’t do this earlier was because I made code changes in quite a few places and I wasn’t confident in being able to test it properly. After developing a testing approach I became more comfortable with getting it ready for deployment.
5/16 - Spent the day working on the new Glossi profile layout which should launch in a few weeks. In the evening, I caught up with a friend who’s working on a startup. We had a pretty interesting discussion about focusing while having a grand vision and how to choose an initial market to target.
5/17 - Made some performance improvements to the newly improved infinite scroll by tracking prior calls and only calling when there’s a chance that something will be retrieved. Met up with an old college friend of mine who’s working on a few startup ideas. Contrasting with his first startup (that was successfully sold), he’s focused on building a company that can generate revenue from day one. I wonder if this is a trend
5/18 - Fixed up minor bugs with the infinite scroll, multiple queues, and went through server logs and fixed a few issues uncovered by the Google crawler having to do with old cached pages.
5/19 - My wife and I held a Miss Representation screening and had a good discussion about women’s portrayal in the media. Watching it also gave me the idea to scrape some data from IMDB and look at the trend of actor vs actress ages in movies. A light Glossi day.
5/20 - Started incorporating Mixpanel into Glossi - amazingly well designed product that’s simple to implement and fun to use. I still have a few questions about tracking certain events but I’ll try to figure them out over the next few days.

Be a Generalist - A Week in the Startup Life (Week 8)

Sun, 08 Jul 2012 16:32:00 -0400

Edit: This is written in the present tense but taking place around 2 months ago. I’m working on catching up with these updates this week.

Things started to accelerate two months into the startup life. There are two of us working on Glossi full time now and we’ve been meeting nearly every day to work together. Amazing how much more you get done when you’re working with someone compared to working alone. I’m starting to appreciate the need to have generalists within a startup, both on the tech and business side. It’s impossible to predict what you’ll need to get done each day and being ready for as much as possible is the best you can do. Luckily for us, there are so many resources available right now that all you need is curiosity and a willingness to learn to quickly ramp up. I can’t imagine doing what I’m doing now without Wikipedia or StackOverflow.

Here’s week eight:

5/7 - Caught up on a bunch of items after returning from Austin. Luckily there were no emergencies that had to be resolved while I was away. Went back to work and started working on the new Glossi profile view by hooking up a bunch of UX things that Marc created to the backend.
⅝ - Nearly done with the “contact me” feature for Glossi but as usual, the devil is in the details. Made sure edge cases worked and also spent some time making sure the the usability was there. For example, focusing on the proper fields by default as well as animating the popup when it’s opened and closed. Improved logic around when we should repull photo albums from Facebook. Quiet day on the meeting/business side.
5/9 - Added a few metric tracking features for our premium clients and spent the day working out minor kinks by looking at user profiles. We also started working on some admin tools to help us monitor and fix issues with broken layouts.
5/10 - Sys admin day of the week - spent the day working on various caching and compression issues. This was driven by our transition to Cloudfront as a CDN as well as our goal to get rid of the various CSS/JS caching issues that prevented us from quickly releasing code. We also ended up running a small instance of memcached to improve performance.
5/11 - Another solid day of feature work. We polished up the actions people could take on moments (Like, Repost, Delete) and are getting it ready for release. I also went back to our old Javascript code and ended up improving it a ton based on what I know now. Big improvements were refactoring a bunch of old methods to use JQuery’s delegate function and the various event handlers.
5/12 - We had the final round interview with ERA in the early afternoon that went well. We grabbed a quick team beer after in Bryant Park and then I headed home to celebrate my mom’s birthday.
5/13 - Didn’t do a bunch of work but ended up catching up on email and read a bunch of saved articles throughout the week.

Achieving Browser Autocomplete

Thu, 07 Jun 2012 19:09:00 -0400

Over the past few days, I’ve been thinking about habits. How do they form? How do they change? And the selfish one - how can you build a product that is habit forming? My cofunder sent me a great Nir & Far blog post that goes into detail about generating desire which is a great read to anyone building a consumer product.

