Bobby Yan's Blog RSS Feed

How One Slow Car Can Jam a Six-Lane Freeway

Thu, 07 May 2026 00:00:00 GMT

You know the feeling.

Traffic has been crawling for miles. You inch forward, tap the brake, inch forward again, and start imagining the disaster waiting up ahead.

A crash? Construction? A mattress in lane three?

Then the jam simply ends.

No wreck. No cones. No police cruiser. Just open pavement and the dawning suspicion that the whole ordeal was caused by something stupid.

Often, it was.

A few cars were driving side by side in the passing lanes, moving just slowly enough that nobody could get around them.

A rolling wall.

If you have ever reached the front of “traffic” and found only that, you are not alone. And while it is satisfying to blame the driver who appointed themselves pace car for the entire freeway, the more interesting question is how such a small obstruction can produce such a large jam.

One slow car in the passing lane does not block six lanes by itself.

It does not need to.

All it has to do is sit in the wrong place at the wrong time. From there, the freeway does the rest.

When traffic is light, the road forgives almost everything. Faster cars go around. Gaps open and close. The system absorbs the annoyance.

But when traffic is already dense, gaps are the whole game. A car stuck behind the slow driver moves right. The driver in that lane taps the brakes. That brake tap shrinks another gap. Another driver moves right. Someone else brakes. The little problem in one lane starts leaking sideways.

The slow car is the match. The lane changes are the gasoline.

The Mechanism

The basic story of freeway flow can be written as a simple equation:

$$ q = k v $$

where $q$ is flow, $k$ is density, and $v$ is speed.

Flow equals density times speed.

That is the clean version. The messier version is that real traffic is made of humans who follow too closely, hesitate, merge late, check their phones, and sometimes decide that the far-left lane is a wonderful place to contemplate the speed limit.

Still, the simple model is useful because it shows how little slack a busy freeway actually has.

A lane has a rough capacity based on headway, which is the time gap between one car and the next:

$$ C_{\text{lane}} \approx \frac{3600}{h} $$

where $h$ is average time headway in seconds.

If cars leave about two seconds between each other, one lane can carry roughly:

$$ \frac{3600}{2} = 1800 \text{ vehicles per hour} $$

Six lanes gives you about:

$$ 6 \times 1800 = 10{,}800 \text{ vehicles per hour} $$

Now put 10,000 vehicles per hour on that freeway.

On paper, you still have room. In reality, you have only about 800 vehicles per hour of slack, which is not much when every lane change requires someone else to make space.

Do not treat these numbers like a design manual. Real capacity depends on ramps, grades, weather, trucks, pavement, sight lines, driver behavior, and a dozen other things. The point is the margin. When demand is already close to capacity, a disturbance does not need to be large to erase the remaining slack.

A slow car in the passing lane doesn't have to delete a whole lane.

It only has to trigger enough braking to burn the cushion.

How One Lane Infects the Others

Start with one slow car in the left lane.

Cars stack up behind it. Some wait. Some brake. Some look right.

One car moves into lane two.

If the gap is large, nothing much happens. If the gap is tight, the driver behind the merging car tap the brakes. That brake tap travels backward.

Now lane two is less smooth. Drivers in lane two see brake lights and shrinking gaps. Some move into lane three.

The same thing happens there. Then lane four. Then lane five.

The original slow car never blocked those lanes. The passing demand did.

That is the non-obvious part. A left-lane camper is not just a slow object. It is a generator of lane changes.

At low density, spare gaps hide the damage.

At high density, every lane change makes someone else react.

This is why a freeway can feel fine one minute and doomed the next. Nothing dramatic has to happen. No one has to crash. Traffic only has to be pushed slightly past the point where drivers can adjust smoothly.

Once that happens, the road begins manufacturing its own problem.

Two Blockers Are Much Worse

One slow car in the passing lane is bad.

Two slow cars side by side are a moving wall with license plates.

With one blocker, faster traffic still has an escape path, even if it is ugly. With two, the escape path gets worse. More drivers stack up. More drivers brake. More drivers hunt for gaps farther right. The disturbance spreads faster because the freeway has fewer ways to route around it.

The road may still have six painted lanes, but it no longer has six useful lanes.

That distinction matters.

Usable capacity is not just a question of asphalt. It depends on speed, spacing, predictability, and safe gaps. A few vehicles in the wrong place can damage all four at once.

This is the central point: left-lane camping is not merely rude. In dense traffic, it can turn a wide road into a narrow doorway.

A Toy Model

The model above is simple on purpose.

Cars have different target speeds and following distances. They react to traffic ahead. They change lanes when boxed in. They slow down for tight merges. A hard minimum gap keeps them from overlapping.

The blue angled cars are changing lanes.

The red bars show cars that have hit the minimum-gap constraint and must slow down.

Try it in free flow first. The cars mostly sort themselves out.

Then click 1 blocker.

