A Few Pointless Thoughts

Cruises for the Rest of Us

Santa Cruz de Tenerife.

Hello! I don't know what happened. Sorry. I think I fell asleep for half a year. OK, I really need to put up the rest of my photos from my US trip. But before that...

I'm in Barcelona! And instead of flying, I took a ship.

I was originally thinking of crossing the Atlantic on a freighter, but an article by a globetrotting programmer pointed me to a somewhat more comfortable option: cruise ships. Wait, wait — before you think "how posh", hear me out! As it turns out, during certain times of year, the big cruise lines ferry some of their ships across the Atlantic to prepare for their summer Mediterranean cruises. Since these sailings only go one way and happen during typically non-vacation times of the year, they are often very cheap. What's more, the article pointed to these cruises as a fantastic way to get programming work done. Sure, internet is a no-go, but if you can get past that, you basically have an idyllic floating office (food service and hot tub included!) for a work week or two. Given my needs, it was a no-brainer.

The author of the article had a website called CruiseSheet to help find these sorts of cruises. This is where I started my research.

Cruise Tips

Transatlantic cruises can be ridiculously cheap at certain times of year.

In March, April, and May of this year, I've seen a good number of 2 week cruises going for $300 per person for a 2 person inside cabin. Windowed cabins are usually only $50-$100 more per person. Remember what I said about freight ships? Turns out, they are usually about 3x as expensive, for reasons completely alien to me. However...

The list price of a cruise isn't really its final price.

...you also have to factor in port taxes and government fees (usually $100-$200). There's also gratuities, which usually default around $12-$13 per day. These are deducted automatically, but you're free to change them by talking to Guest Services. (Don't cheap out on this.) Fortunately, you can pay them off with onboard credit. Don't worry: food is included in the price!

For getting a good deal, travel agents don't really bring much to the table.

For the most part, travel agents offer the same exact price as the cruise websites, with a few rare exceptions. If you'd like to explore your options, CruiseCompete allows you to submit a request for quotes from travel agents. They end up mostly being the same; I think I only saw one that offered me a lower price than retail, and I had to buy their travel insurance to get it. One thing travel agents can offer you is a bit of onboard credit, but it's usually not much for cheap cruises like the ones I'm describing. I've only seen $50 at most in my price range.

But if you flex your capitalist muscles, you can save a few hundred bucks!

Most cruise lines have something called a "shareholder benefit", where if you own a small amount of their stock (usually 100 units, or about $6000 total), you can get $200-$250 onboard credit for these long cruises. Yes, it's a lot of cash to toy with — but if you have the means to do it, all you have to do is buy the stock, e-mail them a statement, and sell it off when it gets a bit higher. You'll end up with free onboard credit and quite possibly a bit more money in the bank to boot.

I just had to go and try this! As soon as I bought the stock on my Vanguard account (4 days processing), I sent Royal my statement. They got back to me in a week telling me that the credit would show up in my onboard account by the second day (which it did). The credit went on to cover the entirety of my gratuity ($12 per day by default), $50 of internet (about 10 minutes per day — more on that in another article), and part of a haircut. On the downside... I got unlucky and managed to buy the stock at its peak. A week after I bought it, I was down $400. Currently, I'm only down about $140, which means that I'm actually ahead by $110 if you factor in the onboard credit. Still, I think I'll wait until the stock surpasses the price I bought it at, which it probably will, given that the trendline has been pretty much linear for the past year. As with any transaction involving the stock market, YES, it's a gamble, and YES, there are potentially massive risks involved — no question about it. But for $250 for free... maybe worth it.

(Do note that the shareholder benefit might not be compatible with other discounts and onboard credit offers, though some people have suggested that travel agent credit doesn't count. I don't know!)

Single supplement sucks.

If you're travelling by yourself, you'll in all likelihood have to pay 2x the price. Pro: you have the entire room to yourself. Con: a burning hole in your pocket. Note that Norwegian Cruise Lines has rooms on some of their ships designed specifically for singles, allowing you to waive the single supplement. The one I saw cost $650 and was a swanky-looking inside suite.

