Shubham Jain's Blog

Algorithmic Feeds Need to Be Banned

Wed, 25 Feb 2026 10:11:13 GMT

"What Orwell feared were those who would ban books. What Huxley feared was that there would be no reason to ban a book, for there would be no one who wanted to read one. Orwell feared those who would deprive us of information. Huxley feared those who would give us so much that we would be reduced to passivity and egoism. Orwell feared that the truth would be concealed from us. Huxley feared the truth would be drowned in a sea of irrelevance."

— Neil Postman, Amusing Ourselves to Death: Public Discourse in the Age of Show Business

I often wonder if it's simply my age or the world is truly going to sh*t. When you've an opinion like that, it also means you're disregarding facts that prove the opposite: poverty is down, child mortality has fallen, wars are fewer, people live longer, and work fewer hours.

I don't dispute that the world is getting better in those ways, however, I am convinced that it's also getting worse in many. For eg, it's the decline of slowly leisurely activities and increase in cheap dopamine seeking. In US, 40% teenagers were reported to be using smartphones more than 8h per day. And I bet most of it was on scrolling social media feeds. The number of people who didn't read a single book in 2023 was alarmingly 46%, a radical jump from just 24% in 1990.

Some may dismiss these concerns by pointing to historical moral panics—early 20th-century warnings about novels, crosswords, or even cycling supposedly corrupting society.

What makes the current situation troubling is its correlation with a broad and visible decline in mental well-being.

From 2011 to 2022, adults who experienced poor mental health rose from 35.7% to 42.5% according to BRFSS, from 31.1% to 35.8% according to NSDUH, and from 18.7% to 20.5% according to NHIS. Source
“One of the notable trends we observed was a sharp rise in the rates of anxiety, depression, and stress in adolescents and young adults over the past 20 years. Rates increased by more than two-fold for those aged 16–24, and understanding what is underlying these patterns is of critical importance.” Source

And I suspect that the real picture is far worse. Because, having anxiety, stress is quite different from being clinically diagnosed with it.

I see people around me and it's all too evident. People are more isolated, more absorbed in feeds, more socially anxious, and increasingly pessimistic about the future.

As someone on HN put it, “the future has lost its appeal to me.” I couldn't have put it better how I feel these days and I suspect everyone else too.

What is the source of this pessimism? I remember being techno-optimist for the most part, thinking how technology is making everything better. Where did all go wrong?

In my view, they did with the invention of algorithmic feeds. Social media had a bad rap from infancy, but gradually these platforms started ignoring what you explicitly chose—your interests, your follows—and started showing things to increase engagement, for eg, based on what others happened to like. The turning point was TikTok. That's where things went rapidly downhill. The idea was simple: discard user intent altogether in favor of pure engagement optimization. Relevance didn't matter anymore, it all about reducing friction to extend the time spent on app. Seeing it succeed, it was only a matter of time when everyone else followed.

Algorithmic feeds don't care whether you feel miserable, angry, or anxious after showing you something. They don't care about the quality of the content, nor whether it is shallow, sexist, misogynistic, or socially corrosive. Their sole objective is time-on-platform.

Our attention spans already bad as they were have fallen off the cliff. We are often angry at the world or anxious about the future and slow leisurely things are just not fun and enjoyable anymore. It's the Neil Postman's worst fears realized—a dystopia not through censorship, but through constant stimulation.

We spend an enormous amount of time debating artificial general intelligence and the hypothetical risks it may pose, while largely ignoring the fact that algorithimic feeds have already been doing the damage. Instead of showing what we chose, they are exploiting our worst cravings and making us all misrable.

It is no accident that rage-bait—content solely created to provoke outrage—has become a quick and dirty way to rise on social media. Content goes viral not because it is insightful or good, but because engagement that results from you being angry.

So what can be done?

The worst part is this is how no one seems to see it as a problem. Most politicians, regulators are chasing the easy targets: restricting it for teenagers, or censoring some content, or preventing fake news, but no one is asking why we are allowing algorithimic feeds to slowly corrupt our minds.

The issue is that much of the problematic content doesn't violate any law but still is poison that shouldn't be on anyone's feed. The easiest example is painting feminism as some evil propaganda against men, and how women's place in the kitchen is natural order of things. Such views have gained steam in the recent times purely because of social media.

