Jack Dorsey's schedule

He works 16 hours a day. 8 at twitter. 8 at square.

Monday: Management meetings and “running the company” work
Tuesday: Product development
Wednesday: Marketing, communications and growth
Thursday: Developers and partnerships
Friday: The company and its culture

Sahil is doing 12 am to 5 am sleep and rest every minute working marathons.

Strong caveat on SSL usage on AppHarbor

.NET PaaS (Platform as a service) seems to have 2 major vendors at the time of writing. Microsoft Azure and AppHarbor. 

[EDIT: Please see the paragraph below. I'd like to weaken the caveats expressed here. I've had more time to analyze the situation and talk to Rune Sorenson over at Appharbor]

We were having a good time using AppHarbor until we deployed SSL on our website. Now, I am thinking of moving from Appharbor to Azure because of SSL woes on AppHarbor. This post applies to you
- If you are considering a platform to pick (Microsoft Azure or not)
- If you want to have a custom domain (like www.example.com)

By default, AppHarbor does SNI SSL (which is not supported by many mobile browsers and older browsers especially on Windows XP. I have gotten warnings with Google Chrome version 13 with alarming regularity on Windows XP).
[EDIT: Heroku does state that SNI SSL is buggy with Chrome. Basically SNI SSL is shitty. Just don't use it. Its ok for prototyping though.]

When I contacted AppHarbor support, they proposed a service which costs $100 a month for an IP based SSL. (They say its due to Amazon's EC2 pricing for an public IP). The cited model here is Heroku which also charges $100 a month for IP based SSL.

While I do like AppHarbor quite a bit, $100 per month just for SSL seems quite high to me. I exchanged a few tweets with Aaron Stannard about Azure and SSL. 

1. Azure doesn't seem to be charging extra for SSL.
2. Azure's SSL worked fine on all the browsers on Windows XP.

So $100 a month buys you one small compute instance at most cloud providers. We already use GitHub BitBucket and have an internal Jenkins build server. I am guessing we'll roll our own deploy for Azure and move on.

----------------------------------

EDIT: Actually I'd like to weaken the caveat.

I simply love using Appharbor. Even with the pricing caveat.

What is AppHarbor giving away for free:

- One shared compute instance.
- The deployment is insanely quick and frictionless. 
- We use mercurial, but if you use git, they also run your unit tests. (Not sure if they run unit tests for mercurial folks. We have our internal build server, so we don't bother).

- Rune of AppHarbor and I exchanged some emails:
     * Hostname based SSL is $50 a month on Appharbor.
     * IP based SSL is $100 a month on Appharbor.
     * Heroku has $20 a month for hostnamed based SSL and Microsoft Azure is free for hostname based SSL.
- Rune specifically mentioned prices might come down in the future. I sincerely hope they do.

We run some part of our service on Microsoft Azure as well: 
  - We used Azure for a heterogenous situation. We had to run a JAVA process for an API call to an external vendor*. Appharbor doesn't support JAVA.
  - With BizSpark, Azure hosting is free for us at this time (for our needs).

So even after competing against BizSpark free giveaways of Azure compute instances, Appharbor is still my number one choice for a frictionless deploy experience.

* External vendor is the Government of India: We are forced to use a JAVA API (IKVM.NET did not work for certain packages we had to depend on). So we couldn't be all .NET. 

Working for closed sourced companies

After hanging out at Hacker News, GitHub, twitter and other places where the hacker community gathers, I've come to the realization that the only way forward for a good hacker is to contribute to open source projects. Working for fundamentally closed source companies then is a very bad idea. I don't always mean open sourcing the core fundamental Intellectual Property which is the secret sauce of the company[1]. But the company should have a strong ethos of open sourcing anything which is auxiliary and useful to the community at large. 

I worked for Data Domain in Santa Clara which later got acquired by EMC. When we were a startup there was very little time to think about open source and community. I also was just out of college and didn't think too hard about community. I wasn't hanging out on github and hacker news. Once we got acquired, I got a lot more time on my hands [2]. What I realized was that EMC was strongly about patenting [3]. There was never any program or intent internally to open source anything. It was ok to publish papers (with patents submitted before hand) with management's approval, but our team of 100 engineers never open sourced a single thing to the best of my knowledge. We did some really cool stuff in C and some of those things could really have been contributed back to the community. But we never did. I now feel sad about my lack of imagination at the time (maybe I could have convinced management. I never did try.)