A queue forms behind the slow car. Some drivers try to escape right. Those lane changes consume gaps. The next lane starts braking. Braking creates more lane changes.

If you wait long enough, you will see that little slowdown turn into something close to a standstill.

Click 2 blockers and it gets uglier.

The passing path narrows. The queue forms faster. The disturbance spreads farther across the road.

This is not a calibrated model of any specific freeway. It ignores ramp geometry, weather, crashes, pavement, aggressive drivers, trucks, and plenty of other variables.

That is fine. The claim here is narrower.

Even in a toy model, the bottleneck appears when traffic is near capacity.

Below capacity, the road absorbs the obstruction.

Near capacity, a small rolling bottleneck can create a large jam behind it.

What This Is Not Saying

This is not an argument for unlimited speed in the left lane.

It is not an argument against speed enforcement.

It is not saying every slow car jams the freeway.

It is not about people briefly passing, preparing for an exit, avoiding a hazard, or making room for emergency vehicles.

Those caveats matter because the left-lane conversation can quickly become a referendum on whether other people are driving exactly the way you want them to. That is not the useful version of the argument.

The useful version is simpler: when a road is crowded, unnecessary surprises are expensive.

A driver who sits in the passing lane while traffic stacks up behind them creates surprises. Drivers behind them must decide whether to wait, brake, tailgate, or pass on the right. Drivers to the right must react to sudden merges. Everyone gets fewer good choices.

A predictable road is a safer road.

A road full of forced decisions is not.

Slow Is Not Automatically Safe

The usual defense is:

"I am going the speed limit."

A driver sitting in the left lane at the speed limit while traffic stacks up behind them may think they are making the road safer.

They are doing the complete opposite.

They increase speed variance.

They encourage right-side passing.

They force more lane changes.

They create more braking.

They make the roads less predictable.

None of that is a safety win.

The safety benefit of lane discipline is predictability. Faster traffic knows where to pass. Slower traffic knows where to be. Everyone has fewer surprises to process.

Pass on the left.

Move right when done.

Let the road sort itself with the fewest possible forced decisions.

California Already Has the Idea

California already has a version of this rule.

California Vehicle Code 21654 says that a vehicle moving slower than the normal speed of traffic should use the right-hand lane, except when passing or preparing for a left turn.

The California DMV handbook also describes the far-left lane on a multilane road as the passing lane. It tells drivers to use the left lane to pass or turn left, and to avoid weaving.

That does not mean “speeding is fine.”

It means speed enforcement and lane discipline solve different problems.

Reckless speed is dangerous.

So is turning the passing lane into a slow-moving roadblock.

Both ideas point at the same target: keep traffic predictable.

Enforcement Should Be Boring

The rule doesn't need to be complicated.

Keep right except to pass.

Enforcement does not need to read minds. The only behavior it needs to look for is this:

A vehicle has a lane on its right. There is at least one car behind it. And there is no apparent reason it cannot move right.

CHP already has a traffic complaint process for hazardous traffic conditions and traffic-related issues. Treating left-lane camping as a traffic problem is not about rewarding impatience. It is about removing unnecessary conflict from a system that depends on smooth gaps.

The goal is predictability:

Fewer surprise lane changes.

Fewer surprise right-side passes.

Fewer surprise brake taps.

Fewer six-lane freeways reduced to one-lane problems.

Thoughts on Tesla FSD 14.3

Sat, 11 Apr 2026 00:00:00 GMT

Some context

We bought a Tesla for one reason: FSD.

If all we wanted was an EV, there are other cars I would much rather have. I mean, the Taycan Cross Turismo is just stupidly gorgeous. But that wasn't the point. We wanted self-driving.

Part of what makes that funny is that I actually love driving. I don't just mean I like cars. I mean I genuinely enjoy being behind the wheel. Long drives don't bother me. Road trips don't bother me. I've done enough driving to know what good driving is like, and what bad driving is like. So for FSD to impress me, it has to clear a pretty high bar.

I've been around Teslas enough over the last few years to watch FSD go from "interesting demo" to "okay, this is becoming real." In 2021, it felt like something I would only ever trust on highways. By June 2024, it was getting somewhere, but it still felt more like a gimmick I would show friends than something I actually wanted to use around town. By February 2025, after test-driving the lineup again, I was shocked by how much better it had gotten, especially in city driving. By July 2025, it felt good enough that I actually wanted one for road trips.

As a side note, FSD on the Cybertruck felt noticeably worse than on the rest of the lineup when I tried it in February 2025. At one point it tried to back us toward a shopping cart return. At another it straight up ran a red light on a highway ramp. We joked that that was not the kind of hurry we had in mind when we tested the Hurry mode. So no, this was not some clean sweep across every Tesla.

Why 14.3 feels different

The biggest improvement, by far, is following distance.

That sounds like a boring answer. It is not.