Regulating these feeds would mean recognizing the problem in the first place, and I don't hope that will happen anytime soon.

The ideal regulation would be to ban or heavily regulate algorithmic feeds. If you don't follow someone, or a page, or an interest, there's no reason for you to get that stuff on your timeline. Only things that should be shown is what you explicitly choose. And if it does, it should be easy enough to nuke it from your timeline.

Yes, the feed would grow dull as it was in when Facebook was still early, but it also means our sanity will return. The social media of 2016 was bad in many ways, but our debates were centered around how seeing bright parts of others lives made us feel worse. I would have never imagined that I would be staring at re-emergence of loud and unashamed racism and misogyny.

Banning algorithim feeds won't magically make social media clean, but it would a decent reset.

Will AI Disrupt my Field?

Tue, 24 Feb 2026 11:16:59 GMT

[Satire]

Don’t EVEN ask the question. The answer is YES.

Cybersecurity? Models will detect, exploit, patch, and write the postmortem for the zero-days that haven't been discovered yet.

Food delivery? That burrito will be assembled, priced, routed, and reviewed by models negotiating with other models. And who will delivery the food? Of course, AI self-driven vehicles. The only human in the loop will be you — a temporary inefficiency, which will be eliminated when it simply starts predicting your hunger.

Therapy? Replaced by a system fine-tuned on every transcript, not exhausted by your mind-numbing ramblings of your childhood trauma.

Teaching? Replaced by adaptive tutors that detect cognitive fatigue and adjust difficulty, tone, and encouragement levels dynamically. On second thought, you may not even need that since there would be no point of learning anything.

Movies, videos, music? AI will be your personal studio running 24x7. It will drop a surprise album at midnight referencing that oddly specific memory from 2014 you never told anyone about. Celebrities won’t age as they will just virtual, personalized personas existing just for you.

Every workflow, every profession will reduce to an API call with billing rates attached. The GDP of entire sectors would collapse into server racks.

Your think your expertise will have any value? It will represent a fleeting arbitrage between what models cannot yet do and what they will accomplish next quarter.

Building a brand, slide deck, app, or startup will mean nothing. Any idea you can think of will have already been simulated thousands of times to produce an optimized version of everything possible. Think you can do better? The AI would've already priced in your optimism, your skepticism, and your eventual acceptance.

The value of everything except AI overloads would approach zero. Anything that matters would be replicated, optimized, and eventually distributed at marginal cost.

So, when you ask whether your field is safe, whether your niche is defensible, whether your uniqueness gives you leverage, understand this: your pitiful little existence, your insights, your "hard-won knowledge" is just a raw material for whatever is coming next. Your nothing more than lamb that gets bred, fattened, and eventually sacrificed.

Fear is Just Perceived Lack of Control

Thu, 25 Sep 2025 11:16:59 GMT

Wingsuit BASE jumping is one of the most dangerous sports in the world. Studies indicate that it has a fatality rate of nearly 1 in 500 jumps. This means that every active participant faces alarmingly high odds of meeting with a fatal accident.

I always wondered why did these jumpers took such enormous risks for tiny amount of thrill? Couldn't they see statistics? To me, the answer was obvious: they were adrenaline junkies. Addicted to the sport just like an addict is to his drug. For them, nothing could replace the rush of skirting the edge of death.

However, reading Flow by Mihaly Csikszentmihalyi, I came to another conclusion. In the section called Paradox of Control, he explains.

It is usual to explain the motivation of those who enjoy dangerous activities as some sort of pathological need: they are trying to exorcise a deep-seated fear, they are compensating, they are compulsively reenacting an Oedipal fixation, they are “sensation seekers.” While such motives may be occasionally involved, what is most striking, when one actually speaks to specialists in risk, is how their enjoyment derives not from the danger itself, but from their ability to minimize it.

So rather than a pathological thrill that comes from courting disaster, the positive emotion they enjoy is the perfectly healthy feeling of being able to control potentially dangerous forces.

Yes, thrill-seeking may be part of their motivation. But that explanation alone is too simplistic. BASE jumpers often believe they are in control, and that they are better at it than everyone else. It’s not that they don’t feel fear, but rather that they see their abilities as equal to the dangerous forces they’re confronting. Their relationship with fear is very different. It's like a challenging hurdle they have to cross, and not something that can potentially result in their death.