Cruise prices can change drastically on an almost daily basis.

This was a bit surprising to me. When I saw the price of an oceanview room on a cruise I was considering drop nearly $300, to below the price of an inside cabin, and then back in a single day, I figured it was some sort of fluke. But no: if you look at the price graphs for the cheaper rooms on a price tracking site like CruiseFish, you'll probably see weird spikes a few times every month, sometimes to absurd lows. It sometimes pays to be patient, unless...

The perfect time to buy?

...you wake up one day to find everything sold out! I spent just a little too long deciding on my cruise, and I refreshed one morning to suddenly find that the prices for the cruise I was favoring spiked by almost $500 overnight. A quick call revealed that the cheap oceanview cabins had all sold out. I asked about a different cruise, and it turned out that there was only a single room available at the good price. I quickly nabbed it. Turns out that even for these lesser-sailed routes, you'll start seeing the cheap rooms sell out about a month in advance, and probably a bit earlier than that. (Note that if prices drop within 48 hours, most cruise lines have a price guarantee that you can redeem for onboard credit. Be vigilant after you buy!)

Some Data

I was interested in the long-term behavior of these prices, so I picked 5 cruises out of the ones I was considering and tracked their cheapest inside and outside cabin prices over the course of about a month. Here are the results, adjusted for the single supplement. ($0 means sold out.)

As you can see, pricing behavior varied wildly between the three companies. Norwegian barely changed their prices. Celebrity changed their prices occasionally. Royal's indoor cabin prices remained mostly stable, but their outdoor cabins fluctated wildly and unpredictably, almost to the very last day. (And even with their indoor cabins, you could save a hundred bucks just by waiting a day.) As we got closer to the sailing date, prices rose all around. Cabins were selling out on a whim, then coming back a few days later.

Based on my findings, for the category of cheap transatlantic cruises out of the US, I think 2 months ahead of time is a good time to buy, with 1 month being the point at which things start selling out rapidly.

In Conclusion

After a month of nail biting and frantic refreshing, I decided on a 16-day Royal Caribbean cruise from New Orleans to Barcelona. My original intent was to land in London, but the 16-day was cheap enough that if I really wanted to get to England straight away, I could simply take the train for close to the price difference. The pre-gratuity price for the voyage was $1100 for a windowed room. (Non-windowed was $200 cheaper, but basically my entire family forced me to take the windowed room by helping pay the difference.) Going with a friend would have cost $550 per person, but sadly, no friends were available. At the time I booked the ship, a flight would have cost me about $600. Transport pricing is unfortunately pretty obtuse, and there are a lot of factors to consider. I've heard, for example, that round-trip plane tickets tend to be cheaper than one-way, and that there's the occasional rare deal where you can fly into a European country for only $200-$300. But all in all, even without factoring in the benefit of transportation, I think the cruise was a fantastic deal. Depending on how you look at it, I got to spend $50-$80 per day total for a great office, fantastic food, and the rare experiencing of crossing the ocean (the ocean!!) by sea.

Late 2013 15" Macbook Pro: Intel Iris Pro 5200 vs. Nvidia 750m (And Other Stories)

I recently got a high-end 15" Macbook Pro. The 13" model I was using before had served me with faith and dignity over the years, but as my appetite for high-performance apps increased, the poor guy just couldn't keep up like it used to. In the past, I would have only considered upgrading to another 13" laptop, but a lot has changed over the years. Computers have slimmed down. I've slimmed up. A 15" device just didn't seem like the back-breaking monster it used to be.

The other big factor in my decision was graphics performance. Among all the Macbooks currently available, the high-end 15" Macbook Pro is the only one with a discrete graphics chip still inside. You get access to an integrated Intel Iris Pro 5200 for everyday use, but the OS can also switch you over to a powerful Nvidia 750m when the polygons have to fly.