Before 14.3, FSD had this obnoxious habit of tailgating, especially at highway speeds. Mad Max is the profile that most closely matches how I naturally drive, but older versions would still sit way too close to the car ahead at 80 to 85 mph. That didn't just feel annoying. It made the whole system feel less trustworthy. A lot of the time, supervising it was actually more tiring than just driving myself, unless I was exhausted or there was nobody in front of me.

Tesla currently has five FSD profiles: Sloth, Chill, Standard, Hurry, and Mad Max. They mostly affect how aggressively the car picks speed and changes lanes. But the spacing problem used to show up in all of them. Even in Sloth, FSD would still sit uncomfortably close if traffic ahead was moving at a "normal" highway pace. That isn't a style preference. It's just bad driving. Either pass, or stay the hell back.

In 14.3, that finally feels substantially better. The car leaves more room. It's much less tiring to supervise. And that's the part people miss if they haven't used this stuff a lot: the difference between a system that is technically capable and a system that is actually pleasant to live with often comes down to details like this.

On paper, "better following distance" sounds minor. In real life, it changes the entire experience.

Honestly, this one fix alone makes me prefer FSD over most human drivers on the road. At night, probably over me too, and I think pretty highly of myself as a driver.

What's still annoying

The stop-sign behavior is still too robotic.

I get why Tesla wants to be conservative here. I really do. But doing a full, exaggerated three-second stop at every stop sign does not match how traffic actually flows in the real world. It can confuse other drivers, irritate the person behind you, and create these weird awkward moments that don't feel any safer.

The good news is that 14.3 finally makes the accelerator-pedal override feel usable. If the car has already stopped and I want it to go, I can now gently press the pedal and have it move forward in a predictable way. Earlier versions technically let you do this too, but they often followed it with an abrupt hard brake that made the whole interaction worse. In 14.3, the override still feels like a workaround, but at least now it behaves like a sane one.

The steering-wheel twitching needs work too.

Sometimes the car gets into this weird indecisive mode where it does a few quick little left-right jitters, like it can't decide which way it wants to go. Fortunately, this has only happened to me at low speeds so far, but it is unsettling when there are other cars around.

Destination parking is another one.

FSD has gotten so good at navigating to a destination, but it still struggles with parking once it gets there. And sometimes it parks absurdly far from your actual destination when there are clearly spaces much closer. To be clear, this is pretty low on the list of things to fix. It's a minor annoyance, not a major flaw.

What really needs to go

There is one change in 14.3 that I genuinely do not understand: if you exceed 100 mph at any point during a drive, FSD gets disabled for the rest of that drive.

To be clear, I do not mean FSD refuses to engage above that speed. Obviously it already has a hard limit at 85 mph. I mean that if you go above 100 mph while driving manually, the system seems to punish you afterward, even once you're back at a speed where it would otherwise be perfectly happy to operate.

I have no idea what this is supposed to accomplish. It doesn't make the experience feel safer. It just feels arbitrary and annoying.

Sure, someone might say, well, you should never be doing over 100 mph in the first place. Let's agree to disagree.

If FSD is already taking me up to 85 mph trying to pass a slightly slower truck that's doing 83 to 86, I feel safer just taking manual control, getting the pass done quickly, and moving on, instead of sitting next to that truck forever. On a car as quick as a Model S Plaid, 85 to 100 is basically a one-second event. That would be even more true in the new Roadster, if Tesla ever ships that damn thing. And yes, I want that car to come out already. Penalizing FSD afterward for that kind of manual intervention just seems dumb.

And before someone says, "Why not just slow down and move right?" Sometimes that is not the best move either, especially if you've already committed to the pass and now have someone stacking up behind you. And to be fair, if your overtake is taking 30 seconds, they're not the impatient asshole for tailgating you. You are.

If you're going to pass, pass. Don't sit there pacing another car forever. In the real world, the safest move is often to just complete the maneuver cleanly and move on.

What I would love to see

There is one feature that would genuinely change the game: let me get out when I arrive and let the car park itself.

I pull up to the restaurant. I get out. The car goes and finds a spot on its own. When I'm done, I summon it back. Honestly, I don't mind walking to the car later if I need to.

Tesla almost certainly has all the technology to do this already. It feels more like a regulatory and liability problem than a technical one. But whatever the reason, that is the kind of feature that would put FSD so far ahead of the game that it's not even a competition anymore.

Onto a few smaller quality-of-life improvements while I'm at it:

The auto windshield wipers need a sensitivity adjustment. Most other cars have had this for years. Some people want the wipers to kick in at the first raindrop. Some people, myself included, want a lot more drops on the windshield before a wipe. Right now, Tesla's auto mode is way too eager. It wipes at the slightest hint of moisture. If you leave it in auto, you'll be replacing your wiper blades way more often than you should just to keep them working well when you actually need them. Just give me a slider. This is a solved problem.

