or Connect
Head-Fi.org › Forums › Head-Fi Special Forums › Head-Fi Bloggers › Jason Stoddard › Schiit Happened: The Story of the World's Most Improbable Start-Up
New Posts  All Forums:Forum Nav:

Schiit Happened: The Story of the World's Most Improbable Start-Up - Page 47

post #691 of 14744
Quote:
Originally Posted by AK7579 View Post
 

How about some Olde English? 

Olde-English-1.jpg

Oh Dear God!  Not OE!

post #692 of 14744

Looks like I'm drinking bum wines tonight.

post #693 of 14744

 

Wow listening to this through my DT880s and Schiit stack in FLAC. EPIC! :]

post #694 of 14744
What's funny about OE 800 is that it's formulated differently in the US depending on which state you are in. In so called "semi dry" states it can be as low as 2.5% abv. So it tastes like schiit with no benefit. However in a beer country like Canada we always get the full 8%. If you can drink two 40s of it and still stand straight you are a better man than I biggrin.gif
post #695 of 14744

Excellent reading, thanks for sharing

post #696 of 14744

The health and safety stuff brings back memories. The warehouse I was in was a pretty tight shop when it came to rules (no daisy-chaining power cords even -- though in every other respect the place was total anarchy) but the importer across the park that handled incoming Dell and Apple shipments (among a crapload else) had a reputation for forklift madness, especially as they had quite a few gas-powered ones and a lot of gung-ho workers, one of whom got skewered... 

 

That brings up a famous story though:  An OHSE (or rather the Australian equivalent) officer was pulled over by a cop for speeding and given a $160 ticket. As the cop went back to his car, the worker pulled out his camera and took a picture, then went and fined the cop $900 for not wearing his fluorescent safety vest.


Edited by Currawong - 4/4/14 at 10:37pm
post #697 of 14744
I don't even know who to support in this scenario Currawaong.
post #698 of 14744

What comes around, goes around, I guess. Reminds me of those stories where people with authority may let it go a little to far, only to be surprised with a tax audit from hell.

post #699 of 14744
Quote:
Originally Posted by Currawong View Post
That brings up a famous story though:  An OHSE (or rather the Australian equivalent) officer was pulled over by a cop for speeding and given a $160 ticket. As the cop went back to his car, the worker pulled out his camera and took a picture, then went and fined the cop $900 for not wearing his fluorescent safety vest.

Perfect example of a race to the bottom.

post #700 of 14744

As an HR Director, wine lover and proud owner of an Asgard 2 and Uber Bifrost, I love this installment.

 

Where else will you find the test as to whether someone is an independent contractor or an employee (thanks Jason for not going through what could be up to 20 factors!).

 

Can't wait until next Wednesday! 

post #701 of 14744
Thread Starter 

Chapter 11:

USB Sucks! Or, Mike Joins the 21st Century

 

Okay. Time for me to take a step down. Until now, most of the book has been about my designs, but now, it’s time to talk about Mike Moffat and DACs.

 

But first, a scorecard. At this point in time, we’re early in 2011. Say, 9 months old. We’ve introduced:

 

  • Asgard
  • Valhalla
  • Lyr

 

And I’d talked about upcoming DACs and other fantasy products with 6Moons, further deepening my “don’t talk about it” dilemma. Again, if I could go back and punch myself in the face, I would. But I really wanted everyone to know that we weren’t just going to be about value products.

 

Anyway, on to Mike.

 

If you know a little bit about the history of digital audio, you know that Mike was one of the first guys to take digital seriously, and the first to introduce a standalone DAC. Yes, you can thank Mike for all the separate DAC vs CD player, DAC vs sound card, DAC vs the D/A in your phone/computer/Blu-Ray/oven/lawnmower arguments we have today.

 

Now, some others use technicalities to claim the “first DAC” prize, but the fact is: Mike and his company, Theta Digital, were first. The technicality is that their first DAC was actually a DAC and preamp, the Theta DS Pre. Back then, the idea of a DAC and preamp combined was, well, more than slightly strange. So that led to the introduction of the Theta DS Pro shortly afterwards, so confused audiophiles had a component to use with their uber-expensive preamps of the time.

 

This same Mike Moffat also designs all the Schiit DACs. So, it’s not like we just decided to get into the DAC market—we have the guy who started it all.

 

It’s funny. Shortly after we introduced the Bifrost DAC, I got an email from a prospective customer that went something like this:

 

“Hey, this looks interesting, but I’m wondering what your credentials are in digital design…let me know, please?”