At first, I naturally assumed that the 750m would kick the 5200's butt; this was a separate ~40w component, after all. But as I started to dig through forum posts and benchmarks for my research, I discovered that while the Iris Pro usually lagged behind the 750m by 15%-50%, there were a few recorded instances where it matched or even surpassed the Nvidia chip! Some people blamed this on drivers, others on architecture. Were the numbers even accurate? I wanted to find out for myself.

There were a couple of specific questions I was looking to answer during my testing:

  • How good is the maximum graphics performance of this machine?
  • How does the Iris Pro 5200 compare to the 750m?
  • How does Windows 7 VM (Parallels) graphics performance compare to native Windows 7 (Bootcamp)?

3DMark

For my first set of benchmarks, I configured a Parallels VM to run off my Bootcamp partition with the following settings: 4 logical processors (which I think gives you 2 physical cores), 12GB RAM, 1GB video memory, DirectX 10, and 1440×900 resolution. (I also ran a test with 2 logical processors, which caused performance issues, and also with 8 logical processors, which caused my system to seriously freeze up.) I then installed the 3DMark demo on Steam and ran the default suite of tests. Finally, I turned off the VM, changed the energy setting from "Better performance" to "Longer battery life" in order to make the VM use integrated graphics (verified with gfxCardStatus), and ran the same tests again.

(Oh, before I give you my results, I should mention that none of these tests are scientifically rigorous. I tried to be as accurate as possible, but I'm no Anand Shimpi. Also, even though the results of my VM tests were consistent with everything else I tried, I'm not sure how much the numbers were skewed given that they were running through Parallels' DirectX to OpenGL layer.)

3DMark Vantage in Windows VM
Integrated Discrete Discrete Performance Over Integrated
Ice Storm 52223 60862 116.5%
Graphics 55459 69401 125.1%
Graphics Test 1 252.4 fps 326.2 fps 129.2%
Graphics Test 2 230.8 fps 280.7 fps 121.6%
Cloud Gate 6035 6415 106.3%
Graphics 7067 7709 109.1%
Graphics Test 1 29.0 fps 28.6 fps 98.6%
Graphics Test 2 32.6 fps 40.4 fps 123.9%
Physics Test 12.7 fps 12.8 fps 100.8%

The Physics test only measures CPU speed and can thus be ignored. Insofar as benchmarks are concerned, it looks like the 750m's gains over the Iris Pro are moderate: 25% in the basic test and 9% in the more intensive test. Do note, however, that the second number is an average, and that Graphics Test 2 in Ice Storm also showed ~25% improvement.

I then ran the same test in native Bootcamp. Unfortunately, it's impossible to switch to integrated graphics in Windows, so this is only useful for comparing discrete performance between native and VM.

3DMark Vantage in Windows Native
Discrete Windows Native Performance Over Windows VM
Ice Storm 80004 131.5%
Graphics 101785 146.7%
Physics 45745 107.5%
Graphics Test 1 447.9 fps 137.3%
Graphics Test 2 437.3 fps 155.8%
Physics Test 145.2 fps 107.5%
Cloud Gate 10258 159.9%
Graphics 12601 163.5%
Physics 6215 153.8%
Graphics Test 1 55.0 fps 192.3%
Graphics Test 2 54.6 fps 135.1%
Physics Test 19.7 fps 153.9%

As one might expect, Bootcamp clobbers Parallels. The more demanding Cloud Gate benchmark shows much bigger gains than Ice Storm, implying that the VM is better suited to slightly older games.

3DMark in Bootcamp also added one more test at the end, possibly due to the fact that Parallels only supports DirectX 10. I'm putting it here for completion's sake.