Can we have an adjustable turn-signal volume? This one gets especially annoying with FSD, because it will happily sit there at a light with the blinker clicking the entire time while it waits for a turn. When I'm driving, I can at least choose to signal later. With FSD, I just get the full soundtrack whether I want it or not. Either lower the default volume, or let folks set it themselves. Porsche somehow figured out how to make a turn signal sound that doesn't drive me insane. Tesla can too.

Bottom line

14.3 is not the first version of FSD that impressed me.

It's the first version that actually changed how I drive.

That difference matters. Plenty of technology demos are impressive for five minutes. Very few become part of your actual life. FSD 14.3 crossed that line for me, and it mostly did so not through some flashy new trick, but by getting better at the boring stuff: spacing, smoothness, predictability, trust.

That's why this version feels different. It didn't just get more impressive. It got easier to live with.

Life-changing Things

Fri, 10 Apr 2026 00:00:00 GMT

This is a growing collection of things I've personally found to be life-changing and some of my thoughts on them. They are arranged mostly in the order of wow factor, not importance or usefulness.

ChatGPT

Generative AI has completely changed the way I operate. I don't even think twice about asking it for a second opinion, a simple question, or a quick sanity check anymore. It's nothing short of amazing.

A recent example: we went on a hike with Zuki, our dog, and two mornings later, he started peeing what looked like blood. Needless to say, we panicked and took him to the ER. Long story short, he was diagnosed with primary IMHA, or immune-mediated hemolytic anemia. Before we knew it, we were talking to doctors who were throwing around words like hemolysis, PCV, hematocrit, and so on. It was overwhelming fast. And when the ER gets busy, and it does sometimes, you feel bad for hammering the doctors with questions.

When you need to get up to speed on a field you know almost nothing about, ChatGPT is a godsend. Within minutes, I had enough of the basics to understand what was going on and come up with better questions to ask. Hemolysis is the breakdown of red blood cells. PCV and HCT are measures related to how much of the blood is made up of red blood cells. Bilirubin is a waste product associated with that breakdown. These are just tiny examples, and we went much deeper than this, but hopefully they give you a sense of how useful this is. Going from completely lost to meaningfully informed in minutes is incredible.

Claude Code

I love building things. The problem is that I don't code fast enough to build all the things I want to build. Claude Code has basically obliterated the time and effort between having an idea and having a working prototype.

I was already doing this style of building before people started calling it vibe coding, and way before agents became a thing. I still remember when Augment in VSCode was giving me better results than Cursor with the latest Claude models, and even that already felt like magic. Today, as of April 2026, Claude Code feels like cheating. It's hard to imagine what this will look like five years from now. Actually, scratch that, two years from now.

Tesla FSD

This is real living-in-the-future stuff. Honestly, this might be the most mind-blowing item on this list.

I love driving, and I don't just mean I like cars. I mean I genuinely enjoy the act of driving. I also consider myself a good driver, and to me that means safety first, speed when appropriate. For context, I've done same-day Bay Area to Yosemite and back trips solo, which is about six hours of driving. I've also done a Bay Area to Vegas road trip where I averaged at least eight hours of driving a day for four straight days, on top of the hikes, because we spontaneously decided to visit the Grand Canyon after already booking our Vegas hotel. Driving has always felt like second nature to me.

That's what makes Tesla FSD so impressive. Since version 14.3 especially, it has completely changed how I drive. I basically don't do normal driving anymore. I delegate at least 90% of my normal driving to FSD, and it handles it shockingly well. What's left for me is the part I actually enjoy: spirited driving, when I want it. I'm going to write a separate post just about FSD 14.3, because I have a lot more to say about why that update hit so differently, and what I think is good and bad about it. But for now I'll just say this: even as someone who genuinely loves driving, this feels life-changing.

drTung's Smart Floss

Expandable floss is already one of those things that feels weirdly life-changing once you discover it. drTung's Smart Floss is the best version we've tried, and it's not even close.

It pulls out so much more than regular floss that normal floss just doesn't do it for me anymore. This stuff ruined any other floss for me. You have to try it.

Apple Silicon Macs

I've been a Mac user my whole life. I don't see that changing anytime soon. Macs are unusually good at both ends of the spectrum: basic things are simple and intuitive, yet it's still hackable and tunable underneath. Everything just feels natural on a Mac in a way Windows PCs never quite have for me. I mean, the trackpad, oh boy, is still in a league of its own. Give me an Apple trackpad and I might give PCs a try.

What made Apple Silicon Macs special was that Apple finally matched that software experience with hardware that felt properly sorted out. The Intel years got rough. The chips ran hot, the obsession with thinness made the thermals worse, and too many of those laptops felt compromised. Remember those butterfly keyboards that broke if a speck of dust got in? Then the M1 arrived and the whole thing snapped into place: fast, cool, quiet, and absurdly efficient. These days I don't replace my laptop because it no longer meets my needs. I replace it because I just want even more speed. Apple Silicon feels like endgame laptop computing.

