By: Neel Mehta, Parth Detroja, and Adi Agashe
35 MINUTE AUDIO / 5,000 WORDS (20 PAGES)
For audio version join Plus
Would you like to be “fluent” in the language of technology? Read this book summary to familiarize yourself with the terms and strategies used in the tech world and become confident in your ability to discuss them with others. Whether you are looking to switch careers into the tech space or merely looking to grow your knowledge and understanding of the industry, this book summary will give you the foundation and insights you need. Understand the “what” and the “how” of key technologies, like how computers teach themselves new things, how Google Drive is like Uber, or how self-driving cars work. And more importantly, learn why the tech giants have made some of the business decisions that they have. Why did Google make the Android operating system free? Why did Microsoft acquire LinkedIn? Why does Nordstrom have free WIFI in their stores? Get under the hood of these questions and many more in this book summary.
First, understand how some of today’s most innovative technology works: Learn how companies like Netflix and Spotify easily tailor recommendations for you based on past viewing history. Understand how apps continually improve, decide on new features, and edge out their competitors with A/B tests. Speak intelligently about the shift towards the cloud and how Google Drive is a lot like Uber. Decode acronyms like API, NFC, and GAN, and apply your understanding in analysis of key business decisions of the world’s largest tech companies like Apple, Google, Facebook, and Uber. Gain familiarity with ever-improving innovations like self-driving cars and personal assistants.
Then, understand why tech companies make the decisions that they do: Why did Google make their Android operating system free? What motivated Microsoft to acquire LinkedIn? How do apps manage to generate revenue (other than via ads) if they’re free to download? Why was it a smart business decision for Nordstrom to install free Wi-Fi in all their stores? And, what’s behind the “software as a service” trend and the net neutrality debate?
How does it work?
This section outlines the “how” and “what” behind today’s key technologies. The goal of this section is to provide a solid foundational working knowledge from which to analyze business decisions, and therefore talk about them intelligently.
How do companies like Netflix and Spotify make recommendations for you?
The “recommendation” technology is quite ubiquitous today. For example, Amazon uses it to offer new products or services to you, while Netflix uses it for TV shows and movies and Spotify for music. But how does this technology actually work? With so many customers, there’s no way that these companies could employ the number of employees needed to make personal recommendations to each customer. There are two main types of algorithms.
First is a function called “collaborative filtering.” As customers shop, listen to music, or watch Netflix, they are slowly adding to their personal data cache that that company owns. For example, Netflix could keep a list of all the TV episodes you’ve watched in the last month. Finding new TV show recommendations for you using collaborative filtering involves comparing your “list” to others’ lists. When Netflix finds another customer’s list that is, say, 80% similar to yours, it has found a good match for collaborative filtering. After concluding that your list is somewhat similar to the other customer’s, Netflix will make the assumption that your overall tastes in television shows might be similar. Then, it will take a closer look at your respective lists of TV watched in the last month. Netflix will recommend to you the 20% of shows on the other customer’s list that you’ve not yet watched, and likewise for the other customer with the shows you’ve watched but they haven’t. Essentially, collaborative filtering works by identifying other customers who may have similar tastes and preferences as you, and seeing what they’ve found and liked that you may not yet have.
The second main way that companies make recommendations for you is by creating a personalized “taste profile.” For example, each song in Spotify is classified within a genre and a sub-genre. As you listen, Spotify keeps a running tally of the number of songs you’ve listened to within each genre and subgenre. Once a certain genre or subgenre gets enough “hits,” it will be added to your taste profile as a preferred type of music. Then, Spotify will begin recommending more songs that you’ve not yet listened to within those genres.
While recommendation algorithms are certainly more complex than described above, “collaborative filtering” and using “taste profiles” are the two main ways that companies can make solid recommendations to you based on “past listening patterns.”
How do apps successfully improve?
Ever wonder why apps like Facebook or Snapchat often change their interface and functions? Why all the changes, and how do they know they’re headed in the right direction? The technology behind these decisions is rooted in “A/B tests.” In the tech world, an A/B test is a process that companies use to decide between two (and in some cases more) alternatives – “option A” and “option B.” Because of the quantity, ease of access, and constant updating of their customer information, companies can simultaneously introduce option A and option B and determine which performs better among customers by measuring “clicks” or other identifiers of positive reception.
For example, the Washington Post regularly issues two versions of the same headline on their website. After analyzing the data, “developers decide which version is better and show the winning version to everyone.” Companies like Tinder, Buzzfeed, Upworthy, Facebook, Snapchat, Amazon, and many, many more utilize A/B tests to make critical business decisions about everything from the wording of headlines to new app capabilities and the most effective ads.
How is Google Drive like Uber?
Regularly utilizing Uber can greatly decrease one’s overall transportation expenses, depending on one’s lifestyle. In an urban area, taking Ubers instead of owning and maintaining a car can be far less expensive, when factoring in the costs of expenses like parking, gas, maintenance, insurance, and licensing. This “outsourcing” of an everyday need and shedding of responsibility for ownership is in fact the Google Drive model, or rather the “cloud computing” model in general. With the cloud, you can outsource the costs and maintenance associated with storing files. Instead of being on your hard drive, your files are stored in “the cloud,” or really on someone else’s server. “Google Drive is like Uber for computers. Instead of owning your own car or computer, you can get your files or transportation on-demand from anywhere with an internet connection.”
How can Apple Pay process transactions so seamlessly?