Alex Honnold, the renowned rock climber best known for free solo-ing El Capitan, touched on something similar.

I differentiate between risk and consequence. Sure, falling from this building is a high consequence, but, for me, it’s low risk.

So, yes, Alex realizes the stakes are high (death), yet believes his control is so complete that falling would be practically impossible.

Fear is Perceived Inablity to Control

Fear is obviously a useful mechanism that keeps us safe from genuine dangers. But fear is also a limiting force which might stop us from doing things when the actual threat is minimal. Things we might like and enjoy.

We assume that the fears are deep part of our personality. The endless catalog of phobias doesn’t help: acrophobia, arachnophobia, aquaphobia. By giving every fear a clinical name, we make it seem like they are all hardwired.

I believe most fears don’t come from the danger itself, but from our perceived inability to control a situation.

We fear water because we imagine ourselves drowning. We fail to see that once swimming skills are embedded in muscle memory, drowning becomes almost impossible (at least, in normal conditions).

We fear heights because we can’t control the intrusive thought of falling even when we are super-safe.

For the longest time, I was dead-scared driving because I was sure I would run into an accident. I had a deep-seated fear of (Indian) traffic. I couldn't see that once I learn driving and practice enough, I can drive on auto-pilot basis just as million others do on daily basis.

I did exactly that. Over a period of time, the fear dissipated, replaced by a trust in the ability to control any situation even in chaotic driving conditions of India.

From this perspective, even BASE jumping starts to make a strange kind of sense. It’s not as if someone wakes up one morning, straps on a wingsuit, and leaps off a cliff. They likely progressed through various stages, grew more and more skilled, and then started the penultimate sport of BASE jumping. They understand the risks, but they also think they have the skills to minimize it.

Of course, it's not for everybody. And no matter how confident the Wingsuit flyers are, it's undeniable that not everything will be under their control. One little mistake is all it takes to die a horrible death. But what's interesting to me is how humans can learn to minimize fear even in the riskiest of situations.

Over the years, I’ve learned to reframe fear not as an ingrained trait, but as a shifting relationship with control. Fear is there because we perceive ourselves to be powerless, threatened in certain situations. No surprise how quickly fear dissipates once we acquire enough skills to understand and control the exact situations. And that’s exactly why the seemingly dumb advice to “just face your fear” can actually work. It forces exposure to build the very skills.

Most of our fears require only a tiny amount of skills, or tweaking our thought process, before we realize how misplaced they were.

The Easiest Form of Mediation

Thu, 03 Apr 2025 11:16:59 GMT

Meditation is great — most of us can agree on that. But like exercise, it’s been overcomplicated. There are just so many forms of it: guided, silent, body scans, vipassana, kriyas. There are apps, videos, retreats, books, and self-proclaimed gurus. Everyone’s selling their version of meditation.

I also think its benefits have been oversold. Only a few years back, I was convinced that it was a path to everlasting bliss. An intense form of meditation called Vipassana is often touted as a tool to permanently transform your mind.

I don't believe all that these days. And honestly, the whole “leave the material world behind” thing gives me cult-ish vibes that I’m not comfortable with.

Meditation, to me, is just a tool to denoise the brain—and that’s become crucial in an age of constant overstimulation. Denoising helps you tune into the present, improves your focus, and helps soften the grip of negative thoughts.

Transcendental Meditation

Once I reframed what meditation is—a tool to denoise the brain—I had much more realistic expectations. After experimenting with various techniques, I found one that fits perfectly: Transcendental Meditation.

Don't worry, the name makes it sound advanced, but it’s actually the simplest one I’ve tried.

That said, it’s promoted by an organization that is borderline scammy (trying too hard to sell something), so good information is weirdly hard to find without paying. But the technique itself is surprisingly easy and effective.

Here's how to do it:

Just close your eyes and repeat a simple word or mantra in your mind (like, om, shrim, or namah). It should be pleasant sounding and ideally meaningless. Don’t try to control your thoughts. If your mind drifts, gently return to repeating the mantra.

That’s it. Sit on a chair, sit cross-legged, or lie down, doesn’t matter. Set a timer for 15–20 minutes, or go by feel. Do it in a park, or at home; in the morning, evening, or at night.

Everything is up to you.

Why it works?

Like I said earlier, meditation is just a way to denoise your brain. Whether you focus on your breath, your body, or a mantra, it’s all aiming for the same outcome.

