iOS 7 Killed Skeuomorphism but Metaphor Lives On (and Always Will)

“With the complete roll-out of iOS7 comes a new era for user experience,” the overzealous tech luminaries might say, “that is free from the shackles of skeuomorphic metaphor.” No longer will we rely on cheap images of wooden shelves and folders to understand the way applications work.

The fact of the matter is that, even with the (near) death of skeuomorphism, it is impossible to escape metaphor. In fact, you cannot conceive of much at all without metaphor. Before I receive any hate mail from militant anti-skeuomorphists, let me explain what exactly I mean by metaphor.



Conceptual Metaphor Theory was proposed by a linguist from UC Berkeley named George Lakoff, and it states that all abstract thoughts are understood through metaphorical comparisons to other more basic, universal concepts. One of my favorite examples of this is related to intimacy¹. In English, we talk about intimacy mostly in terms of proximity and temperature. Having a close relationship means you are very intimate. A warm person is one that is intimate with others. So how did these metaphors come to be? Why don’t we use something like verticality to denote intimacy?

The key maxim here is that metaphors for understanding such abstract ideas always break down to fundamental human sensory experiences that are generally learned in infancy. When you were a baby, who was the person in your life that you were most intimate with? Your mother. She held you close against her warm body when you were scared or hungry. As we grow up, we use these early experiences of intimacy to explain and understand our relationships with others. We can only conceive of intimacy in terms of human warmth and physical proximity — INTIMACY is WARMTH and INTIMACY is CLOSENESS.

Our understanding of different user interfaces is based on the conceptual metaphors as well:

  • We know to press the button on the top to increase volume because of the conceptual metaphor MORE is UP.
  • We know to look for inactive programs in the bottom of our screens and below other windows because OFF is DOWN.
  • We know to add similar files to the same directory because of the conceptual metaphor RELATIONSHIPS are CONTAINERS.
The file system on most operating systems is still based on a literal filing cabinet metaphor, which is why it is not common for one file to exist in multiple non-hierarchical folders -after all, they don’t in real filing cabinets!

The file system on most operating systems is still based on a literal filing cabinet metaphor, which is why it is not common for one file to exist in multiple non-hierarchical folders -after all, they don’t in real filing cabinets!

Our understanding of an interface, according to skeuomorphism however, is based on literal metaphor:

  • We know to press the button on the top to increase volume because of thats how it is on physical boom boxes.
  • We know to look for inactive programs in the bottom of our screens and below other windows because it’s the same with papers we haven’t looked at in a while getting lost under a big stack.
  • We know to add similar files to the same directory because that’s how filing cabinets work.

((The file system on most operating systems is still based on a literal filing cabinet metaphor, which is why it is not common for one file to exist in multiple non-hierarchical folders -after all, they don’t in real filing cabinets!))

It’s not a poor assumption², but it’s quite shallow. As you might notice, these literal metaphors are actually understood via conceptual metaphors – even the physical objects! It’s conceivable that if the metaphor MORE is UP did not exist, there might not be any vertically-oriented volume controls on boom boxes. It’s not boom boxes that make digital volume controls intuitive, its the underlying conceptual metaphors.

At the end of the day, skeuomorphism is more than a terrible design aesthetic; it’s an obstacle to doing really great design. Great designers create experiences based on what they know about how humans think, and thought is made up of conceptual metaphor, not just book shelves.

¹ Some other crazy metaphors to think about are ECONOMIES are BUILDINGS, LIFE is a JOURNEY, SEXUALITY is a WEAPON. You can find a list of other metaphors here.

² It is actually a great mechanism to increase early adoption for more ground-breaking technologies. When a new kind of technology enters the market, it is almost always compared to existing technologies to help users understand it — one example is the first cars being called horseless carriages. The iPhone was especially unprecedented, so it was probably strategic for Apple to base much of their UI on real-world things to drive adoption.