Roborock

Back in the early 2010s, my parents bought one of those early robot vacuums from LG. It was expensive, but it didn't clean well. We used it maybe once or twice and gave up. That pretty much wrote off the entire category for me.

Fast forward to today, and Roborock is a completely different story. I'm sort of a clean freak, and these things are shockingly good. The game changer isn't just that they mop. It's that they mop well. That distinction matters. Vacuuming is nice. A robot that actually leaves the floor feeling properly clean is a different level. It's one of those products that quietly removes a huge amount of life friction.

Fiberglass Reinforced Tape

Not enough people know about fiberglass reinforced tape. I'm serious. If you've ever moved or packed heavy boxes, you know regular packing tape is terrible. The 3M or Scotch stuff is not even decent. Layering doesn't help because if the bottom layer doesn't stick well, the whole thing starts coming loose anyway.

Fiberglass reinforced tape is a whole different beast. I can wrap it around a 50- or 70-pound box once and literally lift the whole thing by the tape. That sounds absurd until you try it. I just don't see anyone going back to regular packing tape after this.

Practical Idealism

Sat, 05 Aug 2023 00:00:00 GMT

People love to split the world into dreamers and doers, like they're two different species. I think that's mostly a story both sides tell themselves to justify not doing the harder thing.

The harder thing is holding both at once.

I've been around enough "visionaries" to know the type. They're great at dinner. They'll tell you about the company they're going to start, the book they're going to write, the movement they're going to lead. Ask them what they actually did this week and the answer is always a little thin. They have the destination. They don't have a Tuesday.

The pragmatists aren't much better. They pride themselves on being "realistic," which in practice usually means they've quietly pre-negotiated a smaller life and want you to join them. Their imaginations died somewhere around 25 and they've been calling that wisdom ever since. I don't think of them as villains. I think of them as people who mistook a ceiling for the sky.

The people I actually care about are the ones who manage to be both. And I used to think that was a rare talent. Now I think it's more like a habit — a specific way of holding tension that you can train.

Here's the version I've ended up at:

The dream protects you from the grind, and the grind protects you from the dream. If all you have is the vision, you burn out the first time reality pushes back. If all you have is the checklist, you drift, because a checklist can never tell you whether the thing on it is worth doing. You need both, and you need to let each one correct the other, on a loop, forever. The vision pulls. Reality files down the edges of the vision. Then the vision pulls again from a slightly smarter place.

This is the part in the essay where I'm supposed to trot out Steve Jobs and Elon Musk and the Wright brothers. I'm going to skip it. Not because those examples are wrong, but because reaching for them is a tell. It's what you do when you want the idea to sound important without having to defend it yourself. I'd rather just defend it.

The annoying part — the part nobody likes to hear — is that there's no clever trick for any of this. You just have to sit in the discomfort of wanting something much bigger than what today allows, and then doing today anyway. Most people can't stand that feeling for very long, so they pick a side. They either shrink the dream until the Tuesdays feel reasonable, or they let the dream stay enormous and never get to Tuesday at all. Both moves are really just ways of turning the volume down.

The people who end up building anything are the ones who refuse to turn the volume down.

A few things I try to keep in mind, for what it's worth.

If your goal doesn't sound a little delusional when you say it out loud, you're aiming at something you already know how to do. That isn't ambition. That's a chore with extra steps.

If you can't describe what you're going to do about it on Monday morning, the goal isn't real yet. It's still a mood.

And purists on both sides will tell you that the middle is for cowards. They're wrong. The middle is where anything actually gets built. The poets will call you a sellout, the cynics will call you naive, and somewhere in between, one of you is quietly shipping.

Honestly, most of the things I'm proud of in my own life didn't start as plans. They started as half-formed obsessions I couldn't let go of, and at some point I got embarrassed enough about how little I was doing about them to just start. Not a strategy. Not a five-year plan. A small, ugly first step I'd be willing to show a friend.

If you're waiting for the version of yourself who has it all figured out, don't bother. That person is a ghost. The only version of you who ever gets to build the thing is the one sitting there right now, slightly underprepared and slightly embarrassed, doing the Tuesday.

How to Work Hard

Thu, 15 Jun 2023 00:00:00 GMT

A lot of people have a very strange relationship with hard work.

They think if it feels awful, it must be good for them. If they're stressed, behind, sleep-deprived, and vaguely miserable, they assume they're doing something important. Add a packed calendar and a permanently green Slack dot and now it really looks like work.

Maybe even impressive work.

But if you've been around enough smart people, you eventually notice something awkward: the people who look the busiest are often not the ones getting the most done.

You've probably seen this type. Always swamped. Always apologizing for the delay. Always saying things like "this week has just been insane." They work late, they work weekends, they work through lunch, and somehow the output is still... fine. Not terrible. Not amazing. Just oddly average for all that visible effort.