Repeating a word in your head is especially effective because, oddly enough, the mind can only think of on one thing at a time. When you repeat the word, it gently cuts off wandering thoughts. Over time, this starts to clear the mental clutter. The mind calms down, and you become more in tune with the present and your own thought patterns.

When I practice it, I feel less overwhelmed, more focused, and the urge to doomscroll quiets down. At night, something similar happens, and it's quite helpful in helping me fall asleep. I like this analogy: mental clutter is like having too many tabs open in your mind, and meditation is a tool to press "Close All."

People who espouse transcendental meditation often make much bigger, more spiritual claims about it. But as I have said, my idea of mediation is modest.

Maybe those deeper effects show up with long-term practice, and if they do, great. But I am okay not having that expectation.

The Truth About Bangalore’s Water and Hair Loss

Wed, 02 Apr 2025 11:16:59 GMT

The idea that Bangalore’s water causes hair loss—a modern myth—began around a few years back, and today, it's treated as an unquestionable fact. Like many such myths, it thrives on casual observations: someone notices their or their friends' receding hairline and quickly concludes that something in the water must be to blame. And from there, the story spreads.

In India, we lack the scientific or statistical groundwork to rigorously confirm or refute such claims. But with high confidence, I’d say the supposed link between Bangalore and hair fall is most likely a spurious correlation.

Hairfall != Hair loss

First, it's important to differentiate between hairfall and hair loss. Factors like pollution, poor hygiene, or hard water can make hair dry, frizzy, or brittle, which may lead to increased hairfall — hair breaking or shedding from the shaft. However, that doesn't necessarily mean permanent hair loss.

True hair loss involves the detachment of hair from the root, often leading to reduced density or bald spots. It’s entirely possible for someone to experience noticeable hairfall yet still maintain thick hair, simply because the follicles remain healthy and intact.

Can Hard Water Cause Hairfall?

There are two studies on the subject of hard water and hair breakage that I have found, and they have given mixed results.

A study published in the International Journal of Trichology investigated the impact of hard water on hair's tensile strength and elasticity. Researchers immersed hair samples in both hard water and distilled water over 30 days and found no significant differences between the two groups.

Conversely, another study examined hair samples from 70 men, comparing baseline tensile strength to hair treated with deionized water and hard water. The findings indicated a significant decrease in hair strength after treatment with hard water, suggesting that hard water may weaken hair and increase the risk of breakage.

But here’s the thing — both studies on hard water didn’t measure hair loss, they measured hair damage. They took human hair samples and treated them with hard water in lab conditions, then measured the effect on hair's strength and elasticity. What they observed was how the hair shaft responded, not whether hair fell out from the root.

So there’s no scientific evidence linking hard water to actual hair loss.

Another set of questions

If hard water truly caused hair loss, it raises a number of obvious questions — questions that are rarely addressed when people casually repeat this narrative:

If bad water damages hair, why does male pattern baldness always follow a specific pattern — receding from the temples, thinning at the crown, and progressing in defined stages? If water were the cause, wouldn’t the entire scalp be affected more or less equally?
In a 2019 survey, Bangalore's water quality was found to be significantly better than New Delhi’s, yet Bangalore gets most of the blame. Why is that? Shouldn't cities with worse water quality have even more severe hair loss cases?
Why does the issue predominantly affect men and not women, even though both genders are exposed to the same water?

So why the myth?

Like most myths, it’s hard to pinpoint its exact origin — but I have two theories.

First, male pattern baldness typically begins in the mid to late 20s, which just so happens to be the age when many people migrate to Bangalore. While I don't have hard statistics, Bangalore is a major tech hub and has a large number of senior/lead positions. People often are attracted to Bangalore because they have hit the ceiling elsewhere.

Now, if hair loss begins after someone moves to Bangalore, it’s tempting to draw a cause-effect relationship. Once water was casually blamed, the idea spread like a viral meme — and over time, it solidified into a definitive fact. Many large residential complexes in Bangalore also rely on borewell or groundwater, which makes people suspicious of the water quality — which might contribute to the myth.

My second theory (though less likely) is that young people living away from home might neglect nutrition, which can play a role in hair health. Lifestyle changes, stress, irregular meals — all of that can add up.