Why You’ll Take $6,000 Over Better Health

A few months ago, I met Andrew, a financial planner whose boyfriend, Jake is a doctor. As he described his work to me, I couldn’t help but remark on the similarities between financial planning and medicine. Both he and Jake have patients/clients with whom they communicate on on a consistent basis. They are also both responsible for maintaining important components of their patients/clients lives – their finances and health respectively. But Andrew later shared some interesting anecdotes that shed light on an unexpected difference between these two professions.

Andrew receives frequent calls from many of his clients, sometimes just to chat about family or current events. One client even requested that he spend three days with her in the Bahamas, drinking margaritas and picking out timeshares – but mostly drinking margaritas. Jake – the doctor – on the other hand, has difficulty maintaining even annual contact with his patients and when they come in to talk there is less opportunity to build the same kind of relationship that Andrew has with his clients; its straight to business.

While Andrew and Jake might not be the most quintessential exemplars of their professions, the truth is, people seem to hold their checkbooks closer to their hearts than they do their health. We make health sacrifices every day in favor of saving money, saving time, propelling our social standing, etc. Why don’t I bike to work instead of taking the bus? Why don’t I prepare a salad instead of grabbing a massive steak burrito? Why don’t I play football instead of sit on the couch watching other people play football? It’s about time and money.¹

It might not even be that we care about time and money more than our physical well-being, it’s just that they are easier for us to think about. There are two issues we encounter when conceptualizing health that do not exist for time and money: quantification and cause and effect. To demonstrate, let’s try a thought experiment:

What would you do with $1,000,000? If you don’t live in California where Uncle Sam pockets a good chunk of your money, you might buy a house or a Tesla as those who’ve made it like to do. Or if you’re a pragmatic Silicon Valleyite keeping up with the Joneses, you might invest in a startup some guy you met on Caltrain co-founded. Either way, there is a clear vision of what you can and cannot do with this money all the way down to the last cent.

Now, what would you do with excellent health? There’s the first problem – what the hell does excellent health mean? While no one has come up with any sort of dollar equivalent for health,² the quantified self movement is pushing towards it.

Those involved in this movement try to measure every aspect of their lives. They’ve got fit bits and fuelbands and gizmos and gadgets galore to measure everything from sleep to step count. The problem is, at the end of the day, all we can do is put this data into pretty graphs showing upward and downward trends. So you walked 200 more steps than yesterday? Cool story bro. Honestly, the body is a really complicated place and we’re still a long ways away from accurately quantifying health even across just a few metrics for specific diseases.

Ok, back to the scenario: assuming we had some way of deducing that you were in the top 10% healthiest people in the US (like you would be in the top 10% wealthiest with $1M), what does that actually get you? A faster mile time? Probably. Better looking abs? I hope so. How about a longer life? Perhaps. The issue is that even with great health, there is still uncertainty that lies in genetics and the environment.

I have this strange disposition to have disproportionately large biceps. Cool right? Nah – it’s to the point where if I work out for an extended period of time (which I don’t), I start to resemble something like Pop-Eye the Sailor Man. Humor aside, there is an array of chronic conditions that range in seriousness and visibility that cannot be changed even through medical intervention.

Suppose you do not have any of these genetic predispositions. If you’re eating right and working out with the aim to live longer, you might as well not leave your house because the probability of you dying in a car accident is higher than you think – unintentional injuries was the 5th most common cause of death in the US in 2011. The environment we live in, like our genetic makeup, is beyond our control, but plays a massive role in our overall health.

The point is this: we favor money over health most of the time because it is better quantified such that we understand exactly what it gets us. But building on that, the moral of the story of the financial planner and the doctor is that gratitude comes from understanding.
So, Andrew, you’ve helped me divest in Zynga and now I have $6,000 more dollars to kick around this week? Great, I’ll buy you copious libations and also that Les Paul Guitar I’ve had my eye on. You are essentially giving me the guitar.

So, the exercise regiment you’ve suggested seems to be making my heart stronger? Thanks, Jake… But it’s hard for me to be truly thankful when I don’t know (and truly understand) what this actually gets me in life, unless of course the stakes are higher.
So, you’ve removed a malignant tumor and there are no more traces of cancer in my body… How can I ever repay you doctor?