Then there's the other kind of person. Much calmer. Much quieter. No theatrical suffering. No big production. They disappear for a while and come back with something sharp, finished, and high-leverage. What took everyone else a week of noise took them an afternoon of actual concentration.

That's when it hits you: a lot of what gets mistaken for hard work is just friction.

And friction is not the same thing as effort.

There is a version of work where you're basically dragging yourself across the floor. You keep checking your phone. You open the document, then somehow end up in your inbox, then on some unrelated tab, then back in the document, then suddenly you're making coffee like the coffee was the real assignment all along. Every small step requires negotiation. Starting feels heavy. Continuing feels heavier.

You can get things done this way. People do it all the time. But it is expensive. It burns a ridiculous amount of energy for the amount of progress it produces.

Then there's the other mode.

You're still working hard, sometimes very hard, but the energy actually transfers. One thought leads to another. The problem starts opening up. Your attention locks in. You stop needing to whip yourself every five minutes just to stay on task. Hours pass, and instead of feeling depleted in that gross, scrambled way, you feel used well.

That is also hard work.

In fact, I think that's the real thing.

Hard work is not the performance of strain. It's not looking exhausted in public. It's not turning your life into a stress cosplay. It's sustained, honest engagement with something difficult.

It is hard to think clearly for long periods of time. It is hard to keep pushing after the first draft disappoints you. It is hard to stay with a problem after the novelty wears off. It is hard to tell yourself the truth about what is and isn't working. It is hard to make something good.

That kind of effort counts whether or not it looks dramatic from the outside.

I think a lot of people secretly distrust work that feels fluid. If it feels smooth, they think they must be cheating. If they enjoy it, maybe it doesn't count. If they get into a groove, they start wondering whether they're being "disciplined enough," which is insane if you think about it for more than five seconds.

If the work is flowing, that's not a character flaw. That's the goal.

The point is not to become better at suffering. The point is to become better at working.

Those are different projects.

Some of the worst work habits come from confusing the two. People pick work they feel no connection to, define it in the vaguest possible way, let themselves get interrupted every six minutes, sleep like garbage, and then act surprised when every task feels like pushing a refrigerator uphill.

Of course it feels bad. You set it up badly.

A lot of productivity advice is weirdly moralistic, as if the solution is just to become a more obedient little machine. I don't buy that. Your mind is not a forklift. It matters whether you care. It matters whether the task is clear. It matters whether you have momentum. It matters whether your brain has been blasted to dust by notifications all day.

Interest matters more than people like admitting. You do not need to be intoxicated with passion. You do not need to hear the voice of destiny. But you usually need some kind of spark. Curiosity. Ambition. Pride. The desire to get good. Something.

If the work is completely dead to you, everything gets harder. Discipline can paper over that for a while, but it is a terrible long-term fuel source.

Clarity matters too. Vague tasks are where motivation goes to die.

"Work on startup." "Write paper." "Fix life."

Your brain hears that and immediately develops a sudden interest in cleaning the kitchen.

The people who work well are often doing something very unglamorous: they make the next step stupidly clear. Not "build the company." More like "draft pricing page," "email these two people," or "figure out why this query is slow." That's not small thinking. That's how real momentum works.

And then there's rhythm, which is deeply underrated.

Some people rely on emotional drama to get moving. They avoid, delay, stress out, and finally hit the point where panic becomes rocket fuel. This works just well enough to trap people for years. But it's a terrible system. You're basically borrowing energy from your future sanity.

Rhythm is less exciting and much more powerful.

Sit down at roughly the same time. Work in a way your brain recognizes. Protect the beginning of the session. Don't spend the first twenty minutes nibbling at texts and email and then complain that you "just couldn't get into it." You never gave focus a chance to show up.

That early window matters a lot. Once your attention fractures, everything feels heavier than it should.

Another thing that helps: don't stop when you're fully lost. Stop while you still have the thread.

Leave the next move in plain sight. Write yourself a note. Leave the next paragraph half-started. Keep the engine warm for tomorrow. One of the dumbest things people do is end a session in total cognitive fog and then expect their future self to restart gracefully from nothing.

That is an unfair handoff.

Feedback matters too. Work gets much easier to stay inside when it talks back. If your loop is too long, motivation leaks out. That's why some kinds of work feel naturally engaging and others feel like shouting into space. The fix is to create shorter loops on purpose. Test sooner. Share drafts earlier. Break things into pieces small enough that progress becomes visible.

Visible progress is energizing. Invisible progress is where souls go to die.

And then, of course, there is the thing people keep pretending isn't real: rest.

If your brain is cooked, your work will feel cooked. That's not weakness. That's just how brains work. Sleep matters. Walking matters. Exercise matters. Real breaks matter. Not fake breaks where you poison your attention with six apps and come back even more fragmented than before.