Most importantly, I think people underestimate how common hair loss is. Several studies show that about 30% of men start experiencing male pattern baldness by age 30, and by 50, that number rises to around 50%.

Don’t blame the water. Blame your genetics.

ChatGPT is the Teacher I Always Wanted

Tue, 01 Apr 2025 11:16:59 GMT

In 2012, I asked a question on Server Fault: How do servers handle thousands of connections per second when a regular home internet connection would be too slow for that? I had just set up Apache on my home IP and was genuinely curious about large-scale servers. This was the response I got :

Hm, I hate to say it, but it’s not a very SMART question.

It’s like asking, “I have a car, but when I want to transport a lot of people it’s too small. How do big companies do it?” — they don’t use a car.

Similarly, web hosts DON’T USE a slow home connection. Isn’t that obvious?

They have data centers with massive backhaul — multiple 10-gigabit connections. That’s how.

Looking back, maybe it wasn’t the smartest question — but I was 18, and genuinely curious about what kind of internet infrastructure powers large-scale services.

I have always been a nerd about too many things. When something catches my interest, it consumes me, and my brain fills up with a flood of questions. Google and Internet have always helped in chasing down the answers, but it's not always enough. That's because, as an amateur, it's hard to piece bits and pieces of information scattered around.

Fast forward to today, I have access to something remarkable: a chat interface that lets me ask anything, and without judgment or condescension, it will always answer. Here are a few questions I’ve asked recently:

In 1760, expeditions were sent to observe the transit of Venus to measure the Earth-Sun distance. Were they successful?

From the universe’s frame of reference, can we say we’re traveling at the speed of light, since the space is expanding at that rate?

What’s the hard speed limit a spaceship could reach before it would destroy itself?

The LHC already pushes particles to 99.99999% of light speed — why do we need an even larger collider?

Can the chemical reactions that led to life happen again today? If not, why?

Every answer has been spot on. While it’s become common to say that AI is making us lazy or dumber, I’ve found the opposite to be true. Having instant access to answers has made me more knowledgeable and curious. I’ve found myself far more immersed in scientific questions now than I was just a few years ago.

A big part of it is also the non-judgmental nature. People are often afraid of sounding stupid, so they hold back from asking the questions that actually matter to them. But with a machine, that fear just isn’t there.

ChatGPT feels like the teacher I always wished for — someone who’s endlessly patient, deeply knowledgeable, and never makes you feel small for asking anything. It answers without ego and never gets tired of follow-up questions.

And for someone with a curious mind, that’s not just convenient, it's an incredible gift.

WiFi Passwords Should Be Dumb

Thu, 13 Mar 2025 11:16:59 GMT

Why do people complicate WiFi passwords to the point where they either have to look it up themselves or recite them like instructions—'Capital A, everything small,' or 'At the rate, 54321, dollar sign'?" It's a password that you have to share often and input frequently, so it seems foolish to make it complex.

Your WiFi is not a bank account. There's little incentive for anyone to hack into yours (except maybe to get free internet). If you're worried someone can use your Internet to do something nefarious, remember that there are far less risky ways to do shady stuff (VPNs, residential proxies) as compared to using something that requires close proximity.

Even for a determined attacker, hacking the WiFi password is extremely difficult—unless you use a very commonly used password.

A possible attack vector on WiFi (WPA2) is capturing the 4-way handshake — the short exchange that happens when a device connects to the network. Tools like aircrack-ng can sniff this handshake packet from the air, without needing to be part of the network.

This handshake packet contains data that includes a PBKDF2 hash, which is derived from the Wi-Fi password and the network name (SSID). Once an attacker has this hash, they can try millions (or even billions) of passwords offline, without needing to interact with the network again.

Here's the catch though: PBKDF2 is a computationally expensive function.

PBKDF2 takes your password and runs it through a hashing algorithm — in the case of WPA2, SHA-1 — thousands of times. With WPA2, that number is fixed at 4096 iterations. So, just to check one password, the computer must run SHA-1 4096 times. Even with a powerful system, that would only amount to testing a few hundred to a thousand hashes per second.

The effort/reward ratio is simply terrible when it comes to brute-forcing. Unless you're using a super-commonly used password (e.g, 12345678), I wouldn't worry about anyone hacking your WiFi.

Last of all, is the password even a secret? Most people don't have any qualms about handing it out to anyone who comes to the house. How many people use something like guest networks to restrict access or check access logs of their router?