So from now on, I will try to work for companies fundamentally committed to open source. Right now, I have my own startup. We are soon going to open source a non-core but a useful project to the ASP.NET MVC community [4].

To conclude, open source aggressively. Luis (a fanatic Linux kernel contributor) tells me, that is step-1. After that comes good documentation, marketing, getting contributors who are as motivated. I'll tell you of the journey when I get there.

-------

[1] My friend, Luis works on the Linux wireless drivers for Atheros and has been ardently pushing the company to open source more of its IP. He is a staunch FOSS believer and believes that even with core IP open sourced, there is little or no threat to a company's ability to survive. I disagree with him (Cheap Chinese manufacturers could potentially eat Atheros lunch if all its IP was open source. But I digress).

[2] The place went from a magical fun place to shoot me till I am dead mind numbing. Maybe I matured, but a bunch of key smart people left.

[3] This is true for most large companies inside and outside of the valley. Friends at Cisco, Microsoft, etc report the same emphasis on patenting. As an aside, I am fundamentally opposed to software patents. I never wanted to write any patents for EMC and I never did. The Chief Architect of Data Domain and I had several long chats about software patents. Real world business examples of patent trolls made his arguments valid. 

[4] We write most of our code in C# and asp.net MVC. For the uninitiated it is the Microsoft clone of Ruby-on-rails. Being part of the asp.net MVC community has been interesting (asp.net mvc is opensource). It is not fundamentally as open as the Rails community, because of the belief in Redmond's one true way. But the open source framework so far has made me learn a lot. The aggressive use of lambda expressions and other higher level functional goodness has made me a better programmer. I am now learning Lisp to cover the gaps in my education.

Key insight about the product?

So the Individual tax filing market folks aren't passionate about a tax filing website. They interact with it only once a year for few minutes. Where is the passion? It is with the CAs, the tax practitioners.

So my new hypothesis is that we'll have to acquire a lot of invested users. Folks who have to use the software day in and day out.

We have a desktop version of the software but we'll soon deliver a SaaS version. With JQuery magic, the SaaS version can be sufficiently light to interact with. Interestingly in my conversations, convert rate of the SaaS product is high. I can basically create a reality distortion field easily.

As an aside: Can I bring the passion to Individual tax filing market. I am not sure. I will have to think hard about it. People are passionate about tax savings. So the magic may lie there!

Bill Gates never took a day off in his twenties

Clearly I have a lot of catching up to do. I hope he did go out to play and exercise.

So one thing is circumstance. Microsoft grew gangbusters in Bill's twenties. But PG also pointed out that all he did during the startup days was: work and go running. Clearly I am not on that level. Yesterday Dad told me he worked 14-15 hours days in his twenties, thirties even. Only Sunday is the day he seems to get any rest even today. So clearly I am not functioning at the same level of dedication.

Short productive bursts. Not insane long hours of inefficiency. But growth and carefully orchestrated plans. 

So here's what I am going to do. Carefully planned days. Lot of execution.

I never make new year resolutions. But having a plan is not bad at all.

My plans
  • Grow startup to reach $1 M in revenues by January 31st 2012. Need to figure out what needs to happen here, but this is the bottom line goal. $1M today is a lot less compared to Bill's time adjusted for inflation.
  • Ultimate Frisbee. Get to top 10 players level in Delhi this year by December 31st 2011. Need to run every day for that.

I'll break it down further in future post. But lets go.

More to life than IITs

Dear 17 year old,

(who did not get into IIT yesterday)

Don't be too hard on yourself. Its one exam, one day of the year. Some folks test well, some don't. All those coaching classes try to control that variable. Sometimes it works, sometimes doesn't. Also its a big country. Acceptance rates are so low that it is embarrassing. There's a chance that you'll write the JEE again and get in. Lot of people I know did that. Some succeeded, some didn't. Maybe you will never go to an IIT. There is no reason to get distraught if you never get in. You'll be just fine.