 

To which I replied something like this:

 

“Well, other than having the “father of the DAC,” Mike Moffat, on our team, we can recount all the stuff he brought to the table in terms of DAC design. This includes:

 

  • Much experimentation with early 2-chassis player/DAC designs with shared clocks before the SPDIF standard was approved
  • The first standalone DAC
  • The first to use custom digital filters in a DAC
  • The first to use DSP to run those digital filters
  • First true time domain optimized digital filters, based on math perfected with a U of Iowa Professor Emeritus of Mathematics and a RAND Corp mathematician
  • The first to identify jitter as a cause of audio degradation—before Mike, “bits is bits”—you can thank him for literally every jitter argument we have today
  • The first to work to minimize the causes of jitter
  • First to use a Stanford interval counter for jitter analysis, before “convenient” measurement via JTest
  • Presented an AES paper on jitter in digital audio
  • The first to introduce the AT&T ST-optical interface to address jitter issues
  • The first with upgradable DACs at Theta (from the beginning in 1986)
  • The first to make an upgradable surround sound processor at Angstrom
  • The first DTS-capable surround processor at Angstrom
  • Multiple firsts in entertainment digital media distribution engineering at Digi-Flix

 

But even before Mike was doing digital, he was doing audio, at Theta:

 

  • One of the first in the tube revival in the 1970s
  • The first non-12AX7 implementation for audio (revolutionary in its day)
  • The first “no overall feedback” tube stage in the 1970s
  • One of the first passive preamps
  • Worked in the Chilean jungle looking for oil using the first 8-bit A-D converters prior to Theta
  • In 1934, many years before he was born, God appeared to Mike and revealed to him the formula for amazing digital audio, which he has inscribed on 12 lead tablets…

 

The prospective customer’s response:

 

“Oh, then you’re going to annihilate pretty much everything, then?”

 

Ha. If it was only so easy.

 

 

When Mike Says USB Sucks, You Listen

 

Okay. Enough of the bragging. The fact is, I’m excited to have Mike Moffat as a partner, and I’m proud of his resume. He’s contributed quite a bit to the digital audio realm—just look at Sony’s original stance of “perfect sound forever,” and their current frank discussion of jitter-reducing measures in their audio products. Quite a turnaround.

 

But to get back on topic, let’s talk about DACs. Not Mike’s original idea for a DAC, the one he came up with shortly after we started the company, but Bifrost.

 

An aside: Mike’s original idea for a DAC is what eventually turned into Yggdrasil. Yes, we’ve been talking about it for that long. Yes, we’re really late. But it has grown and morphed over time. And, in the early part of 2011, we weren’t ready for a DAC that cost 8x as much as an Asgard.

 

Once we were focused on the real goal—an inexpensive DAC in the same size chassis as Asgard, Valhalla, and Lyr—that’s when the arguments started.

 

“And it’ll have optical, coaxial, and USB inputs,” I opined to Mike.

 

“USB?” Mike gagged, miming sticking a finger down his throat to induce vomiting. “USB is for children and fools. Why would you want to use USB for audio?”

 

“Because it’s really popular,” I said. “Everyone has it—“

 

“If everyone was dressing up in tutus, would you?” Mike shot back.

 

“No, but…”

 

“No, but you’d say you did, just to be popular.”

 

“We have to have USB,” I pressed on. “There are a ton of people who only have laptops for sources, and most of those only have USB.”

 

Mike grumbled something under his breath, then spat out: “But USB sucks. It just sucks. It was never meant for audio. It’s an all-purpose, packet-based grab-bag that might be fine for printers or hard drives, but it’s just crap for streaming. You can recover the clock from the packet clock, barf, or you can have the computer and DAC do some negotiating and guess at the clock, barf, or you can turn the whole car around and drive it from the back seat with the computer providing the clock, barf.”

 

Note: the above, for the more technical, is Mike’s take on isosynchronus, adaptive, and asynchronous USB implementations.

 

Second note: also, remember, this is early 2011 we’re talking about. Adaptive USB 1.1 was kinda the de facto “good solution”, with most audio components using the truly terrifying TI USB input receiver/DAC/headphone amplifier/car washer chips that weren’t even adaptive. Some guys were fooling around with USB 2.0, but implementations were thin and software was iffy. We know. We tried all of them.

 

“We have to have USB,” I told Mike, firmly.

 

Mike grumbled again.

 

“Remember Angstrom,” I said, bringing out the big guns.

 

Angstrom was Mike’s surround processor company. It would probably still be around today, except for various life issues that aren’t mine to talk about. But it’s also the company where we discovered that going against the grain might not be the best idea, even if it ends up being the right idea.

 

Angstrom brought out one of the first inexpensive Dolby Digital decoders on the market, the Angstrom 100. It was a great product. But Mike didn’t want to do video switching. Video switching, he said, had no place in a no-compromise home theater audio product. What’s more, the new HDMI standard was imminent, and that was a whole new ballgame. The whole video switching deal was going to be changing, and fast. So why put it in a product when it was going to be obsolete in less than a year?

 

Yes, it made sense. To us. Unfortunately, to customers, it made a lot less sense, especially when Angstrom never brought out the promised separate video switcher.

 

“But Angstrom was right,” Mike said. “Look at all the processors today, you throw them away when HDMI changes. Now we have the same thing with USB. You know there’s gonna be a better way to do it in a few months.”

 

“I know. But USB is a must-have. We’re dead in the water without it.”

 

Mike sat for a long time, saying nothing. Then his eyes lit up. “Make it upgradable,” he said.

 

I shook my head in confusion. “Make what upgradable?”

 

“The USB input. Put it on a separate card. So you can change it when the technology changes.” Now, Mike was excited. He jumped up. “No. Not just USB. Make the whole thing upgradable, so you can swap the DAC as well! That’s the way you do it! Not just a throwaway product, something you can keep for as long as you want.”