The truth is, I’ll take $6,000 over a “healthier heart” any day, but when it comes to life and death, its a different story. There is no amount of gratitude that I could give a doctor who saves my life or the life of a loved one, such a deed is simply not quantifiable.

¹ Obviously there’s social utility to some of this as well. Staying out late drinking copious libations is clearly destructive to your health, but its great for your social standing.

² Though, there was a film released in 2011 called In Time, which was about a world where the economy is based on time and individuals’ wealth lies in the minutes they have until they die – did I also mention Justin Timberlake is the lead actor?

The Silicon Valley Is Not an Industry-Agnostic Tech Haven

I have lived in the bay area my whole life, moving in 30-minute increments from Walnut Creek to Berkeley to San Francisco, but it wasn’t by choice that I never left. The opportunity and energy in Silicon Valley is unparalleled by anywhere else in the world. Plus, I’ve grown accustomed to the day-to-day efficiencies (like same-day delivery) that come with an innovation-based economy. Being somewhat of a jet-setter though, I have been hopeful that a similar atmosphere could be recreated elsewhere such as NY (Silicon Alley), LA (Silicon Beach), London (Silicon Roundabout), Tel Aviv (Silicon Wadi), Berlin (Silicon Allee) or even Lagos, Nigeria (Silicon Lagoon).¹

There has been a lot of talk about replicating elements of the Silicon Valley to foster a similar culture of innovation elsewhere. Several nations such as Canada and Chile are even engaging in advertising campaigns to entice entrepreneurs in the Valley to do their startups outside the US with tax incentives and better immigration policy. Most discussion of Valley clones usually ends with the same conclusion I came to in the third grade when I rationalized feeling different from the other kids: we’re special, we’re unique, my mom makes the best lunches at school.

There is indeed validity to these sentiments. It is actually very tough to recreate culture, and after all, the culture is a large reason why the Silicon Valley became what it is today. Sure, venture capital is essential, but it is the culture that made people willing to throw their millions of dollars into the Apples and Googles that kickstarted the industry.

What I do believe, is that the Valley will struggle to become an industry-agnostic technology innovation center and that there is significant potential for other regions to develop industry-specific tech communities.

The Valley as a Bubble of Consumer Goods and Services and Advertising

Let’s take a look at some of the products and revenue streams of some well-known tech companies. Google creates search engines and other consumer products and gets most of its revenue from advertising. Facebook built the biggest social network in the world and has also been monetizing its massive user database through advertising. Amazon is one of the most popular online marketplaces and makes money by taking a percentage of items sold on the site, but it also receives significant revenue from advertising as well. Are you seeing a pattern yet?

The Silicon Valley finds its talents mostly in two domains: consumer goods and services and advertising.

Wait a second… technology permeates every industry! Whether its focus is in healthcare or construction, every business is becoming a digital business. This is true, except that technologists themselves do not permeate every industry, just the ones they know.

The typical² story we hear is that Startup Stan goes to Stanford, majors in Computer Science, meets Co-founder Craig, grows his niche consumer-facing app startup after graduating, gets funding from some head honchos on Sandhill Road and eventually gets bought by Big Tech. The question is: when does Startup Stan have time to learn the ins and outs of hospital management or construction field force management. He doesn’t and that’s why he sticks to his guns in the consumer space. We’ve seen the alternate scenario of entrepreneurs innovating in spaces they don’t really understand end in disaster over and over again – if your name is Elon Musk, forget I said that.³

My favorite example (excuse my schadenfreunde) of this comes from a hackathon I went to where the winner was a group of amateur hackers that created an application to help people manage their diabetes. They were selected by top venture capitalists and tech execs to be part of a prestigious startup accelerator despite the fact that the diabetes app market is over-saturated by existing products with deeper functionality and that many people with diabetes don’t even own smart phones (90% of patients have Type 2 diabetes and typically come from poorer backgrounds).

If these developers went to medical school, things might turn out better for them. Fortunately for the rest of us, there’s no such thing as Consumer School where you go to learn how to maintain friendships, go to bars and buy expensive things online. Well, I guess you can go to Business School for that… Regardless, the point is that we already know how this all works.