Along these lines, I decided to be a bit introspective and see which products and sites are a part of my habit. A simple way was to type each letter of the alphabet into the Google Chrome address bar and see what site autocompletes. Here goes:

analytics.google.com
bankofamerica.com
cad-comic.com/cad
docs.google.com
eventbrite.com
facebook.com
glos.si
heroku.com
instapaper.com
joinblended.com
klout.com
linkedin.com
maps.google.com
news.ycombinator.com
optimum.com
plus.google.com
questionablecontent.net
reader.google.com
startupmullings.com
twitter.com
udacity.com
voice.google.com
wixlounge.com
xkcd.com
youtube.com
zerply.com

After excluding my sites (glos.si and startupmullings.com), we can organize them into the following categories:

Entertainment (the comic sites - xkcd, QC, CAD; Youtube; Google Reader)
Social Networks (Facebook, LinkedIn, Google Plus, Twitter)
Utilities (Google analytics/docs/voice, Bank of America, Instapaper, Eventbrite, Optimum, Heroku)
The rare letters (Zerply, Udacity, Wix Lounge). I’d like to include Klout on this list rather than admit to browsing it but I don’t know if that will be believable.

Every consumer site should strive to get to browser autocomplete status for some users rather than being semi-popular to more users. Being useful to a few passionate users and growing with their help is a much better approach than trying to immediately appeal to the mass market.

And although this exercise may be embarrassing, I’d love to see what others have as their 26 sites.

Documenting the Startup Life

Mon, 04 Jun 2012 10:23:57 -0400

Below are the weekly posts I’ve been writing to document my startup journey. I’m currently writing these on a 4 week delay but should catch up one of these days. The goal is to help me improve my writing, develop my startup thoughts, and provide a glimpse as to what startup life is about (at least to me) to those interested. I initially meant to only do a month of these but based on the feedback I’ve received I’ll continue to write these indefinitely. Let me know what you think.

A Week in the Startup Life (Week 7)

Sun, 03 Jun 2012 12:47:00 -0400

A pretty exciting week - very productive first half and then an awesome trip to Austin. It’s good to take a break every once in a while to relax a bit and get some new perspective. Travelling is a great way to get some new experiences which can help influence the decisions you make for your product. Just as it’s important to have diversity in the workplace to surface a wide range of ideas and perspectives, it’s important to also have variety in your experiences.

Here’s week seven:

4/30 - Some serious coding today. Improved our image extraction for RSS feeds and deeplinks. Improved performance of our album display code. Started working on some new Glossi social features that let you interact with other people’s moments better. Ended up meeting with the Glossi team in the morning to discuss our roadmap and met a bay-area entrepreneur who was in town for lunch.
5/1 - Had an interview with ERA in the morning about attending their program. We could have definitely done a better job explaining what we’re about and our goal but on the whole we did a pretty good job. Spent the day working with the cofounder and in the evening hosted the first NY Web Tech Scaling meetup. Now I need to find a presenter for the next meetup.
5/2 - Discussed the past week’s progress with a potential client and went over the launch plan - definitely a lot of small things that need to be taken care of before launch. Met up with the cofounder to code and ended up improving a variety of small things. I’ve definitely improved in the Javascript department and ended up simplifying a ton of code.
5/3 - 5/6 - A short trip to Austin for a wedding of a high school friend. We went with a large group of people and had a great time. I’ve never been to SXSW so this was a good introduction to Austin.

A Month in the Startup Life (Week 6)

Sun, 27 May 2012 10:50:56 -0400

So this week had some highs and lows. We were very productive with Glossi and got a lot done but my grandfather passed away after a battle with cancer. I’m going to miss his motivation and inability to remain idle. I remember him visiting our house for some sort of celebration and quickly start raking leaves - after which he’d yell at us kids to clean up the piles. I’m amazed by his work ethic and wish I can bring half as much to Glossi.