Some of the best ideas arrive when you're not squeezing them to death.

The funny thing is, once you start working this way, work can stop feeling miserable all the time. It can even feel good. Not easy in the sense of trivial. Easy in the sense that the energy is connecting. The machine is not grinding itself apart. You are doing something difficult in a way that makes sense.

That is what most people miss.

They think hard work should look like self-punishment. They think the badge is the suffering itself. So they end up optimizing for the feeling of effort instead of the result of it.

But those are not the same.

If your work always feels like dread, friction, and inner warfare, that does not automatically mean you are noble. It may just mean your system is bad.

Real hard work is not about making everything feel heavy.

It's about building a way of working where your attention, energy, and ambition all point at the same thing. When that happens, the work can look calm from the outside. It can even feel natural while you're doing it.

That doesn't make it less real.

Usually it means you've finally stopped confusing hard work with working badly.

How to Survive Bay Area Driving

Tue, 17 Jan 2023 00:00:00 GMT

I drive a lot. I like driving. I've done enough of it in enough places to know what good driving looks like. And I'll just say it plainly: Bay Area driving is bad.

Not because everyone here is awful. Plenty of people are perfectly fine drivers. The problem is the spread. On the same stretch of road, you can have one person driving like a cautious adult, one person drifting around like they're still processing what a lane is, and one person cutting across a gore point because their exit apparently came to them in a vision. And of course, there's the lunatic who thinks too highly of their car and their own reflexes, weaving through traffic like it's a video game.

That mix is what makes Bay Area driving exhausting.

The roads themselves do not help. Potholes, faded lane markings, random dark stretches where street lighting seems optional.

But the roads aren't the real problem. The real problem is who else is on them.

A rough field guide

There are basically two kinds of bad drivers. The ones who don't know how, and the ones who do but choose not to.

The first group splits into the incompetent and the inexperienced.

The incompetent are the genuinely alarming ones. They make moves that suggest they do not understand spacing, speed, right-of-way, or sometimes basic geometry. The inexperienced are different. They are not good yet, but they are often at least trying. They hesitate, brake awkwardly, miss chances, and generally move through traffic with the energy of someone taking a final exam they did not study for. That can still be annoying, but it is usually more forgivable and more predictable.

The second group splits into the reckless and the assholes.

The reckless are overconfident and impatient. They weave, tailgate, dive into tiny gaps, and act like every commute is a qualifying lap. The assholes are a little different. They may be competent enough behind the wheel, but they the road like a status contest. You signal, they speed up. You leave space, they take it. You make a reasonable merge, they act like you've personally insulted their bloodline.

And then, thankfully, there are the good drivers. Competent, aware, patient, predictable. The ones who know how to keep traffic moving without turning every lane change into a hostage situation.

Everybody starts in the incompetent bucket, because everybody sucks at driving on day one. Most people put in enough miles to graduate into inexperienced-but-safe, and from there they drift toward either the good column or the bad one. Most of us live somewhere in the middle, nobody's a saint, and even a mostly-good driver will occasionally do something boneheaded.We're human.

What bothers me

Here's the part that actually bothers me: good drivers get punished for being good.

Leave a safe following distance. Someone cuts into it.

Signal early before a lane change. The guy next to you takes it as a cue to close the gap.

Try not to pass on the right. The guy camping in the left lane thinks doing 5 over the limit buys him a permanent lease.

The reckless drivers are basically extracting value from everyone who's trying to play nice. It's a tiny, stupid version of the paradox of tolerance.

The twist is, in order to stay safe, good drivers end up doing things that in a saner world only the jerks would have to do. They stop leaving a cushion. They signal late on purpose, so nobody has time to close the gap. They pass on the right. None of them actually want to drive like that. They do it because the alternative is getting cut off, getting boxed out of every merge, or getting stuck behind a slowpoke forever. That is not them turning into bad drivers. That's them adapting. And it's a vicious cycle, because every good driver who starts driving a little less nicely makes the road feel a little more hostile to the next good driver.

What to do out there

STAY RIGHT UNLESS YOU ARE PASSING. If there is an open lane to your right, you should be in it. The only real exceptions: you are about to turn left soon, or slower traffic is merging in from the right. Otherwise, if you are sitting in the left lane with an open lane to your right, you are making the road worse for everyone else, and you are part of the problem.

Let others in when it is safe. If all it takes is easing off the gas to make room for somebody who actually signaled, do it. If traffic has already stopped and a car is trying to merge in, let them in. Almost zero cost to you. Pay it forward.

Know and follow the right-of-way. Please. FFS. It really isn't that hard.

Don't ever make somebody else slam on their brakes. If your maneuver requires another driver to stomp the pedal, it's not a good move. This should be obvious. It is somehow not obvious enough.