The Tech Industry is most successful in domains that the average person understands without much education. Mark Zuckerberg is an expert in personal relationships. He’s had millions of interactions with people over his lifetime; but then again, so have you and I. The difference is that he had the technical skills and the timing to create the technology infrastructure for personal relationships before you or I did. Now take a hospital EMR system: in addition to timing and technical chops, you would also need as deep an understanding of healthcare management as you have for personal relationships.

Considering this, will the Silicon Valley be the center of all technology innovation or just that pertaining to consumer products and services and advertising?

Niche Industry Opportunities Outside the Silicon Valley

When I was in high school, I thought I was going to be a professional songwriter, so naturally I applied to schools in LA and New York and had aspirations to live there permanently. Why? Well, the music industry is centered in these cities and if you want to be anybody in music, you’d better move there. While the Silicon Valley is the center of tech right now, big industry-specific companies won’t move to Palo Alto just to attract entrepreneurs to innovate in their space. It’s up to the entrepreneurs themselves to embed themselves in their startup’s industry.

If I have a financial services startup, does it make more sense to be in Palo Alto or New York were many financial services companies have head quarters. In terms of gaining access to industry subject-matter-experts and potential acquirers, New York is the correct answer.

So, am I suggesting that all the financial services startups in the Valley should GTFO? No. Not at all. At the end of the day, bountiful venture funding is the lifeblood of the vibrant tech industry we have in the bay area, and is more or less unparalleled elsewhere (if not in amount of capital but willingness to fund startups). Opportunities for funding is indeed growing outside the Valley. Waze (Tel Aviv) was rumored to be in talks with Google for a $1.3 Billion acquisition, and SnapChat (LA) has been valued between $500 Million and $1 Billion. When these companies are acquired it will create a new generation of millionaires that will likely turn around and fund startups in their respective ecosystems. Once money is flowing, I believe that each Silicon [[insert idiosyncratic noun here]] will become specialized technology innovation centers.

¹ Seriously, these are real. Wikipedia says they are.

² There are indeed many exceptions to this story, but this one seems to find its way into tech media every day.

³ Though, I can’t imagine Musk threw in millions of his own money into SpaceX and Tesla without gaining a deep understanding of the space and auto industries.

Could the Canadian Oil Frontier Mute Musk’s Electric Efforts?

My mother and father were die-hard hippies in the 60’s and I went to school in Berkeley, so naturally the electric car was a big deal for me growing up – it tugged at my environmentalist, tree-hugging heart strings. But for a long time, that’s just about all the electric car was good for; it was not really a serious competitor in the auto industry. That is, not until the recession hit…

As budgets were being slashed left and right in a country wrought with frivolous spending, Americans looked inside themselves for change. Gas prices were always the topic of conversation, “can you believe that gas is at $3.43 today?”. Feelings of pain from both the pocketbook and the heart for supporting such an expensive and destructive necessity plagued the developed world.

The hybrid, though still somewhat costly, was an answer to these woes. With their prices dropping every year, more and more Americans have been able to shed themselves of the shackles of big oil – at least at a constant RPM on the highway.

Today, fully electric cars are making a comeback partially thanks to technology-entrepreneur-turned-space-explorer-slash-automaker, Elon Musk, who recently announced that his company, Tesla Motors has plans to release a $30k electric car, a departure from the $70k+ luxury vehicles the company currently produces. There has been a lot of hype about cheap electric cars fundamentally changing the auto landscape for good – I agree, but unfortunately, this might not be true for a long time thanks to new scientific breakthroughs to help Canada access 176.8 billion barrels of mostly untapped oil reserves.

The UAE became the land of indulgence and wealth because of immense oil fields in the region. What many fail to realize is that Canada has a comparable supply of oil, but it has been largely inaccessible because it mixed into what is called bituminous sands, or tar sands.
Tar sands are nasty concoctions consisting of sand, clay, water, and petroleum, making them extremely difficult to separate out and extract the petroleum from. Long story short, scientific advances are making it much easier to mine these deposits. And cities in Alberta, Canada where they are located are booming as a result.