Why not make them easy?

WiFi passwords should be easy, dumb, and pronounceable. Something you can remember and say easily, and the other person gets it without confusion. Two easy words. A single phrase. A single long word.

"recycle bin"
"writing legend"
"air force"
"common sense."
"hexadecimal"
"house is blue"

And that will be enough! Security is important—but Wi-Fi passwords don’t need the same level of security as your online accounts.

The Audience Matters

Tue, 07 Jan 2025 12:50:02 GMT

There's an interesting story behind making DDLJ (Dilwale Dulhania Le Jayenge).For those who don’t know, it’s one of the most successful Bollywood films ever, with the longest theatrical run in history. I always considered it to be a run-of-the-mill crowd-pleaser movie. I am not being harsh here. Aditya Chopra, the director himself, admits:

My film is not different. I’m making the most commercial, clichéd, tapori (pedestrian) movie. I’m making the oldest story in the world.

What's interesting about the movie is how its success was no accident. It was born from nearly a decade of research to understand the Indian audience

He (Aditya Chopra) saw every movie in the theatre FDFS (First Day, First Show) since he was 12 - just to study almost clinically what the audience thought, where they laughed, where they left. He tracked the BO numbers. He did the analysis, put in the hard work, and studied the business of cinema quite literally. He used those insights to inform his filmmaking process and predict what’s next. He understands the psychology of people in general, of Indian people very deeply, and what connects with them and what doesn’t.

This approach is quite different from the creative projects I’ve always admired—the "Don't care about the audience" or "Ship something you want to see" kind of projects. The movies, shows, and video games I love have all been driven by a passionate vision to make something great and not how the audience is going to react.

Aditya flipped this approach on its head. He wanted to understand what makes the Indian audience tick before he committed to a project. Make no mistake, he was still creative—the locations, dialogues, songs were all well-thought. However, he worked within the bounds of what he believed would eventually succeed.

Caring About Something That Works

Rick Rubin is a big advocate of letting your creative spirit do the work. You care and everything else will fall in place. In his book The Creative Act, he says:

In terms of priority, inspiration comes first. You come next. The audience comes last.

All that matters is that you are making something you love, to the best of your ability, here and now.

It’s a beautiful idea. But here’s the problem. Sooner or later, you need to create something the audience cares about. That's because, without their validation, you’ll struggle to make money and sustain what you love.

Sure, you can ignore the audience and create only for yourself. There are enough success stories to make free-spirited creativity sound like a tempting idea. However, there are just as many failures. The basic fact can't be ignored: to have a shot at making money, you need to build something people are willing to pay for. You do need to care about the audience.

The other approach is to figure out what is working in the market, and create something with that in mind. The most successful indie-hackers I’ve seen do this very well. LevelsIO, for example, has been pretty amazing in spotting trends and building something valuable around them. When Generative AI was just taking off, he created Interior.ai, and it quickly succeeded in becoming a $50k MRR side project.

Danny Postma entire approach to product-building centers around researching what people are already searching for via SEO tools, selecting the ones with money-making potential, then slowly building your website to rank well for those keywords. He, too, has a long track record of successful projects.

Of course, you still need creativity. It's not easy to sell junk. But by focusing your creativity on something with a better chance of success—something people already want—you can drastically improve your odds at making money. You treat creativity as a tool, not the goal.

Maybe You're Not Sick of Programming

Thu, 27 Jun 2024 11:16:59 GMT

Time and again, I encounter posts, people, who believe that they are sick of programming and can't wait to retire to do something worthwhile (farming, traveling, for eg). Maybe they are and programming was just an ill-suited career, but higher chance that they're not actually sick of programming, but fed up with:

Bureaucratical bullshit to get simple things done
Lack of product vision where everyone from sales to marketing to support is driving it.
Lack of ownership in product; the inability to effect a positive change.
Pointless meetings and discussions.
Process, procedures, having the upper-hand, and getting things done taking a back-seat.

Or maybe, they just need a break and do something different. Even renowned artists, writers hit a block and must work hard to replenish their motivation.

Speaking from personal experience, the burnout stage can skew our perspective unfavorably. It can make us overlook programming's immense potential for creative fulfillment and underestimate hurdles in other pursuits. Farming, for instance, while enjoyable for a few weeks, can be incredibly taxing (both mentally and physically) in the long run. Perhaps, it would serve anyone better to fix why they're failing to realize the creative potential of programming vs. throwing it all away.