I am not presenting stats. It'll be mostly anecdotes. Folks I met in IITs weren't always the smartest. Many were average students who tested well and/or could bust their butts for long hours. The first is not an important skill in life. The second takes you surprisingly far. (Caveat: Some of the kids were quite brilliant).

So you didn't get in. What next? Find the 'best' college for you and go there. You'll find out that many around you will hate engineering after the first year. Maybe even you. Still try to learn. As much as can. Or if you aren't feeling like it, just learn as much Math as you can. Personally, I really like computer science. So I'll say maybe just learn to program really well? Go sit in those CS courses. You can audit any course you feel like. Or like Ranchord Das of 3 idiots, you can go sit there and learn. No professor is going to kick you out. There's MIT open courseware too. If you don't get into a good college, that shouldn't stop you from learning.

So I know of extremely intelligent folks who didn't get in. They are pretty successful today. In life, you'll realize success is very subjective. Its also unfortunately relative for many. But a combination of hard work, hustle, patience and ambition get you pretty far.

Don't be bitter about not getting in, ever. Don't belittle the guys who got in. Most of them got in when they were 17. It wasn't a conscious decision. Parents made them do it :)
If a potential employer/institution overlooks you just because of your degree, its their loss. Work hard though. That will give them less of an excuse. Don't self select yourself out though. Apply, call, try to get in. Hustle. Everybody likes a hustler. (Read Vinod Khosla's story of how he got into Stanford MBA school).

Good things about getting in an IIT: A sense of self belief and the "need" to make a mark. You get in, and you "believe" you are destined for bigger things. Ambition grows. So you work hard to get to them. Also there's a lot of precedent of people doing good things. So you can emulate these people. This is very useful. Here the breeding feels institutional.

What you can do: Talk to a lot of people in various places and find out their aspirations. Your ambitions will grow too. Dig deeper and figure out what needs to be done to get there. I see many kids not interacting with folks from other institutions (other IITs, arts and commerce colleges).

Also take everything with a pinch of salt. Get a lot of data points. Do what makes sense.

Takeaway from Paras' talk on customer conversion.

Paras (of visualwebsiteoptimizer.com) gave a talk at Triggr (triggr.in) and touched on the following:

There are three types of visitors on a website:

1. Buyer - She'll buy anyway. Will figure out/muddle through the UI and get things done.
2. On the fence - This is the person we need to have a conversation with and convert.
3. Browser - This person will not buy today.

This "on-the-fence" person convertion is what I'll work next on in cleartax.in

The feedback is that we need to have a way to get engagement before creating a login. Login creation scares off a lot of people. Rightly and wrongly. This has been pointed out to me by both Pratiroop (of jaamun.in) and Sarvesh (of internshala.com)


But first we are fixing up issues for the "Buyer". Minor issues remain in going from a login to efile to checkout to done flow. I am taking a crack at that right now!

India is a value market

Until it isn't. Without having seen data on either side. It might be something entrepreneurs need to crack open.

I hear this assertion from other struggling entrepreneurs, mostly anecdotal data points. But USA is a value market too in my opinion. It should be true in India too.
One key realization (a fairly obvious one) is that human intervention while prohibitively expensive in the US is infinitely cheap in India. I think this is changing due to schemes like NREGA but its a long term trend. (NREGA's future sustainability notwithstanding. Human labor is bound to get expensive considering the insane inflation).

Lets do a bit of analysis:
Selling to the US market for a clearly identified pain point has the advantage of selling to a mature market. An example is visualwebsiteoptimizer where corporations in the US whip out their credit cards and sign up. (A Value service compared to Omniture etc). Paras tells me Indian corporations want a in-person demo while he is selling to a global audience from his room over webex. So much clearer and quicker sales cycles.

I need to study and quantify Indian corporate IT + Technology spending. Mostly I am trying to figure out which markets to focus my energies on in a potential new future venture.
Since I have a couple of products aimed at the Indian customers already in hand, I need to figure out a sustainable sales and marketing strategy.

(An incomplete post. I'll ramble on to distill a few things. Using Paul Graham's mechanism of writing essays to clarify thoughts.)