Here’s week six:

4/23 - A productive day of coding with very few distractions. Almost finished up albums and did a bunch of work improving the way we pull short content. Made some improvements to our Google+ integration by going through some debug logs and identifying some areas for improvement.
4/24 - Added thumbnails to the Glossi album displays and spent time with Git merging the various branches together. Did a bit of code refactoring. Finished up a blog post.
4/25 - Started discussing the release plan with the client as well as sharing the album display feature we’ve developed. Did a bit more work in the afternoon around refactoring our album display and started thinking about developing this type of display for Facebook albums.
4/26 - Started working on JS/CSS compression and cache invalidation. In the past we’ve run into issues with browsers caching our JS/CSS files instead of loading the new version causing a crappier user experience so we’re trying to make this a non-issue. This required us moving a bunch of JS code from our template into JS files.
4/27 - My grandfather passed away a few days ago and we had the funeral today. He had an amazing life - fought in World War II as a teenager and moved his entire family to the United States. I’ll miss him and his need to always be working.
4/28 - Pretty light coding day. We improved our deeplink code after looking at some examples of failures. We also removed a bunch of code and pages that were no longer used.
4/29 - Not a lot of coding today but set up some meetings for the following week to meet with other entrepreneurs working in a similar space. There are lots of companies with lots of different ideas trying to tackle this space.

Trend of Actor vs Actress Age Differences

Wed, 23 May 2012 23:45:00 -0400

I recently watched Miss Representation which documents how the portrayal of women in the media affects women’s roles in society. It raised many interesting points and definitely got me thinking. If you haven’t seen it already you should definitely check it out. One of the points was that there’s a huge pressure to cast female roles with young actresses whereas it doesn’t matter so much for the male. I was sure this was true but I wanted to see how big of a deal it actually was, take a coding break, and play around with some data. The goal was to replicate the results as well as provide some tools for others to do similar analyses.

I took a quick look at the IMDB site and realized that they did not have an API available. I looked at a few open source alternatives but they all seemed like overkill for what I wanted to do so I decided to just write a quick Python script to scrape the pages I needed. I started by pulling the top 50 movies for each decade (via http://www.imdb.com/chart/1910s - http://www.imdb.com/chart/2010s) and then pulling the top 5 cast members for each movie (via http://www.imdb.com/title/tt1375666/fullcredits#cast). I had to actually look at the actor/actress pages as well in order to pull the birth dates as well as the sex. After loading this data into a database it was a very simple query to run the analysis and then Google Spreadsheets to clean it up.

Not surprisingly, it turns out that over the past 11 decades, the average actor is 41 while the average actress is 32. Interestingly, during the 1980s they were almost the same but the gap has been widening since then.

I may be a bit late to the “Don’t learn to code” debate but I think this illustrates that coding is a pretty useful skill to have. It’s not about being able to develop enterprise applications but more about automating some work and being able to scratch a curiosity itch. If this data were publicly available and everyone had the tools and abilities to do these types of analyses I believe we’d be in much better shape. Maybe someone can take the work I started and leverage it to discover something new.

Note: The code to scrape IMDB is posted on github but note that it’s definitely crude and hackish at times. My goal was to get the analysis done as quickly as possible so I didn’t spend too much time refactoring.

Selling to the enterprise? Target the consumer

Tue, 22 May 2012 11:13:00 -0400

A trend I’ve been noticing more and more is enterprise sales being done bottoms up. The typical approach is to offer a free trials or have some sort of freemium product. Each sign up is then treated as an inbound lead that is assigned an account manager. Within two weeks of signing up for New Relic I was contacted by an account manager who helped answer my questions and helped me get New Relic set up for Glossi. Working with him, we were able to get a longer trial period and a discounted price for when we’re ready to upgrade. HubSpot found that inbound leads cost 61% less than outbound leads. If having a strong SEO and Social Media presence drops acquisition costs that much imagine the drop caused by having a usable product. Although we’re a small, scrappy startup that’s quick to try new products and services, I believe this approach will become the standard way of selling SAAS in the enterprise. It’s much easier to get a person to try something new and if you can turn him into a fan, you’re one step closer to getting the company signed up.