Voice is (mostly) Bad UI

Tue, 16 Apr 2024 14:16:59 GMT

Humane's AI Pin is in the news and unsurprisingly, it's been universally panned by reviewers. I don't have anything to add there. I appreciate the courage to invent a new paradigm, but there are no excuses for shipping something so pointless after raising hundreds of millions.

What I want to discuss is voice as an interface.

Movies like Her have led us to believe that voice is the ultimate user interface— why use a keyboard or mouse when you can just talk things out? Just a few days ago, I heard about a talk claiming that it's only a matter of time before every app becomes a voice interface that uses AI agents to accomplish tasks.

I can’t shake the feeling that these ideas reflect a shallow understanding of how the human mind actually works.

Of course, voice as input is nowhere near as polished today, but let's assume it does become so in the near future. Is Voice poised to take over as the universal interface? I don't believe that for many reasons.

Voice is inherently incapable of representing abstract thoughts: When we use tools, we rely heavily on our subconscious and abstract thinking — especially in activities like writing or research. These tasks tap into deeper, often nonverbal processes. I wouldn’t be able to finish an article if I had to narrate the whole thing aloud. That’s because, while I occasionally have fully formed sentences in mind, more often I’m working with fragments — scattered thoughts that need time and space to come together. I have to search for the right words, connect the right ideas, and track down the right information — all of which unfolds more naturally through typing and thinking, not talking.

Voice exercises are part of the brain that belongs to the 'consciousness' category, which limits these capabilities.
Voice makes quite a few things harder not easier: Theoretically, saying, "order an Uber to airport" seems like the easiest way to accomplish the task. But is it? What kind of Uber? UberXL, UberGo? There's a 1.5x surge pricing. Acceptable? Is the pickup point correct? What would be easier, resolving each of those queries through a computer asking questions, or taking a quick look yourself on the app?

Another example is food ordering. What would you prefer, going through the menu from tens of restaurants yourself or constantly nudging the AI for the desired option? Technological improvement can only help so much here since users themselves don't clearly know what they want.
Voice is privacy invading: Most people are naturally self-conscious and don’t like the idea of being seen talking to a computer in public. Even around friends or a partner, it can feel awkward. And it’s not just about making suspicious or unusual requests — people often worry about being judged for completely normal things. Take checking messages, for example. That alone could raise silent questions like, “Why does he need to check so often?” or “Doesn’t he have anything better to do?” The fear of judgment, even over something trivial, can be surprisingly strong.
UI is preferable to understand the limits of software: The UI isn’t a limitation — it’s what helps us understand what a piece of software can actually do. Take Google Maps, for example. If you want a route that avoids narrow or poorly maintained roads, or one that takes you through scenic areas, it likely can’t help — not because of the UI, but because it doesn’t have that kind of data. On the other hand, without the UI, you might never realize you can check traffic conditions between two locations for a specific departure time. The interface reveals the software’s capabilities — and sometimes, its limitations.
Users don't like talking all the time: Using voice for simple tasks like setting a timer or alarm is fine, but relying on it for everything can quickly become frustrating. We often don’t realize how many micro-tasks we breeze through each day thanks to familiar interfaces — checking the calendar, reading emails, browsing messages, or just scrolling the web. Sure, some of that is about passing the time, but even that — the act of killing time — is a deeply ingrained part of our culture, and it’s not going anywhere.
Voice can't resolve real-world ambiguities: Voice-only interfaces can easily lead to misunderstandings. For example, you might say, "Sell 500 shares," thinking that’s 50% of your holdings. But in reality, 500 shares represent your entire holding. The system won’t know your intent—it will just do what it’s told, even if your instruction was based on a wrong assumption.

Of course, you can make the same mistake on UI, but it would be more difficult since there's visual feedback about your action. An audio confirmation prompt won't be that effective as it's not easy to visualize the numbers in your head.

Overall, I'm not convinced that voice is the future. There are too many fundamental challenges with not having a screen to interact with. And even when you add a screen, voice often complicates tasks that could be done more efficiently with a well-designed UI. Even if AI could perfectly understand human speech with zero latency, I still don’t think we’d arrive at the kind of future companies like Humane are imagining.