Just in the same way that Google spiders the Internet (automatically crawling through millions of URLs), there are some spiders that have yet to be made that would crawl existing exciting datasets with the hope of producing some useful utility. Here is a list of spider types that I’ve come up, with from purely informational to physical at various scales. I have attempted to put them in order of when they may be achieved. 



Internet spider- Google / Yahoo / Bing etc

Internet vertical spider - there are large range of companies that scrape/spider/record a specific vertical/horizontal dataset


Space spider - We are already collecting tons of data with telescopes from CMB data to, gravity maps and physical star / asteroid positions. Given the resolution required to capture all data in the universe, we still have long, expensive way to go here. 


Geo spiders - we use satellite imagery to collect data about our planet’s surface and now can use drones to get higher resolution/more real time information. We also collect data on the sea floors. Lidar will give us even better laser based data sets of urban / complex areas. Velodyne currently adds individual data sets to an entire cloud map from all incoming sources (e.g. Google self driving cars) to help build more accurate maps.



Product spider - there will be ramp in the 3D data we have on products / objects given the current scanning revolution that is closely tied to the 3D printing revolution


Bio spider / Gene spider - Robust, nuclear battery based soil/species sampler that captures new genetic data by crawling physically around the earth, relaying data back wirelessly. The objective here would be to find new unknown species and genes that may be useful in synthetic biology or medicine. Credit to Sumon Sadhu for coming up with the gene spider. Alternatively, a modified Bdelloidea could “steal” DNA from its targets.


Nanotech spider / materials spider - this spider would essentially be some kind of scanning electron microscope which would look for exotic and interesting surfaces/materials/objects at the nano level. The data would be useful for the nanotech industry, just in the same way that we have scanned gecko feet to inspire new sticky surfaces. 


Brain spider - there is a ton of data that is yet to be collected locked in our brains - thoughts, ideas, emotions, memories. For this we need mobile, high resolution, non invasive techniques. Major privacy issues on this spider. 


Particle spider - this spider would sit listening for particle interactions on a local level and constantly report back. Would need huge amounts of storage and IO bandwidth. 


Abstract dataset spider - this spider would seek to record all informational data sets connected to the Internet. In a sense this is the Internet itself.

Universe spider - this hypothetical spider would be able to collect all data. There are possible limits to how this machine would be able to store all information on the universe inside it and whether it would be even possible to record certain areas of universe and/or types of interactions or data. 

Time spider - this hypothetical spider would be not be limited by recording what is happening/happened already. It would be able to record data in the future and the past. 


Multiverse spider - this hypothetical spider would be able to record data across multiple universes.


Short post here. Just in the same way that you can have a center of mass/gravity of an object, you can apply this analogy to cashflow.

For example, the number of days in your net terms will be be analogistic with the distance that the mass acts and the amount of cash owed would be analogistic with the mass of that object.

For your business you may have some money coming in on net 30 terms, some on net 15, some on receipt of invoice, some in real time (e.g. on credit card) and some being prepaid. This may give rise to your average “cash in” terms to being, for example, net 20 across all the methods. At the same time, your “cash out” of the business will also have its own exit time. E.g. payroll at the end of the month or bimonthly, rent at the start of the month, supplier costs on net 30 terms, tax on net yearly terms.

When you work out your center of cash in and your center of cash out, you can them compare them to work out the “center of cashflow”. I find this useful for predicting Heyzap’s cashflow or credit turnover period without using excel models.

By mechanical/vector analogy, if you have positive cash turnover period, your business will start to spin in the right way, just in the same way that if your “center of cash out” or “burn force” (representative of your “mass”) is a smaller vector than the “cash in”. However, if you have it the wrong way round, the business will start to spin out of control. 

One of the more practical things you can do to improve your credit term over period is to get payments occurring prepaid or in realtime with tools like Stripe Connect, negotiate contracts to short net terms and to delay spending. 

It amuses me that there was a coincidental mechanical analogy behind “credit turnover period”.

P.s. sorry for not fully explaining the analogy for non mechanics people and not making a diagram.