Turn off your high beams. Modern headlights are already bright enough to interrogate a prisoner. You don't need the high beams on top of that to nuke your retinas. I used to flash mine back as a gentle reminder. Turns out most people either don't know what that means or don't care. I'm seriously tempted to rig up a mirror setup and bounce the light right back at them.

tmux Cheatsheet and Shortcuts

Wed, 12 Sep 2018 00:00:00 GMT

Basics

Start an unnamed session: tmux
Start a named session: tmux new -s <name>
Attach: tmux a #
Attach to named session: tmux a -t <name>
Detach from session: Ctrl + b (or your custom <prefix>), then d
List all sessions: tmux ls
Kill a named session: tmux kill-ses -t <name>
Kill the server: tmux kill-server

Shortcuts

Shortcuts for Sessions

<prefix>, d       # detach from session
<prefix>, s       # select from sessions
<prefix>, $       # rename session

<prefix>, (       # previous session
<prefix>, )       # next session
<prefix>, L       # last session

Shortcuts for Windows

<prefix>, c       # create new window
<prefix>, w       # select from windows
<prefix>, ,       # rename window
<prefix>, &       # kill window

<prefix>, p       # previous window
<prefix>, n       # next window
<prefix>, l       # last window
<prefix>, [0-9]   # go to [0-9]th window

Shortcuts for Panes

<prefix>, %         # vertical split
<prefix>, "         # horizontal split
<prefix>, q         # show pane numbers (when numbers are show,
                      press number to select pane)
<prefix>, x         # kill pane
<prefix>, o         # swap panes
<prefix>, {         # (Move the current pane left)
<prefix>, }         # (Move the current pane right)
<prefix>, z         # toggle pane zoom
<prefix>, <space>   # toggle b/w layouts

Useful Commands

Entering the <prefix> gets you into the command mode, then

:new-window -a          # insert new window right next to current
:swap-window -t [pos]   # move curr window

Miscellaneous

Customizing `<prefix>`

If you wish to change the prefix from Ctrl + b to something like Ctrl + a, which some people prefer, you could create a file ~/.tmux.conf with the following:

# Remap tmux Prefix
unbind C-b
set-option -g prefix C-a
bind-key C-a send-prefix

1-indexing Windows

And if you don't want the 0-indexing for your windows, since 1 is easier to press when switching, add the following as well:

# 1-index Windows
set -g base-index 1

When you are done, be sure

:source-file ~/.tmux.conf

How to Switch to App Store of a Different Country

Thu, 27 Jul 2017 00:00:00 GMT

Have you ever wondered what the top charts of the App Store in another country looks like?

See, https://itunes.apple.com/us/app/spanish-word/id1123591991 takes you to the Spanish Vocab Builder, in the App Store of your country. Now if you change it to https://itunes.apple.com/sg/app/spanish-word/id1123591991, it still takes you to the same page, because this app is available in all territories where the App Store is available.

Apple finds the app by its id, and it happens that if you give it an invalid app id, it'll actually bring you to the app store of that country. For example:

https://itunes.apple.com/us/app/spanish-word/id1234567890 takes you to the US App Store.

https://itunes.apple.com/sg/app/spanish-word/id1234567890 takes you to the Singapore App Store.

https://itunes.apple.com/ca/app/spanish-word/id1234567890 takes you to the Canada App Store.

https://itunes.apple.com/cn/app/spanish-word/id1234567890 takes you to the China App Store.

https://itunes.apple.com/de/app/spanish-word/id1234567890 takes you to the Germany App Store.

https://itunes.apple.com/jp/app/spanish-word/id1234567890 takes you to the Japan App Store.

Note: If you are trying to actually download anything from the App Store of a different region, you'll have to change the country of your Apple ID.

How to Preview Files Online without Downloading Them

Sun, 12 Jun 2016 00:00:00 GMT

Want a quick and easy way to view documents in your browser without having to download them to your computer? Just want to preview some .ppt slides but don't have PowerPoint installed?

Here's how.

A scenario

Let's say you are just casually doing some googling, and you come across a file you want to preview. But my oh my, it's a .ppt file! What in the world are you going to do with that? You don't have PowerPoint installed, and even if you do, you don't want to download the file to your computer just to preview it.

Here's a link to a powerpoint slide on information retrieval I've found:

https://web.stanford.edu/class/cs276/handouts/lecture14-learning-ranking.ppt

If you try clicking on the link, you'll be prompted to download the file, which is what we're trying to avoid.

The solution

It's a simple tweak. Just add https://docs.google.com/viewer?url= in front of the link to whatever file you want to preview, and you're good to go!

For our information retrieval slide example, the link would be:

https://docs.google.com/viewer?url=https://web.stanford.edu/class/cs276/handouts/lecture14-learning-ranking.ppt

Try clicking on the link above. You can now preview the file in your browser without downloading it!

Isn't that neat?

Hope this helps save you some time and clutter. Happy browsing!