So what could this mean for the electric car?

Assuming a perfect market, more supply of oil coupled with lower demand for oil (given the popularization of electric and hybrid vehicles) could mean cheaper oil. Additionally, with bountiful resources and years of experience, established oil companies are in a much better position to manipulate the market than companies like Tesla Motors who are just starting to gain traction. Lowering the price of oil makes gas-powered vehicles cheaper to own and the risk of a consumer-grade electric car less enticing.

Now, now, now.. Let’s not get ahead of ourselves here. I’d like to throw out a few more ideas before we all start proselytizing the death of the electric car (again). Even if the above-scenario actually panned out as described, Tesla could return focus to its successful and growing luxury car line and still be a prominent player in the industry.

At this point, the electric car does not hold a large-enough share of the market to drive oil companies to compete by bringing down the price of gas. There is no doubt, however, that the electric car is growing and big oil needs to seriously consider new long term strategies such as focusing on some of the many other crude oil products like plastic or electricity.

Another important point is that there are several developing economies around the world, namely China, that have an exponentially growing demand for gas. It is also unlikely that the electric car will penetrate these markets any time soon, partially because of a lack of infrastructure to support them.

Additionally, the refineries in the US do not currently have the capability or capacity to handle such a flow. The keystone pipeline was only built to bring oil from Texas to Canada because the reserves were in Texas. Now that the reserves are in Canada, there is a slew of legal and technical hurdles to overcome. Unless the US state department approves projects to rengineer the pipline to flow north to south and cross international boundaries, there is talk that the oil will be piped to Vancouver and shipped directly to China via boat. While it is in the best interest of the US to play along and aid such a project, this is a massive undertaking, and to pay for it, oil companies might choose to keep gas prices at status quo despite its increased supply. And if supply is still low due to low capacity of the refineries, there would be no reason to lower prices anyway.

Only time will tell what will happen to Musk’s venture with the $30k electric car, but one thing is certain: Saudi Arabia can eat its heart out – Canada is doing big things.

The Future of Healthcare Could Be In Rural India

In my last post, I talked about using big data to improve patient outcomes by providing better diagnoses and treatment suggestions, an idea made popular by Vinod Khosla. Despite the great potential for this technology to improve lives as well as cut costs, the unfortunate truth is that most healthcare systems likely just aren’t ready for it. There three main reasons why this might be:


There is a saying that goes “only when humanity is at the brink of destruction, are people willed to change”. While this might be an overdramaticization, the idea goes without saying that we are not at this dire point in healthcare. Across the developed nations, for the most part, people are fairly healthy and receive treatment when they need it. I haven’t been to the doctor yet this year but do I feel healthy? Yeah. Maybe I am just a lobster in a boiling pot, but its easy to get complacent when my friends and family aren’t dropping dead from the Black Plague by the cartload. We are indeed becoming more health conscious, but until the water starts boiling, advanced medical technology is a nice-to-have, not a necessity for survival.¹


We are reaching a point in the United States where the current legislation governing healthcare administration is unbearable. Regulation constricts what kind of care citizens can receive and what kind of data doctors and patients have access to. Since the essential component of a medical decision engine is the data itself, it cannot come to be without improved open-data initiatives and better integration between EMR systems. Google knows everything about my life on the Internet, but its tough today for my doctor to know everything about my health, even if the data was out there.


A medical decision engine using machine learning algorithms needs a lot of data to work – that’s why open data initiatives are so important. Naturally, as early versions of this technology are introduced and gathering data for the first time, they will not be very good, which is why engineers and doctors will have to work together to hard-code some of the capabilities in the beginning. Instead of using complex algorithms do make a diagnosis and treatment suggestion, it would use simple logic. The problem is that a lot of the illnesses people have in the developed world are too complex and nuanced to be successfully diagnosed and treated with a decision tree. Compare the diagnosis/treatment of heart disease (the leading cause of death in the US) with influenza (a significant killer in India).