Aerographite, is lighter than air. Yes, lighter than air. Recently discovered by German scientists in 2012, aerographite is 75x lighter than Styrofoam but the same strength. It is made up of a mesh of interwoven, linked chain of carbon tubes which are 15nm in diameter. 1kg of the material takes up around five cubic meters of space. Think about this, it is insane. I can envisage airships that don’t use hydrogen or helium but use a “solid” tank of aerographite, meaning there would be no risk of explosions from the flammable hydrogen and lower risk of bursting the blimp. Also, you would not need to refill the blimp with hydrogen/helium leading to greater long term efficiency and potential applications that don’t require landing (e.g. surface imaging).


You might ask yourself, how are you going to come back down to earth if you cannot release some of the hydrogen/helium? One could use a high capacity water condenser to condense water vapour into tanks to use as a ballast and/or use engines to guide the airship down.

Typically, a standard Goodyear blimp will has around 200,000 cubic feet of helium creating around 12,500 pounds of lift. Helium is around 7x lighter than air. Aerographite is also around 6x lighter than air. So we can say that the size of this aerographiteship would be slightly bigger than a normal airship.

The material also has an interesting property: after 95% compression of the material, it is possible for the material to return back to its original shape. This means the blimps could be packed into a deployable unit 20x times smaller for later expansion.

Another application could be high altitude wind turbines that hold their position high up without having to refill expensive helium tanks. An advantage of being high altitude would be the wind speed is fast and constant. Altaeros energies Inc should also think about this material for future builds.

Normally it takes around 20 years to get materials into mass production (see history of carbon fibre). The main question remains, when will we have aerographite in mass production?

One other limitation will be creating aerographites that can support themselves with the air inside the material “pumped” out of it (to truly achieve being lighter than air). In practice, getting a real blimp to work will probably be made out of a future version of aerographite or another aerogel that is super strong and light, one that could self support itself without collapsing in on itself if the air between the meshes of nanotubes is pumped out. A partial vacuum will have to be created inside the aerographite structure without crushing the aerogel. This may be the fatal flaw to the idea or at least the next problem to solve (i.e. given that the material is superhydrophobic, maybe a thin shell of water could be put around the aerogel or some similar exotic solution). 

Credit to Millennium Airship Inc for the image.


Just in the same way that “work in progress” aka “WIP” can put an additional burden on a manufacturing assembly line, reducing production efficiency and quality, so too can it place a major burden on software development. “Code in progress” is something that all people involved in making software should be aware of and know how to deal with.

How does it arise?

Business and product priorities are constantly changing in a startup, especially startups that are iterating quickly and adapting their products to new data or client feedback. This means roadmap priorities for small-to-large projects and tickets are also changing. (A ticket is defined as a task or element in a bug/task management CRM).  A project that was important last week can fall by the wayside because of new priorities or suddenly urgent matters. This can lead to engineers jumping to new projects before they have finished other tickets or projects.  As time moves forward, if the team doesn’t return to that project, the “work in progress”, “code in progress” or “CIP” starts to compound and significant projects/tickets can sit around uncompleted for too long.

Engineering problems

CIP is going to create issues for your engineers and engineering team at large:

  • Satisfaction: An engineer will have put hard work into the project/ticket without the satisfaction of releasing it.

  • Quality: Original specifications or communications might start to become fuzzy over time.

  • Block releases: The tickets/projects can also technically block the release of other dependent tickets and block other pull requests.

Managerial complexity

Having multiple unfinished projects/tickets when developing software can sit in the back of a product manager’s mind, clouding their judgement of what needs to be done next and increasing the complexity of load-balancing a roadmap.

Communication noise

Having CIP also increases the noise in communication:

  • Assumption of completion: people start assuming something has been completed if it was started weeks before.

  • Variability: Because of the added complexity in roadmapping, there will be a variability in expected completion date of the project, making it harder to plan QA, launching product, marketing and operational use of the project/ticket.

Production burden

Ultimately too much CIP will lead to reduced team productivity:

  • Focus: communication issues and increased complexity combine  to reduce the focus of individuals and of the team.
  • Quality/speed: Having too much CIP means a frequent reduction to the quality and speed at which you will complete tickets.
  • Impact on creativity: Having too much CIP all the time, means reduced time for alternative unplanned projects. If you always have a project to fall back into, the natural breaks in work that allow for creative time may be impacted.