An extreme case of this would be to initially build a product that’s focused on the consumer and only building out enterprise features when there’s a clear demand for them. A great example would be Dropbox, they initially focused exclusively on making a kick-ass experience for the consumer and only after nailing that down did they release the “Dropbox for Teams” plans. I don’t recall the history of GitHub but they may have done something similar - initially focusing on public and private repositories and then growing into the more enterprise friendly plans. This is a great approach for a product driven startup since you can focus on building your product without getting stuck in the twisted path of custom client work. But when your product and team are more fleshed out, you can focus on the additional revenue opportunities created by going after the enterprise.

A Month in the Startup Life (Week 5)

Fri, 18 May 2012 17:30:00 -0400

The weeks are getting busier and busier. I definitely feel more productive and motivated when I’m sitting next the cofounders and we need to start doing this more often. I attended three NYC Tech Events and am impressed by the growth of the NY Tech Scene over the past few years. There’s so much more support now and I can’t wait to see where it goes - especially with Cornell’s new tech campus on Roosevelt Island.

Here’s the fifth week:

4/16 - Had a call with cofounder to prepare for a client call. Setup a few meetings for the rest of week. Released a new version of Glossi. Got our rejection from YC but surprisingly didn’t make a big impact on my motivation.
4/17 - Went for an early morning run. Spoke with a company looking to hire product managers and sent a few introductions. Met up with another startup cofounder and got some good ideas for Glossi. Met up with a former coworker. Spent some time working outside and finished the first “Startup Life” blog post. In the evening, I went to the NYC BigApps Award Ceremony. Proud to be living in NYC and seeing this tech startup support.
4/18 - Caught up with an old coworker in the morning. Met with cofounder and spoke with our client to go over our progress and discuss release. Spent the afternoon working with cofounder.
4/19 - Went to NY Tech Day event in the early afternoon but didn’t spend too much time there. Spent the latter part of afternoon working out of a nearby coffee shop analyzing our speed performance data and improving the slow components. In the evening went to a Entrepreneurs Roundtable meetup and heard some pretty cool pitches.
4/20 - Worked in the morning but had lunch with a potential speaker for the Scaling meetup. In the afternoon I met with a guy who’s starting an incubator and was able to practice the Glossi pitch - it can definitely be improved. In the evening, went to a classical music concert with the wife and her cousin.
4/21 - Massive day of coding. In the evening went to help setup for the Coalition 4 Queens party and ended up doing a bit of bartending. Great team with a ton of passion - maybe we’ll end up with a tech scene in every borough.
4/22 - Met up with the rest of the cofounders and started working on building out a better UI for photo albums and brainstorming other layout improvements.

Peter Thiel's CS183

Wed, 16 May 2012 16:34:36 -0400

A great blog I’ve recently started following is Blake Master’s notes from Stanford’s CS183 class being taught by Peter Thiel. Peter provides an insightful view of the tech startup world that is valuable to anyone interested in startups and entrepreneurship.

A Month in the Startup Life (Week 4)

Thu, 10 May 2012 23:26:00 -0400

This week was a big release week for Glossi with a stronger than usual technology focus. We ended up implementing the first set of backend improvements and should see a big scaling win. I’m amazed by how easy some of these changes were and wish we could have done some of them sooner. The biggest obstacle turned out to be refactoring old code to better support caching of the different page components and setting up Apache to work with MySQL on a Mac.