So this is why we can’t have nice things: we are complacent, over-regulated, and our problems are too complicated.²

Fortunately (just for the sake of the advancement of technology, it is actually quite unfortunate), there are parts of the world that have both the burning need and regulatory flexibility to make medical decision engine technology a real possibility – India, Africa, and Asia. It is my belief that in the very near future we could start to see futuristic health technologies implemented in the slums of Rajastan before they hit the wealthy suburbs of San Francisco.

Despite having some of the greatest doctors in the world, the health situation in India (which I will focus on for now) is dire. Many people live in extreme poverty, have poor living conditions, are malnourished and have limited access to medical professionals. One of the leading causes of death is from diarrheal diseases – a minor illness in the United States. Clearly, India is in desperate need of better care and the effect that even small improvements could make a huge impact. Additionally, the barrier of entry is much lower than it is in the US because of fewer regulations.

Let’s step back summarize:

Need Effect Barriers
US Low Medium High
India High High Low

Awesome. India is ripe for Vinod Khosla’s medical decision engine. And there are already rumblings laying the infrastructure today. Google is currently testing out blimps that carry wireless signals to remote regions of Africa, Asia, and India and there has been talk about pairing that with cheap smart phones.

Fast-forward and imagine a new India where the Red Cross pays annual visits to villages and slums to equip individuals with Google phones that access blimp-distributed internet and train them on how to use the medical decision engine to serve their community. Foreign philanthropists could fund this endeavor to create monetary incentives for the trained villagers to continue using the technology.

Bill Gates recently published a blog called Why I’m Going to India where he discusses the amazing potential for India to overcome its problems utilizing new advancements in science and technology. I couldn’t agree more that the opportunity in India is abounding and that it will be ground zero in the adoption the latest and greatest health technology.

¹  I will nod to the notion that we are getting to this point with the healthcare system itself (rather than our physical health). We need to find cheaper alternatives to the current model, and technology has proven to be an enabler in this respect time and time again.

²  More complicated than the third world rather.

The Future Doctor: Man or Machine?

Vinod Khosla, a famed Silicon Valley investor, founder of Khosla Ventures, and co-founder of Sun Microsystems, gave a keynote address at the Health Innovation Summit that was hosted by Rock Health in San Francisco last year, suggesting that 80% of doctors’ functions would be replaced by computers in the future. This post serves as a general overview of this domain of thought, providing the background for subsequent posts on the subject.

Becoming a doctor licensed to diagnose real patients takes years of training and hands on experience. Medical school is 4 years, residency is anywhere between 2 and 6 years, and then specialization is another few years tacked onto that.  Medicine has among the most grueling mandated education of any profession, and for good reason. Stakes are high when performing a surgery or diagnosing an illness – many decisions are the difference between life and death.

Because of this seriousness, many regulations have been put in place over the years preventing the adoption of new technology in hospitals and the opening up of patient data. In the next decade however, with the rise of Big Data initiatives, we may begin to see a new kind of patient care based on algorithms rather than human doctors.

What is a doctor but a decision engine?  Input a patient’s symptoms; compare that to their medical history, test results, and a database of all of the previous patients the doctor has experienced or learned about over her education; and output a diagnosis with a probability of being correct. What a doctor does on a computational level is nothing new in machine learning; in fact, the algorithm that Google uses to serve specific ads to their users could be applied exactly to making a diagnosis. Users are served ads based on their current search query, past search queries, social data, and the data of similar users. Let’s look at the metaphor:

user = patient
ads = diagnoses
search query = symptoms
past search history, social data, etc = medical history
other users’ data = other patients’ medical history

But why would Google be better than a real human doctor who went through years of education? Big Data > Small Data. Your doctor has only been able to read so much medical literature and experience so many patients in her career. After a certain point, their memory and ability consolidate experiences is very limited. The diagnoses and decisions that come out of your doctor are based on her experiences and learnings alone. It is the case that many doctors have different opinions about the same affliction – this leads to inconsistent (and sometimes incorrect) treatment.