Wasted Mindshare

Projects that are in progress or almost finished will constantly be on the mind of the everyone until they are released. Not only will the engineers waste mindshare but so will the product managers, management and rest of the team

Solutions to getting the optimal CIP

There are ways to get down to reasonable and optimal levels of CIP:

  • Tag out your CIP: Here at Heyzap, we tag out all the work in progress in asana.
  • Kill the ticket: If the project has been sitting around “in progress” for a very long time, you have to ask yourself - is it really that important that it keeps getting bumped back? In this particular case, you may want to kill that ticket/project.
  • Pass the ticket over to someone else: Yes, passing it off to someone else can be the best way to solve it if they are more specialised or wrote the original area of that code base and have come back from holiday.
  • Modularize projects as much as possible: Breaking down tickets and projects can allow for more frequent releases and prevent CIP.
  • Constant iteration: The length of time something will remain a work in progress is going to be proportional to the project size. 

But! A certain amount of CIP is a good thing

There are various situations where some CIP will naturally be good to have and there are no hard and fast rules as to a safe numerical threshold for the percentage of projects in CIP mode - you have to go with your gut.

  • Change in priorities: Sometimes it is essential to pause a project and work on something else due to priorities. Having a hard rule of zero CIP would be counter-productive in this case.

  • Rotation: Sometimes working between 2-3 projects/tickets can give you time to solve a technical/intellectual or people-based blocker, that is preventing the ticket going out. You might naturally solve the ticket by sleeping on it for example.

  • Waiting: Sometimes waiting can solve the project/ticket e.g. a alternative solution pops up that was a much simpler way to solve the ticket.

  • Inevitability: Some tickets being CIP are unavoidable - e.g. a script running over a few days; there is no way to prevent this from being CIP.


CIP differs from the classic work in progress that is seen in a manufacturing context. Just in the same way that the physical industries deal with WIP, it is worth giving some thought into how we can make better products by being aware of CIP. As software development moves from an art to a science, just as manufacturing moved from an artisan skill to mass industrialisation, we must learn to leverage the subtle effects that govern its mechanics.

I hear the question “what are your DAUs?” frequently bandied around. DAUs are a standard metric for comparing the popularity of social applications/games. However, interpreting DAUs as a narrowly defined metric is wrong. I want to attempt to define it in a stronger way than just “daily active uniques”. Just in the same way that Groupon have been creative on their accounts, it is possible to also be “creative” on DAU measurements. 

Daily: “Daily” is defined in a strong way; meaning 24 hours of time measurement. (Unless your app is being used on Mars, then you will want to switch this to 24h 37m plus or minus a few minutes depending on the day of the year).

Uniques (or in some cases Users): The “unique” component of the definition is currently robust: unique meaning a specific unique user (note: the “fitbit for dogs” whistle need two DAU counts: one for dogs and one for their owners).

Active: This area of the definition has the largest ambiguity. “Active action” is defined as an action that the user initiates by directly choosing to with a gesture of some sort e.g. doing a status update and pressing send or opening your app. “Passive action” is defined as the inverse of “active action”. The DAU definitions can be broken in layers of onion shells that range from the weakest interaction with a user to the strongest user driven action on your own home turf:

  • Layer 1: User made a passive action that they were not aware of, without conscious connection to your brand (e.g. tracking of user time in the app with flurry analytics, where the user doesn’t know they are being tracked). 
  • Layer 2: User made a passive action that they were aware of, without conscious connection to your brand (e.g. Company A using the foursquare API to power a passive action in company A’s app).
  • Layer 3: User made an active action that they were aware of, without conscious connection to your brand (sending an invite txt msg for an app that is powered by twillio (with respect to twillio counting this as a DAU) or authorize.net processing a user driven purchase event).
  • Layer 4: User made an passive action that they were aware of, with conscious connection to your brand (e.g. automatically logging into a website using Facebook Connect ).
  • Layer 5: User made an active action that they were aware of, with conscious connection to your brand (e.g. a user might save their score inside a game but then store it into Heyzap giving rise to a “network DAU” measurement).
  • Layer 6: User made a active action that they were aware of without conscious connection to your brand inside your app/site/content (rather than say on a network).

Layer 6 is the normal case that people report e.g. Path, however, sometimes is it fair to report layer 5 if you are distributed e.g. Disqus. Layer 4 is interesting as it borders what people will constitute as “active”.  In essence, one really needs the user to be conscious of your brand for one to use the DAU measurement as a KPI (key performance indicator). For layer 1, 2 and 3, companies normally use other types of KPIs that are not DAU centric. So next time someone tells you their DAUs or MAUs, you can ask them - how do you really define your DAUs.