3DMark Vantage Fire Strike in Windows Native
Discrete
Fire Strike 1741
Graphics 1790
Physics 9024
Combined 721
Graphics Test 1 8.3 fps
Graphics Test 2 7.32 fps
Physics Test 28.6 fps
Combined Test 3.35 fps

Unigine Heaven

Next, I downloaded the Unigine Heaven benchmark. This benchmark is very convenient because it can be run natively in both Windows and OSX, allowing us to make cross-platform comparisons.

First, I ran the same Parallels test that I did for 3DMark using Unigine's two built-in presets. Unfortunately, it turned out that the Extreme test didn't support tesselation in VM but did when run natively, forcing me to do a few custom runs on the other platforms with tesselation turned off. It also threw up errors when I tried to run the test in OpenGL or DirectX 9, so I had to stick to DirectX 11 in the VM.

Unigine Heaven in Windows VM
Integrated Discrete Discrete Performance Over Integrated
Basic (DirectX 9) 660 (26.2 fps) 856 (34.0 fps) 129.7%
Extreme (DrectX 11 Without Tesselation) 241 (9.6 fps) 416 (16.5 fps) 172.6%

A surprisingly significant result for the Extreme benchmark! Discrete performance is nearly twice as fast.

Next, I did the same test in OSX, using the ever-so-convenient gfxCardStatus to switch between integrated and discrete graphics.

Unigine Heaven in OSX
Integrated Discrete Discrete Performance Over Integrated OSX Integrated Performance Over Windows VM Integrated OSX Discrete Performance Over Windows VM Discrete
Basic 543 (21.6 fps) 870 (34.5 fps) 160.2% 82.3% 101.6%
Extreme (OpenGL) 157 (6.2 fps) 271 (10.8 fps) 172.6%
Extreme (OpenGL Without Tesselation) 230 (9.1 fps) 384 (15.2 fps) 167.0% 95.4% 92.3%

Yikes! Looks like running a benchmark tool in a VM, through Parallels's DirectX to OpenGL layer, and finally through the OSX OpenGL driver is somehow faster than running the same benchmark natively. On the other hand, the difference between integrated and discrete performance is a lot more consistent here. Like in the previous test, we see almost twofold gains.

Finally, I ran the same test in Bootcamp.

Unigine Heaven in Windows Native
Discrete Windows Native Performance Over Windows VM Windows Native Performance Over OSX
Basic (OpenGL) 971 (38.5 fps) 111.6%
Basic (DirectX 9) 994 (39.5 fps) 116.1%
Extreme (OpenGL) 298 (11.8 fps) 110.0%
Extreme (DirectX 11 Without Tesselation) 465 (18.5 fps) 111.8%

Looks like the difference between all three platforms for this particular benchmark isn't too significant.

Metro: Last Light

For my next test, I decided to go a little crazy and give Metro: Last Light a go. Frankly, I wasn't even sure if Parallels was up to the task! As it turns out, modern VMs are a lot more powerful than they look.

Here are the presets I used with the handy MetroLLbenchmark.exe utility.

  • Preset 1: 1440×900, DirectX 10, low quality, AF 4×, low motion blur, SSAA off, PhysX off
  • Preset 2: 1440×900, DirectX 10, high quality, AF 16×, normal motion blur, SSAA off, PhysX off
  • Preset 3: 1920×1200, DirectX 10, high quality, AF 16×, normal motion blur, SSAA on, PhysX off

And here are the results with the typical Parallels shenanigans.

Metro: Last Light in Windows VM
Integrated Discrete Discrete Performance Over Integrated
Preset 2 10.82 fps (1657 frames) 18.7 fps (2940 frames) 172.8%
Preset 3 3.89 fps (589 frames) 6.6 fps (1023 frames) 169.7%

As with Unigine, we see almost twofold gains when using discrete graphics. Makes sense: Metro is one of the most graphically intensive games on PC right now.

Next, I ran all my presets in Bootcamp.