Now imagine if you could merge all of the knowledge of every patient’s medical history into one database and then make treatment decisions from that. Researchers at Indiana University have actually built a system that does something similar to this and gave it to doctors to inform their diagnoses. They found that doctors using their system were significantly more accurate than doctors who did not (

Pretty amazing right?

So, why hasn’t Google already built out this technology? Larry Page was asked just that at Google I/O this year and he responded by saying that the regulations in the industry are too obstructive. As discussed earlier, the healthcare industry is one of the most regulated in the world, and is even more rigid in the US. Unless real progress can be made in consolidating patient records and opening up the data, this future of improved patient care can only be a dream – at least in the developed world.

I believe that it will be in developing nations where regulation is low and the demand for care is high that we will see the integration of this technology first. More on that to come.

Pandora’s Playback Pitfall

Back in high school, I used to commit hours and hours each day perusing magazines, blogs, and record stores, looking for cool new artists like it was my job. There was a thrill in being the first in my group of friends to start listening to a great band, an early adopter if you will.

But since going to college and joining the work force, I have had much less time to commit to this endeavor. For a while, I was relegated to listening to artists that I had listened to thousands of times before – I started to understand why our parents seem to be stuck listening to music from when they were young.Pandora Mobile App Logo

Fortunately, though, I began using Pandora and things changed for the better. Pandora let me passively discover new artists by simply creating online radio stations around artists I knew I liked. Every other song I heard on it was a gem that would have taken me weeks to discover back in high school. The music genome algorithm was quite impressive – sometimes I thought it knew my music preferences better than I did.

Within a short time though, my shining Pandora experience took a turn for the worst. As I gave songs I liked “thumbs up” thinking that it would show me more new music that is even more customized to my preferences, it seemed as if it did the opposite.

Pandora got lazy.

Instead of playing me new music, Pandora played the songs I had liked over and over again; haunting me and eventually making me hate them all. A “thumbs up” for Pandora doesn’t mean “play me more songs like this”, it means “play me this song a lot”.

What’s more is when I sought to escape this terror by creating new Pandora stations, the songs I had liked crept into these stations as well! I couldn’t listen to Bruno Mars without hearing Frank Sinatra! It was horrific.

The more I listened to Pandora, the more it killed my love of music.

Unfortunately, I am still struggling with this problem today. I now rarely listen to Pandora, substituting custom playlists on Spotify and even Spotify radio instead – though, Spotify is not without plenty of shortcomings itself.

What I would really like to see is a redesign of Pandora’s user experience to directly address this issue.

I should mention that Pandora did recently add a feature that allows users to say “I’m tired of this track”, taking it out of regular rotation for a short period of time – I can’t count the number of songs I have done this to. But pressing this subtly placed button every time you hear a song you’re tired of is pretty tedious and it uses up precious song skips. Also, it is perfectly plausible that sometimes I might actually prefer to hear some of the songs I’ve liked over new songs. These things considered, I believe that there is an elegant solution to this issue and it relates to an algorithm I learned in a class in which I built AI to play a version of Pacman

On one particular assignment, we were to write a program that could guide pacman across a bridge with a large reward on the other side – think the biggest, juiciest Pacman pellet ever. The catch was that Pacman had to teach himself what the correct path to the other side was without any prior knowledge. You would receive less and less points every time Pacman fell off the bridge and failed to reach the goal. The algorithm we used to teach our Pacman about the course while guiding it down the right path was a Markov Decision Process and it had an interesting parameter that let you manipulate how often Pacman tried to discover new paths or stick to what he already knew – exploration or exploitation.

Back to Pandora. I am not sure what algorithm is used by the music genome project, but I would be surprised if this “discovery parameter” was not somewhere in the code. My suggestion is to bring this parameter to the interface and allow users to manipulate it as they will.

Imagine a slider that would allow you to toggle between hearing mostly new music, and mostly songs you’ve liked before. Please, Pandora – the brutal onslaught against my favorite artists has gone too far. Consider this suggestion, for the sake of music appreciation.

Also, Jay-Z should not be showing up on my House music radio station… I’ll save that discussion is for another time.