Note: You can also apply this to MAU (monthly active users) as well.

Almost monthly, I’m seeing a new drone startup emerge with an exciting set of abstractions. The commercial drone market is becoming a huge area of innovation, investment and application. Cloud based software platforms for fleet management, automation, aerial imaging, APIs - you name it: the entire software stack for the commercial drone market is being built by startups right now and I want to tell you a little about it. 


Cloud platforms

Drones are going to need well executed cloud based software platforms with APIs. For drones to work at mass scale in cities, most of the functions will have to be automated, we are going to have to open up the data and make standards so that drones can talk to each other via the network.

Drone Control - In AngelPad’s latest batch. Control an entire fleet of drones from the browser. This platform can also automatically help users adhere to flight restrictions and laws (like not flying above 400ft).


Firmware platforms

Companies are emerging that want to be the Intel or Microsoft of the drone market. They want to own the firmware and software stack of the core control systems on the drone.

Airware - Y Combinator funded and recently raised a big round from A16Z. Airware (previously Unmanned Innovations) are making the firmware and hardware/firmware platform for drones. image

Commercial delivery

The prospect of sending mail via drones instead of Fedex is pretty exciting. Food delivery could cut out the car / motorbike and reduce road congestion. At least two companies are working on commercial delivery networks using UAVs. They will face practical difficulties of FAA maximum payload limits and in specific countries not being able to go into congested areas.

Matternet.us - Attempting to build a network infrastructure of UAVs that can pickup and deliver from locations. Tasks like tracking poachers in Africa, to delivering an iPhone charger cable quickly might be covered by them in the future.  I had the luck of meeting someone on the project, who told me he could get a drone delivery port built into my office in 2015. 

Skycatch - commercial delivery of objects. API into drones and set google map waypoints. This is going to be very useful for anyone wanting to have a UAV network at scale.image

Aerial imaging

Satellite imaging not cutting it? Get in close with a drone. Drones may bridge the gap for hi-resolution and controllable imaging of ground assets. Satellite imaging is expensive and cannot go inside buildings. We need the kind of resolution that can track down an illegal poacher or find a lost child.

Mavrx - This team is bringing HD imaging capabilities for commercial purposes using drones. Makes sense for many applications including security, crop/herd management and construction sites. image


There are tons of hardware suppliers emerging but I want to highlight two special companies that have hit the consumer / maker market with success:

Crazyflie - absolutely tiny, 19gram drone. Really fun and hard to control/setup. I’ve been hacking around with the one I own.  

Parrot - the AR Drone is the most successful consumer drone, an entire community has been built around these. We have one in the office - very stable. image


There have been some crazy hacks going on in the drone world:

Wireless charging - drones have limited power, this hack involves allowing a drone to charge wirelessly, so it could go back to a charging spot and hover while charging. Companies like Ubeam will increase the range of charging eventually when their technology becomes wide spread. This is going to help keep drones up in the air without having to come down for recharging.

Dynamic movement - drones can now throw, catch and balance an inverted pendulums. This is a really tricky feat, looking forward to drone made cocktails…

Dogfighting - getting two drones to get in sync with each other is no easy task. This kind of control will help with managing fleet interactions and making them not hit each other in crowded cities.

Construction - The entire construction industry at some point will be automated which will include the use of drones to deploy materials - especially useful for temporary installations.

Other: Gesture controlled dronesSiri controlled drone, Mind controlled drone, Complex maneuvers, Swarm control,

To get more information about the drone community in general, check out Chris Anderson’s DIYdrones site.

Selecting the right business model for your business is crucial. In this post I intend to build on some of the work of Fred Wilson and others in the exploration of web and mobile revenue models. I propose there are two major classes to revenue models: trade methods and trade objects. A trade method would be for example, “licensing”, whereas, a trade object would be the “data”. Here is a fairly exhaustive list, extended from the original collaboration on hackpad. It is fairly interesting to be aware of all the possible combinations of trade methods and objects as it can help predict new startups or guide your own business model choice.imageTrade methods:


  • Normal ads
  • Display Ads - e.g. Yahoo!
  • Search Ads - e.g. Google
  • Text Ads - e.g. Google
  • Video Ads - e.g. Hulu
  • Audio Ads - e.g. Pandora
  • Paid content links - e.g. Outbrain
  • Email Ads - as done by Yahoo, MSN
  • Classifieds - e.g. Craiglist
  • Featured listings - e.g.  Yelp, Super Pages;
  • Recruitment Ads - e.g. LinkedIn
  • Promoted Content - e.g. Twitter, Tumblr
  • Lead Generation - e.g. MoneySuperMarket, ZocDoc
  • Affiliate Fees - e.g. Amazon Affiliate Program
  • Ad Retargeting - e.g. Criteo/perfectaudience
  • Real-time Intent Ad Delivery
  • Location-based offers - ex/ Foursquare
  • Sponsorships / Site Takeovers -  e.g. Pandora


  • Retailing - e.g. Zappos
  • Marketplace - e.g. Etsy
  • Crowdsourced Marketplace - e.g. Threadless
  • Excess Capacity Markets - Uber, AirBnB
  • Vertically Integrated Commerce - e.g. Warby Parker
  • Aggregator - e.g. Lastminute.com
  • Flash Sales:  Gilt Groupe, Vente Privee
  • Group buying - e.g. Groupon
  • Digital goods / downloads - e.g. iTunes
  • Virtual goods - e.g. Zynga
  • Training - e.g. Cloudera (??), -> Coursera
  • Pay what you want - e.g. Radiohead
  • Commission - e.g. SharesPost
  • Commission per order - e.g. Seamless, GrubHub
  • Auction - e.g. eBay
  • Reverse Auction - ex Priceline
  • Barter for services e.g. SwapRight


  • Software as a Service (SAAS) - e.g. Salesforce
  • Service as a Service - e.g. Shopify
  • Content as a Service - ex: Spotify, Netflix
  • Infrastructure/Platform As A Service - e.g. AWS
  • Freemium SAAS - e.g. Dropbox
  • Donations - e.g. Wikipedia
  • Sampling - ex Birchbox
  • Membership Services - ex Amazon Prime
  • Support and Maintenance - ex 10gen, Red Hat
  • Paywall - e.g. NYTimes
  • Voice and video-conferencing - e.g. Uberconference

Peer to Peer

  • Peer-to-Peer Lending - e.g. Lending Club,
  • Peer-to-Peer Gambling - e.g. BetFair
  • Peer-to-peer buying - ex Etsy
  • Peer-to-peer insurance/home/car - ex (??)
  • Peer-to-peer computing (CrasPlan storage, or SETI@home)
  • Peer-to-peer service - e.g. Mechanical Turk, TaskRabbit
  • Peer-to-peer Mobile WiFi/Tethering - ex (??)

Transaction processing

  • Merchant Acquiring - e.g. PayPal (Online / Offline), Stripe (Online), Square (Offline)
  • Intermediary - e.g. IP Commerce (POS 2.0), CardSpring
  • Acquiring Processing - e.g. Paymentech
  • Bank Transfer - e.g. Dwolla
  • Bank Depository Offering - e.g. Simple, Movenbank (spread on average deposits)
  • Bank Card Issuance - e.g. Simple (interchange fee per transaction)
  • Fulfillment - e.g. Amazon
  • Messaging - e.g. Peer-to-Peer SMS, IM, Group Messaging
  • Telephony - e.g. termination/origination in public telephony networks (skype out/in)
  • Telephony - e.g. termination/origination within private telephony cloud (e.g.  native skype)
  • Payment Gateways: Mobile -e.g. Braintree
  • Platform Monetization (“Tax”) - Facebook Credits; iO6 30% cut.


  • Per Seat License - e.g. Sencha
  • Per Device/Server License - e.g. QlikView
  • Per Application instance - e.g. Adobe Photoshop
  • Per Site License - e.g. Private cloud on internal infrastructure
  • Patent Licensing - e.g. Qualcomm
  • Brand Licensing - e.g. Sesame Street
  • Indirect Licensing - e.g. Apple Volume Purchasing


  • Paid App Downloads - e.g. WhatsApp
  • In-app purchases - e.g. Zynga Poker
  • In-app subscriptions - e.g. NY Times app
  • Advertising - e.g. Flurry, AdMob, Heyzap
  • Digital-to-physical - e.g. Red Stamp, Postagram
  • Transactions - ex Hailo


  • Freemium - Free to play w/ virtual currency - e.g. Zynga
  • Subscription-  e.g. World of Warcraft
  • Premium - e.g. xBox games
  • DLC - (Downloadable Content)  - e.g. Call of Duty
  • Ad Supported - ex - addictinggames.com

