HOT The word new as would be seen on a sticker on an item in store inventory trying to attract new customers. Hash


Self described as "a free online platform for assembling models of the world." More interestingly, the mission statement begins, "We believe that most bad things in the world are the product of some form of information failure; from economic collapse and the outbreaks of war and disease, to choosing the right life partner or university degree. We’re on a mission to improve imagination, and help everybody make the right decisions, overcoming information failure once and for all." Sounds cool, right?

We became interested in Hash because:

Joel Spolsky is chairman of the board (associated with creating Trello, Glitch and Stack Overflow). Don't have to be a trend genius to see where that might lead. Also, the mission statement is juicy and data + ai are cool 👏 At the time of trying Hash it has only been live for 2 months, so definitely qualifies as new - we coudn't resist.

Screenshot from hash.ai website landing page

Information Failure?

Screenshot of a poll of our instagram followers where 50% bought that info failure is the root of all problems and 50% didn't.

Of our instagram audience of 15 at the time (are you following us yet? We need you!), the results were split down the middle. Is this information failure in itself? What exactly is information failure? And does hash solve the problem? We dicuss most of that in the podcast 👇

Aparna saying - you know I hate comedy.

Our hash world was modeled after Sima the matchmaker and her singles with lists of criteria.

Here is an example of what Aparna's criteria list looked like from the show:

Aparna's list of criteria.

Turning Aparna into a javascript object was very fun, but it's not clear if that was hash fun, javascript fun or aparna fun. You can see the beginnings of our hash simulation here:

The beginnings of our hash simulation with the javascript on the left and the simulation on the right.

The Fundamentals

There are three fundamental concepts of building a model in hash - agents, behaviors and messaging.


The objects in the simulations. So in our world the agents would be the singles like Aparna and Pradyhuman and the matchmaker, Sima.


The activity of the object. So in our world, we had a matchmaking.js that was attached to Sima’s agent and a dating.js that was attached to the singles. This is where the magic would happen - the logic, the movement, the filtering through criteria.


The trickiest conceptual part of Hash - an agent can’t affect the behavior of another agent. It must a send a “message” to an agent and then the agent can read the message and change it’s behavior.

The beginnings of our hash simulation with the javascript on the left and the simulation on the right.

Tips for making a simulation:

  • The little running person is the play button.
  • Press pause as soon as you can - the simulation does not auto pause so will eat up all of your processing power.
  • Hash Cloud is newly released so you can have it run on their server instead of your computer. Let us know if you're interested and we can try to help you get on that list.
  • Don't bother emailing them for support but they have an active slack channel and you can easily sign up for 30 min office hours on Thursday with all of your questions.


Sadly, we probably created more information failure than resolution. We did not get our simulation running nor solve any world problems, though we spent 3 sessions working on it. Sigh.

Overall, if you like playing with data sets, like being an early adopter and are very comfortable with javascript or python give it a shot and let us know what you think!

Otherwise keep an eye on it but give it a minute to get more performant.

In closing, here's an email from my mom indicating we could be completely wrong:

Screenshot of a email from my mom saying she thought the internet sucked when she first saw it.