Metro: Last Light in Windows Native
Discrete Windows Native Performance Over Windows VM
Preset 1 43.74 fps (7480 frames)
Preset 2 27.76 frames (4744 frames) 148.4%
Preset 3 9.54 fps (1624 frames) 144.5%

I was surprised to see that the gain from running natively over running in VM was only around 50%. For a game like Metro, I expected native Windows to completely blow the VM away. You probably wouldn't want to run the game in Parallels due to the fact that it's a bit jittery and unstable, but still!

Just to see how far I could push this machine, I added a couple of modified runs: Preset 2 with SSAA on, and Preset 2 with PhysX on.

Metro: Last Light Modified Preset 2 in Windows Native
Discrete Modified Preset Performance Over Preset 2
Preset 2 (SSAA on) 16.22 fps (2767 frames) 58.4%
Preset 2 (PhysX on) 21.69 fps (3706 frames) 78.1%

If you're going for performance, SSAA is clearly a killer. PhysX, on the other hand, doesn't make as much of an impact as I expected. Still might be worth turning off: I didn't notice anything different in the benchmark.

Just for kicks, I did one more batch of runs in DirectX 11 mode.

Metro: Last Light DirectX 11 in Windows Native
Discrete DirectX 11 Performance Over DirectX 10
Preset 1 44.91 fps (7679 frames) 102.7%
Preset 2 29.74 frames (5083 frames) 107.1%
Preset 2 (SSAA on) 17.58 fps (3002 frames) 108.4%
Preset 3 10.58 fps (1803 frames) 110.9%

Whoa! It actually runs better? Definitely wasn't expecting that. Certainly not the case on my desktop.

I would have loved to test the performance of Metro on OSX, but unfortunately the OSX port had no benchmark tool and barely offered any graphics options to speak of.

Batman: Arkham City

Finally, I found the one recent game in my Steam library that had an in-game benchmark on OSX: Batman: Arkham City. I ran it on the Low and High presets in 1680×1050.

Batman: Arkham City in OSX
Integrated Discrete Discrete Performance Over Integrated
Low 37 fps 54 fps 145.9%
High 32 fps 44 fps 137.5%

The Unreal 3 engine isn't as fancy as the Metro engine and the benchmarks reflect this. Discrete performance is "only" around 40% faster than integrated.

A few more telling benchmarks would be Far Cry 3, Crysis 2/3, Battlefield 3/4, and Max Payne 3. These represent some of the most powerful engines available on PC right now. I might add a few more measurements later on.

Conclusion

There you have it! A bit of benchmark geekery to shed some light on the new Macbook Pro's graphics situation. What have we learned?

  • The Nvidia 750m is significantly better than the Intel Iris Pro 5200 in many cases, surpassing it by 25%-70% in framerate. With that said, the Iris Pro is powerful enough to run most of the same games at lower settings, which is all the more impressive considering its very low power consumption. I would frankly be surprised if the next generation of Macbook Pros even had a discrete graphics chip, and given the speed at which Intel is improving their graphics performance, that might be OK by me.
  • Bootcamp performs a whole lot better than Parallels, generally by 50%-60%. Performance is also a lot smoother. Still pretty darn impressive, considering that the VM boots up in about 5 seconds and hangs out in your dock while Bootcamp requires closing all your applications and rebooting. For the vast majority of Windows applications, Parallels looks like the winner. (Side note: I noticed really frustrating controller lag in Spelunky until I disabled vertical synchronization in the VM options. Unfortunately, this caused some pretty heavy tearing.)
  • I wasn't specifically testing for it since I knew the answer already, but OSX really suffers in graphics performance compared to Windows. The state of Mac ports in general is honestly a bit sad. Many of them run in a compatibility layer like Wine instead of running natively. Saves usually aren't synced between Mac and Windows. Graphics menus often leave you with far fewer choices, going so far as a single-axis slider for Metro. Crashes and other bizzare problems frequently show up. Unless you're playing games by Valve or Blizzard, you'll be better off running in Bootcamp or Parallels. (And even then, I've noticed weird mouse behavior, network errors, and outright crashes in the Mac port of Counter-Strike: Global Offensive!)
  • In terms of hardware, Macs haven't traditionally been known for their gaming prowess, but all that's changed. If I can run Metro, I can call this a gaming machine. End of story.