Here’s the fourth week:

4/9 - Deployed new version of Glossi with Facebook page integration, caching, and CDN support. Wrote a blog post on Instagram/Facebook acquisition and its effect on startups. Went for a run as a reward.
4/10 - Fixed some minor Glossi bugs that occurred with the release and also made some quick performance improvements. Met with a VC to chat about the personal branding space and got some interesting ideas.
4/11 - Discussed the Letter of Intent with our client and went over our progress. Worked on Glossi premium features. Went for a run.
4/12 - Spoke with a recruiter about some possible consulting work. Attended the ERA incubator happy hour. Wrote a blog post on the Draw Something Zynga acquisition.
4/13 - Analyzed some performance data for the Glossi site to focus on items that could be sped up. Took a much needed afternoon off and went to the NY Auto Show with a friend. In the evening, set Glossi up with a fax number (via HelloFax) in order to receive Letter of Intent. Finalized the time and place and scheduled the first web scaling meetup.
4/14 - In the morning went to the Brooklyn Botanical Gardens with the wife to look at the blooming cherry blossoms. Worked on getting Apache working on dev environment in order to test domain redirects and quickly played around with Nginx/Gunicorn.
4/15 - Went for a run. Got Apache working on the dev environment and also ended up modifying my development environment to use virutalenv. Surprisingly the migration to virtualenv was easy and I should have done it a lot earlier.

I’m still debating as to whether I should continue this weekly trend so let me know what you think.

Prior posts:

A Month in the Startup Life (Week 3)

Mon, 07 May 2012 21:55:00 -0400

This was a pretty busy week in terms of learning. After last week’s scaling mishap I spent this week trying to understand the bottlenecks in Glossi and added NewRelic to help us monitor our system. We quickly noticed that a certain component of our pages was taking too long to render. This led us to implement a caching solution that decreased page load times by 30%. The big lesson here is that it’s critical to monitor your system before trying to scale, otherwise you won’t know if you’re focusing on the right components and will end up prematurely optimizing.

Here’s the third week:

4/2 - Met up with another NYC tech guy to talk about dealing with scaling. Got some pretty interesting ideas and decided to give NewRelic a try. Caught up with an old friend in the afternoon via phone and ended up sending out a few intros. Refactored a bunch of the Facebook backend code to make the page and profile integrations use the same code paths.
4/3 - Spent morning finishing up the Facebook refactoring. Met up with a friend who’s learning to code and spent some time working with him. Started refactoring the Glossi code to make implementing a caching system possible.
4/4 - Had the weekly call with a potential client. Finished up the basic caching work that should lead to ~30% speed improvements but now need to clean the code up. Researched memcached to see how difficult it would be to implement it - seemed pretty straightforward. Went for a run.
4/5 - Worked more on getting speakers and space for the web scaling meetup group I’m starting. Met up with someone about doing a consulting project. Caught up with a former coworker who’s also doing a startup. Had lunch with two other data/ML startup guys. Had a late birthday dinner with the family.
4/6 - More caching coding and cleaning up refactoring from a few weeks back. Ran some errands before having Passover Seder with the family.
4/7 - After getting back to the apartment, worked on testing the caching solution and improved logic around job frequencies and cache expiration. Hacked around with a side project to play with Flask/Heroku/Verite Timeline. Went for a run.
4/8 - Hacked around a bit more on the side project and had to deal with all sorts of annoying unicode/ascii issues as well as the Python datetime library limitations. Did some caching testing and identified a configuration mistake on my end. Met up with cofounder to finish up UX for Facebook page integration and started work on custom themes.

I apologize for not posting this last week. I was gone over the past weekend and didn’t get a chance to finish this post up before I left. I’ll have the fourth week up later this week.

A Month in the Startup Life (Week 2)

Thu, 26 Apr 2012 13:37:31 -0400

This was a pretty hectic week for us - we got a great Glossi write up in TheNextWeb which led to a ton of visits that nearly crushed our server. After this incident, we realized we needed to understand the limitations of our system better and work on a scaling strategy. A lesson for me was that if you have a good product it’s not extraordinarily difficult to get press but it’s you need to make sure your system can handle the traffic burst. This is a good reason to use AWS, you pay a premium for the scaling flexibility but you’re able to deal with traffic spikes.