Trade Objects:


  • Impression (CPM)
  • Click (CPC)
  • Install or open application (CPI)
  • Action inside the app e.g. complete first level of a game or make your first follow (CPA)


  • Apps
  • Virtual goods
  • Videos
  • Games
  • Books
  • Magazines
  • Images
  • Gifts


  • Apps - e.g. whatsapp messenger
  • Virtual goods


  • User data - e.g. BlueKai
  • Business data - e.g. Duedil
  • User intelligence - e.g. Yougov
  • Search Data - e.g. Chango
  • Real-time Consumer Intent Data - e.g. Yieldbot
  • Benchmarking services - e.g. Comscore
  • Market research - e.g. GLG

Peer to Peer

  • Money - e.g. Lending Club,
  • Risk upside (gambling) - e.g. BetFair
  • Risk downside (insurance)
  • Production - ex Etsy
  • Computation (storage, processing and security) (CrasPlan storage, or SETI@home, bitcoin)
  • Service - e.g. Mechanical Turk, Exec
  • Communication - e.g.    shared wifi networks


Some of the business models cited are combinations of trade methods and trade objects e.g. in app purchases suggests both freemium (trade method) and virtual goods (trade object).

1. Github for other verticals. Social sharing, forking projects, backups and version control applied to other verticals.

  • Github. The original github for code. 
  • Cubehero. - Version control for 3D models, fork other people 3D projects. 3D “diffs” actually work.
  • Pixelapse - Version control for 2D photoshop designs, social collaboration
  • Benchling - Social collaboration for biologists, version control gene changes coming soon?
  • Layervault - Collaboration and version control for design
  • Upverter have a chance be the github for electronics

2. Opentable for activities (vayable might vertically integrate in this direction).

3. Small data. Better resolution using super resolution from local sources. E.g. super resolution applied to accelerometer data.

4. Makerbot of the bioworld. This cross references a pitch from the previous post but had better positioning.

5. The stripe of invoicing. Take invoicing into a clean API.

6. Yammer / Hipchat hybrid. Or a scalable hipchat. Take the real time nature of hipchat and take the scalability of Yammer. Hipchat doesnt work well for teams more than 25 people.

7. Social wikipedia with the mechanics of quora. Wikipedia is getting old. The barriers to entry to getting data into the system are too high and we don’t know who really edits the pages. Wikipedia doesn’t feel data rich anymore and is way out of date on many subjects.

After taking many flights last year I began to build up a list of things I’d love to see improved. I’ve thought about every interaction and process that became a nuisance, discomfort or inefficiency to my flight. The airline industry needs to move into a phase where design around user experience becomes the primary objective - profits will follow. Virgin Airlines is leading the way on this front and I fly with them whenever I can - I want to go much further. In this post, I attempt to outline areas in which the airline industry could improve the millions of human hours spent in the air. In many cases, these issues are the results of over-regulation, however, there is scope to out-engineer the issues.

Containerize the airline industry 

One of major issues we see is with the unloading and loading of passengers. Time, energy and human comfort is sacrificed daily because we cannot load/unload people onto planes fast enough. We end up wasting around 20 minutes on both the load and unload cycle (40 in total per trip).

I suggest we take a leaf out the shipping industry. As the book, The Box outlines, the shipping industry was extremely inefficient and non scalable before the standardization of the 40 ft high cube container in the 50’s and 60’s. With standardization, came speedy onboarding, pre loading, flexibility and efficiency.

If we re-engineer planes to have the cabin level detachable, we could add more doors and preload passengers into an airconditioned and powered unit at the gate with a precision conveyer belt to take the container cabin onto the main plane. We already do containerization for luggage. Passenges and crew could preload onto the cabin at the gate sorting out pre flight checks before the base plane was even there! In fact, passengers could load into the glorified container straight after security and not have to walk to their gate. Airports could achieve high packing density and we could have a production line of planes taking off in multiple lines. Passengers would save time/energy and airline would have flexibility to use different planes as the base to carry the cabin. The cabin would become the differentiator, not the entire plane. I understand the engineering complexity of such a system is high but the shipping industry also once had the same barrier to entry and proved the model works.


Take safety online

It would be silly if you had to recheck all safety tests with an independent agency before driving a car every single time. Having to watch the simple safety instructions every flight is a waste of time and energy for the passenger. No one even watches it anyway. We need to take this online and out of the cabin, making it a test that people hold a license for, rather like a driving license.