There are a few other subtle advantages and disadvantages that the discrete graphics chip brings to the table. I might as well mention them here.

  • Pro: OpenCL can allegedly use both the integrated and discrete chips simultaneously. Apple seems to be pushing for OpenCL these days, so we may see significant performance gains in this area as high performance apps catch up. Hopefully, the new Mac Pro will be a cataclyst for this.
  • Con: Discrete graphics really eat through your battery life. While most applications don't automatically switch to discrete graphics, some do and won't switch back until you close them. As far as I know, the only easy way to tell which chip is currently active is to use gfxCardStatus. You can also use this tool to force OSX into integrated graphics mode, unless you're hooked up to an external monitor. However, there's no guarantee that this workaround will keep working with future OS updates!
  • Con: In Bootcamp, you can't switch to integrated graphics. This means that if you run your top-of-the-line Macbook in Windows, you'll have poor battery life. Not so if you buy a Macbook without discrete graphics.

Finally, a quick note on the day-to-day performance of this machine. Holy matchsticks, is it fast! I didn't realize it was even possible to boot up a VM in 5 seconds, let alone run a game like Metro on it. On my previous machine, a mid-range Core2Duo with 8GB RAM, booting up the VM was an ordeal, and often left one OS or the other in an unusable state. Here, it's as smooth as opening any other app. In terms of game performance, I'm super impressed by how well most of my favorite games run, even compared to my desktop. Lightroom is blazing fast when browsing and developing, though full-size previews for my 8MP photos take a second each to generate. And, of course, the Retina display is stunningly beautiful, though it does seem a bit darker than the screen on my previous laptop. Might just be a trick of the eye.

(One minor Bootcamp issue: Windows 7 doesn't deal with DPI scaling very well, so running at the native 2880×1800 with 200% DPI looks kinda bad. Unfortunately, the halved 1440×900 isn't a standard resolution according to the Nvidia driver and looks a bit blurry. Since anything other than 2880×1800 was going to be blurry anyways, I've resorted to running in 1680×1050. Interestingly, since the Windows VM uses Parallels' own graphics driver, it does not have this problem at 1440×900.)

I realize this is one of the most expensive machines on the market, but if you want something that can develop OSX/iOS apps and run Windows and play all your games and be compact enough to travel with for long periods of time and have enough screen estate to do serious work... [deep breath]... and have a beautiful and sturdy design that will last you many years, this laptop truly has no equals. Price is a tough pill to swallow, though.

Other References

Seattle: You're Probably Tired of Seattle By Now

A week into my Seattle visit, I moved from the downtown City Hostel to the confusingly-named Hotel Hotel Hostel in the Fremont neighborhood. (The very same neighborhood I had visited with my friends a few days prior.) The streets were quiet and a little quirky, the nearby water gently lapped at the shoreline, and the smell of chocolate followed you down the street. It felt a little like being on vacation. (Of course, having a fantastic gastropub down the street helped. Did I mention I hate the word "gastropub"? It sounds like somewhere you go to cleanse your body of toxins. But I digress...)

I found myself on the same side of the canal as the University of Washington, one of my favorite university campuses. It was raining when I first visited many years ago, turning the entire Red Square into an enormous skyward mirror. The sun was defiantly shining this time around, but the architecture was no less spectacular.

Suzzallo Library

I spent maybe a little too much time wearing my metaphorical beret.

The Red Square monoliths.

Fortunately, a sign on the nearby wall helped ground me in my present reality.

Turned out the University had a wonderful foot trail that went right along the Montlake Cut, which connects Portage Bay and Lake Washington. I spent a while watching the boats ebb and flow beneath the nearby bridge.

Lots of people were out that day sailing and enjoying the weather.