Though Apple Pay may appear “magical,” it actually uses an extremely secure technology known as “Near-Field Communication,” or NFC. NFC operates over radio waves and, when two devices are in close proximity, enables the transfer of information. This technology is very safe and secure for both customers and vendors. Apply Pay uses an encryption technology whereby the credit card information is not able to be decoded or stolen by hackers.
NFC technology is being used elsewhere as well and is likely to become more ubiquitous over time. For example, Chicago’s public transportation “Ventra” card uses this technology. Companies could begin using NFC in their ads so that potential customers could tap their phones to get more information. Some French cities have NFC stickers available that provide area maps on demand. “NFC is helping blur the line between the physical and digital world, and we think the uses of NFC will get more and more exciting.”
What technologies do Uber, Yelp, and Pokemon Go all have in common?
All applications are made up of “code.” In the case of Google Maps, Google has invested large sums of time and money in canvassing the globe to create their digital maps. Clearly, these maps are important and convenient technologies that can be applicable in a variety of scenarios. Rather than require that each app repeat this process to build their own digital maps for use in their apps, Google makes their maps available through “application programming interface” technology or “APIs.” APIs are “snippets of code” that let apps “talk to each other” and are a common occurrence in the world of technology. They basically allow an app to “pull in” the functionality of another app. In this way, many popular apps actually integrate the insights generated by other companies. For example, Yelp displays a locale on a Google Map as part of the details, and Pokemon Go uses Google Maps to identify your location. Uber uses PayPal’s Braintree API to process payments. Venmo uses APIs to send emails or text messages. They really are everywhere and in some ways have enabled the proliferation of many quality apps.
Why doesn’t Netflix crash when there are spikes in viewership?
Netflix is in an incredible position to handle extreme one-time spikes in viewership as well as growth in viewers over time due to its setup on Amazon’s cloud, Amazon Web Services. In 2008, it began transitioning its content to the cloud. Those the process took over seven years, it is likely that Netflix isn’t regretting the move. Because its content isn’t on its own servers, Netflix has “elasticity” and “scalability.” Elasticity means that “Amazon Web Services will instantly grow or shrink the computing power given to your app as your app’s usage goes up or down.” This means that Netflix no longer has to worry about a spike in viewers due to a new popular show crashing its network. And, when there are lulls in viewership, they don’t have to maintain expensive servers sitting inactive. Scalability in this case means that the cloud allows growth without worry about network capacity. “The video viewed on Netflix has grown 1,000-fold from 2007 to 2015.” Over this time period, Netflix didn’t have to worry about growing their capacity network-wise because of Amazon Web Service’s ability to provide more space, with no headache on Netflix’s side. Elasticity and scalability are two benefits that companies can leverage when using the cloud.
How do personal assistants like Siri or Alexa work?
These personal assistants work by first sending your voice to their company’s server rather than burdening your phone with attempting to process the language. With Siri, Apple’s servers then break down your speech phonetically and compare the sounds against their gigantic database of how others pronounce certain words until it finds a match. In doing so, they create “text” from the “audio file” you’ve created with your voice. From there, there are two main routes Siri could follow. If your request is somewhat complex, Siri may just do a general internet search and share the key results back with you. But, if it is a request more simply handled through an app, Siri will access an app to answer your query. An example of this could be opening the weather app or a calculator to solve a math question. Digital personal assistants are essentially a simple two-step process. Transcribe your query, then find the answer through an app or internet search.
How do computers teach themselves new things?
Computers can learn and improve surprisingly well due to technology called “generative adversarial networks” or GANs. These are a type of artificial neural network. We’ll first outline how a neural network functions and then explain how the GAN works. Humans learn through neural networks – that is, as we receive feedback, we change behavior accordingly. For example, we touch a hot stove and our hand is burned, so in the future we avoid making contact with hot surfaces. Computers have been designed to be wired in a similar way, such that they can learn and make refinements based on feedback and patterns over time. Some examples of artificial neural networks include “text autocorrect” on your cell phone or your email catching spam by picking up on suspicious patterns in the email.
Generative adversarial networks take this process one step further. They are so powerful they’ve even been used to create fake audio and video. So how does this exactly work? Putting GANs into action requires two artificial neural networks – one serving as the “generator” and one as the “discriminator.” The generator’s job is to convince the discriminator that what they are producing is a real version of whatever it is setting out to create (in this case, a fake news video or audio). “The networks get into a sort of arms race, with the generator trying to make more convincing forgeries and the discriminator trying to get better at policing fakes. The networks learn from each other, constantly improving, until the generator is churning out incredibly convincing fake things.” GANs were used in 2017 to create audio of prominent politicians reading their tweets aloud. But, it was fake audio. The politicians never read their tweets aloud. Rather, the audio was created using a GAN. This powerful technology is something to be aware, and perhaps wary, of.
What is the technology behind self-driving cars?
Self-driving cars use a combination of several technologies to orient themselves to the area and their surroundings and make decisions about how to operate themselves on the roads. First, there is the “onboard GPS” to help it figure out its general location. To get the details, it then relies on an “inertial navigation system,” or a set of sensors and compasses that are attached to the car. “These sensors tell how fast the car is moving and in what direction.” Next, to orient itself to its surroundings like people, road signs, and other cars, the self-driving car uses “hyper-detailed maps of the area” that “aren’t your garden-variety Google maps.” The race to develop these incredibly sensitive maps (“precise to the inch”) has led to competition among entrenched car manufacturers to be the first to master or acquire this technology. These maps can only help the car understand fixed objects, however, like curbs and street signs. To detect non-permanent or moving objects like other cars, bikes, and people, the self-driving car turns…