Create more space

Since the removal of free luggage on many airlines and the miniaturization of devices, people are travelling with hand luggage as their sole luggage. We have seen the huge increase in problems this has created in the overhead bin space. We need to increase the size of this space and be more flexible on sizes you can carry on in the first place. We should increase the cabin height and take up more original cargo bay space. I understand they have increased the space on the 787 Dreamliner and the A380 but I have not experienced them yet. I predict they have probably not gone far enough. As a simple fix, we could use some simple cameras to figure out where space is and flag a green LED for free space just in the same way some car parks do this for free parking spaces. Additionally, we could vacuum pack jackets and collapsable items.


Streamline security 

Global entry may be able to speed up border control for people in the US. I’d like to see a worldwide standard for this, where we can be prescreened to be a trusted global traveler in a trusted network. As you probably know, clear are speeding up security checks by making their own privatized security line. Visa and custom forms need to be taken online. People constantly don’t have pens to fill out these forms and they fill out similar information every time. These areas are probably the easiest areas to speed up. Even an array of iPads at border control with a data entry app could work. If people could fill out their visa forms from the entertainment terminal, you could have a standard laser printer on the plane that could print out the data on the standard forms, or better still from the passengers credentials in their Virgin online account.


Maximize comfort

If we fitted meshed chairs (Aeron style) we could allow people to reside in higher levels of comfort and temperature control. Passengers need better localized control on their personal surrounding air temperature and humidity. Right now, the controls are cruder than cars for the 1940’s. I see passengers frequently too hot or too cold, which helps add stress to our immune system while flying helping us receive the common cold. Having highly localized temperature displays on your personal screen might help the user understand how to control their temperature or hydration level. We need to redesign the experience based around the following principles:

  • Maximization of sleep
  • Maximization of entertainment 
  • Maximization of blood circulation 
  • Maximization of passengers control on their environment 
  • Skin breathability 
  • Minimization of disturbance
  • Reduction in backstress through support and maximized inclines.

Right now, little to no deep thought has gone into any of the design of the aircraft chairs. Switching to leather seats at Virgin was a good start for skin breathability and temperature control at the surface of the skin but the seats have a very long way to go. Staggered chairs (like a cinema) might allow more incline on chairs for the same packing density. Also, more exotic layouts may allow various tiers of seating to be produced.


Enhance sleep

Planes don’t seem to be optimized for sleep. Given this is one of the best ways to recover for travel, this seems backwards to me. Some non-exhaustive solutions:

  • Lenses that focuses reading light into the reading area without spread onto other passengers spaces. We could even use polarized light to reduce the ability for someone else to see the light while sleeping.
  • Adding an option to switch off notifications from the captain except emergency notifications. 
  • Contact lens cleaning solution and high quality night masks for purchase.
  • Using blue lights / daylight bulbs to help time zone people in and out of sleep and to prevent jet lag.


Cut the noise

Using some acoustic engineering approaches, we could battle unwanted noise that contributes to reduced sleep and lack of relaxation.  Likewise, we can improve the quality of the experience and let the passengers immerse themselves in their entertainment content. Some ideas for improvement here:

  • Active noise canceling for zones of the cabin: many of the frequencies on the plane are fairly regular and could be cut out with antisound. Speakers could be built into seats or above the seat facing down. I’m aware that some planes do this but I have not experienced quality that is sufficient yet.
  • Use “quiet zones” like you have done on Virgin trains.
  • Using hum, buzz and interference filters on the local audio jacks to bring audiophile quality to the sound. 
  • Adding dynamic range compression to the audio feed so that no volume spikes occur on passager announcements. 
  • Using laser / ultrasonic zoning of noise, which I’ve seen some night clubs use.
  • Noise suppression of the flush in the bathroom/restroom. I think the volume is too high to be safe.

Go functional

Many of the little details in the functionality of components in the plane have been overlooked. Briefly a couple of ideas:

  • Allowing passengers to see the additional data about the plane. Internal cabin pressure, temperature, cabin humidity etc will help passengers understand what they have to do to keep warm, cool or hydrated. 
  • Telling air stewards/stewardesses that your belt is connected with an LED so you can have the belt under the blanket. Frequently, people have their belt on, below the blanket and have to be woken up during turbulance or for landing because the staff do not know if the passenger has the belt on.
  • Using magnets rather than turn fob for securing tray tables. Rather like apple laptops power adapters use magnets.
  • Removing silly things like entertainment controls on your arm rest. These frequently activate the screen when your trying to sleep. Bring it all to the touch screen interface.
  • Better touch screen controls and higher resolution screens. These should be modular so screens can be updated as technology moves on, or so that you can bring your own iPad and hang it on the headrest.