As the coda to my Seattle adventure, I payed a visit to the Smith Tower. It might no longer be the tallest building in town, but it's one of the oldest skyscrapers in the US, if the guides are to be believed. After taking a ride in a vintage elevator (operator and all!), I spent a lovely hour wandering around the airy observation deck and and soaking in the full panorama of the city, 400 feet above the ground.

All in all: great city, fast shipping, A+++, would visit again.

I couldn't figure out where to stick this photo in my narrative, so have it here.

Next: paying a visit to to the land of milk curds and honey gravy!

Seattle Scenery

There are many places you can work while travelling. You can stay at your hostel. You can go to a café. But let it be known that none of them will have the grandeur of the Seattle Public Library.

On the tenth floor of the Central building, from behind the saftey of the honeycomb window lattice, you can enjoy the ethereal imprint of the raindrops against the brusque geometry of the downtown skyscrapers.

When your workday is over, you can take a stroll down south and walk along the waterline.

The wooden piers creak quietly as the boats come in and out of the docks.

On the opposite side of the Space Needle, Kerry Park has some fantastic views of the skyline. From there, if you're as entranced by industrial equipment as I am, you can walk a little bit further north towards Queen Anne Hill. Little roads and staircases coexist with grandiose mansions in this well-to-do neighborhood.

On the very top are three broadcast antennas, reaching towards the distant sky.

The walk back towards the water is quite beautiful, with the city unfurling right before your feet.

Next time: a visit to the university and a trip up a tower!

Seattle: Of Lands and Locals

I spent a couple of wonderful days in Seattle hanging out with an old friend, a new friend, and a baby.

We met up a few times at my friend's house for dinner and board games. The place was usually brimming with family and friends (great folks, the lot of them) as well as their children, who ran around and did whatever it is that children do. Looking back, I've spent a surprisingly small amount of time around kids ever since I was a kid myself, and it was inspiring to once again experience the kind of unreserved enthusiasm and curiosity that tends to fade as you get older. (One of them even asked if my camera was a Nikon!) I was also happy to see that the next generation of gamers was being raised by people with impeccable taste: Rayman Legends, Duck Tales, Fez.

My friend's 1-year-old was particularly interesting to watch. I never realized how quickly toddlers picked up on things! You showed him how to use the shutter release on your camera, and suddenly he was almost snapping photos all by himself. One time, I poked his finger when he was pointing at me; a second later, he was touching his own two fingers together, seemingly fascinated by the concept. Having recently spent about 5 minutes trying to turn off a hostel TV set, I was clearly outclassed. Kids and their pliable brains.

Anyway, it was a lot of fun. As a souvenir, I even managed to pick up some arcane knowledge, including how to shuffle cards the cool flippy way and how to play mahjong just like an old Japanese lady. (I say Japanese because the rules my friend taught me were for the Japanese variety of the game, which I didn't even know was a thing. Also, mahjong is really fun!)

The four of us also took a short day trip to Fremont.

This charming little neighborhood is located on the north shore of the Lake Washington Ship Canal, crossed by the Aurora and Fremont Bridges. The feeling here is a lot more laid-back than in lower Seattle, brought on in part by the tree-lined streets and the calming waterfront. (Of course, it also helps to have a few bearded guardians watching over you.)

After stopping by a parking lot coffee shack, we crossed the Fremont Bridge and walked along the shore in hopes that the bridge would salute for a passing ship. Sure enough, we weren't disappointed.

Up close, the bridge revealed its beautiful geometry and faded colors.

We also stopped by a neat little music store. They had everything from electric bass ukuleles to wooden dulcimers. I briefly considered buying a harmonica to round out my hobo-nouveaux lifestyle, but ultimately decided against it.

A huge thanks to my friend for the warm welcome! I couldn't have asked for a better way to start my journey.

Next: more Seattle, Vancouver, and a bunch of other things that have been on the backburner for a while. (I promise the next article won't take nearly as long as this one.)