Here’s the second week:

3/26 - Got a writeup in TheNextWeb and spent the entire day dealing with scaling issues so our dev work had to be put on hold. Ended the day by grabbing a beer with a buddy to unwind.
3/27 - Met with cofounder to code and refine vision. In the afternoon meet up with another startup founder and send intros. Start writing blog post on our MySQL to MongoDB migration.
3/28 - Met with cofounder and had our weekly call with a client. Started working on Flickr integration. Met with another startup in the afternoon. Got home and almost finished up the Flickr integration. Finished up the Glossi tech blog post on our MySQL to MongoDB migration. Responded to a ton of emails in the evening.
3/29 - Got interviewed by a blogger to discuss how we recovered from our crisis on Monday. Finished Flickr integration and started working on Facebook Page integration. Went for a run.
3/30 - Met someone and got a ton of intros to scaling experts and chatted with one for an hour or so and got some good ideas. Wrote a blog post. Hacked on a side project to play around with a new timeline jquery library and learn Heroku/Flask while at it. Tried to find a venue and a time for the Skillshare classes I wanted to teach. Started setting up a meetup group but backed out at the payment step - will probably go through with it in a few days. Started filling out a questionnaire to get featured on an interview/startup/entrepreneur blog. After speaking with the team we decided we will not apply to Startup Chile this session.
3/31 - Ran some much needed errands in the morning. Went for a run. Celebrated my birthday in the afternoon. Not a very productive day but need to have one of these every once in a while.
4/1 - Worked some more on Facebook page integration. Complete the meetup group setup process and created the NY Web Scaling Meetup.

Growth of Consumer Comfort with Technology

Wed, 25 Apr 2012 13:05:00 -0400

In a previous post, I discussed the factors that allow small teams to create products that can be exposed to millions of users within a few months. In this post, I want to take a deeper look into why consumers are so much more comfortable with technology now compared to 20 years ago and try to see where this leads. Since customers are what cause our businesses to grow, we need to be cognizant of what drives their behavior in order to plan for the future. Wayne Gretzky’s father famously said “A good hockey player plays where the puck is. A great hockey player plays where the puck is going to be” and I’m hopeful that we’ll be able to see where the consumer puck is going to be.

To me, the major driver is Moore’s Law. We’ve seen computation speeds double every 18 months for the past 50 years. This has obviously led to faster computers but has also led to exponentially reducing costs. This has been a huge economic driver and is allowing computers to be more accessible than ever. Our cellphones are more powerful than what was used to land on the moon. These increases in computation also led to the rise of the modern web. It went from being a military/academic project that dealt with text data to something that’s distributing pictures and videos to whoever is interested.

More importantly, improvements in computation led to improvements in usability. Even if we had modern browser standards like CSS3 and HTML5 in the 1990s our computers would be too weak to handle them. We would not have any of the modern innovations (AJAX, DOM manipulation) and our web pages would be static without any rich media content. If we never got past the command line, how many people would have computers in their home? How many smartphones would exist? I’d argue that the usability improvements are what led to the massive consumer adoption of tech products. Of course, computation, cost, and usability are all intertwined but computation and cost alone would not have led to the consumer adoption we’ve seen.

What does this mean for the future? I see usability becoming even more native with us not realizing that we’re even using a computer. We’re already seeing this emerging with Siri and Google Glasses. As long as our computation speeds continue to improve these technologies will become better and better and will recede more and more into the background. Of course, this is all dependent on Moore’s Law holding, with many saying the pace will decrease by 2020. I’m optimistic that we’ll come up with something but even if we don’t, as long as we computing costs keep on dropping, via Koomey’s Law, we should still see the benefits as we move more and more computation to the ever cheaper cloud. It’s difficult to imagine what would happen if our computation speeds stop increasing the way they have been over the past 50 years.