Prevent illness 

As I understand, it is a common misconception that aircabin recirculated air makes you ill. It is the interaction with coughing and sneezing people that make people frequently ill after air travel. We need to look into specifically how we could reduce this issue. Maybe by coating common surfaces with silver or offering vitamin C tablets to everyone on the plane as part of the meal. There must be a solution to reducing the likelihood of getting a cold. Reducing direct drafts from the poorly designed temperature control systems would be a good start.

Entertain me

Restricted usage of electronic devices is ruining our entertainment experience. We must lobby hard to have these restrictions removed as they have little evidence backing up the risk of interference. I suggest we have a standard passed that will allow devices to be compliant during takeoff and landing by having clean EM outputs. Alternatively, we should electrically shield critical flight systems from interference so we can use all devices at any time on planes. This includes wearing noise canceling headphones at anytime during the flight unless the cabin staff is waving at you and needs to communicate.

Let’s fly

The ultimate flight should be one where you want to jump back on an airplane. Until Virgin Airlines opened up, it’s as if innovation on the experience of the passenger just stopped in the 1960’s and has been going backwards ever since. We still have a long way to go though.



When I first came across the mathematical simulation John Conway’s Game of Life I played with it for hours, constructing all sorts of weird and wonderful creations that seemed to mimic nature.  The Game of Life was the Minecraft of my generation; an elegant,  2D, monochrome, zero-player, non-social version of it.  From such simple rules, emerged what looked like “life”: self replicating structures that interacted with the chaos of the canvas. Over the last few years, I’ve been thinking of the variations of the Game of Life and will explore them in the post.

Walking russian dolls 

As outlined in this video, it is possible to nest the Game of Life within itself. It is possible to nest the Game of Life at infinite levels of simulation, building a fractal Game of Life. It is also possible to run the Game of Life recursively, i.e. the largest level of nesting interacting with the cells of the smallest level of nesting, a kind of ”Droste world wrap” if you will. Visualizing this will be pretty interesting as you will have to curve the grid. In a more meta, “Droste-esk” way, one can use Minecraft to build a Game of Life simulation.

Discrete to continuous 

Normally the Game of Life is run on a grid using integers of discrete cells, however, it you make these discrete blocks more continuous by using floating point numbers you get results that look a lot more like nature. Intuitively, this makes sense as the degrees of freedom for nature are much higher (either infinite or on a plank scale grid) than that used in the classic, discrete Game of Life.  When we zoom out to view biological life in nature, it looks suprisingly similar (in ways) to the floating point simulation that Jamie Zawinski created.

Quantum Game of Life

I started wondering whether one can model nature with a quantum Game of Life where each cell is modelled by a qubit. Here is a great page explaining how a 3D quantum cellular automaton works. It would be interesting to see visualizations of a N dimensional, quantum Game of Life and take the Zawinski’s floating point simulation concept across to the quantum realm.


Constructing interesting shapes in the Game of Life can take a long time. It can be done programmatically using Golly if you want to save time. Tom Robinson used Golly to build a generalized text or image printer in the Game of Life. I’ve yet to see a generic 3D printer inside the Game of Life, however, Paul Slocum has physically printed his Game of Life results.

Other avenues

The Game of Life is not over yet. Here are some other avenues that are worth thinking about. As usual, if I have missed anything out please email me or ping me on twitter.

  • Genetic Algorithms can be used to chase a fitness criteria which could lead to some interesting large scale structures that may be hard to program manually with Golly or imagine in the first place.
  • Graph automata - Using the graph nodes and edges rather than the cells themselves 
  • Falling sand game - A popular, fun, with a more complex ruleset
  • Game of Life in polar coordinates [no link]
  • Game of Life in the complex plane [no link]
  • Game of Life with a continuous variable (0 to 1) as the “state” of grid element or graph node/edge [kudos to Max Atkin for that idea]. Not sure what the rule set would be here.
blog comments powered by Disqus