 

“The whole thing? Upgradable?” I was skeptical. “Mike, you know we’re talking about a product that only costs a few hundred dollars.”

 

“Right. And that’s the brilliant part. You can buy a throwaway DAC, or you can buy ours.”

 

And that’s how Bifrost became upgradable. A great ending to the “Not Invented Here” syndrome—which I’ll cover some more in the business bits part of the chapter.

 

“And,” Mike said, pointing a finger in the air, “We’ll also make sure it’s bitperfect to the DAC, instead of using an asynchronous sample rate converter.”

 

“Is that good?” I asked. (Remember, I’m the analog guy.)

 

“Is it good? Is it good, he asks?” Mike said, recoiling from me as if I’d just asked if a Michelin-star restaurant was better than McDonald’s. “It’s an absolute necessity if you don’t want to throw away all the original music data, and create some mathematical-abortion-mishmash of interpolated crap, especially if you’re at a non-binary multiple of the original sample rate, like going from 16/44.1 to 24/192. No, wait. Let me guess. You’re gonna tell me, in your infinite marketing wisdom, we have to crap everything up to 24/192 so we can have a number on a datasheet. Oh, boy.”

 

“No, I’m not gonna tell you that,” I said. “But what are we talking about here? How hard is it to keep everything bitperfect?”

 

“It’s a pain in the ass,” Mike said. “We’ll need a microprocessor to switch the clocks, we have to reset the DAC when sample rates change, we’ll need a hard relay mute, stuff like that.”

 

“Expensive?”

 

“More than just throwing in an ASRC chip and being done with it,” Mike said. “And we’ll need firmware.”

 

I sat quietly for a bit. Firmware meant Dave, the unsung code/digital hero of Schiit Audio. He’s the third guy on the engineering team. But bringing him on meant even more expense, and we were still struggling to simply keep shipping on time.

 

But if Mike wants something, he gets it. So we got Dave. And we started working on the first prototype of what would become Bifrost. This was a bigger project than anything we’d taken on before, involving digital design, firmware, system integration, and analog design. I did only one thing on Bifrost—the discrete analog output stage. And that was plenty. Analog electronics in a digital box, being fed by the relatively noisy output of a D/A chip, is a whole different ballgame than a preamp or power amp. In that kind of environment, analog electronics like to oscillate. So, many iterations of compensation and filtering were in order. That took a good piece of my time.

 

While I was working on that, Mike and Dave were doing the rest. This included:

 

  • Evaluation of the various USB input solutions
  • Integration of the SPDIF inputs
  • Microprocessor-based clock management
  • DAC evaluation (one of the good things about a modular DAC is that it lets us try all the leading candidates for D/A ICs out there—we settled on AKM because it’s what sounded the best, plus it’s one of the best-measuring DACs out there)

 

Fun fact: the first layout of Bifrost had the input selector switch backwards, pointing inside the chassis. I joked with Mike that we could put a cantilever behind the front panel button to activate it, and I think for a few moments, he actually believed me.

 

Of all “the rest,” what took the longest was by far the USB input. Back then, you had the option of licensing code for USB 1.1, using a standard USB 1.1 input chip (the one I mentioned that has all the other stuff tacked onto it), or using one of three different USB 2.0 input solutions. The problem with the USB 2.0 inputs was that they were all kinda beta-ish in one way or another. One needed drivers for Mac and PC, despite Macs supporting the USB Audio 2.0 Standard natively. One was really ambiguous about their drivers and licensing.

 

And then there was C-Media. C-Media was an obscure Taiwanese company that has just introduced their CM6631 USB 2.0 input receiver. And by “just introduced,” I mean, “the datasheet was labeled Version 0.9, and the USB firmware programmer crashed after successfully programming one device—every time.”

 

“Version 0.9 on a datasheet,” Mike said, grimly, when we got the first docs. “It’s a beta. Run away. Run far, far away.”

 

But C-Media was helpful, providing the support we needed to get their device up and running. And when it was running, it sounded pretty good. And by “pretty good,” I mean “as good or better than anything else we tried.”

 

Mike was less impressed. “It’s not complete crap,” he said.

 

Later, we’d make it the USB input better with some tweaks, to the point where we thought we had one of the better-sounding implementations out there. That was where we launched with the original Bifrost—but that happens much later in the story. (And today, with the new CM6631A, I’m finally totally happy with USB. In fact, I use it most of the time at home.)

 

 

A USB Mode/USB Audio Standards Primer

 

Okay. Let’s go to the “useful data” side of things. A lot of people are monumentally confused about USB audio input. So here’s a guide to the whys and wherefores—something I should probably put on our site.

 

USB Mode Versus USB Audio Standard. A ton of people are confused when I say things like, “Modi uses USB Audio 1.0 Standard over USB 2.0.” They think I mean that it uses USB 1.0 as a transmission protocol. Actually, it doesn’t. It runs USB 2.0 at 480MBPS, but transmits audio using the USB Audio 1.0 standard.

 

Still confused? Okay, let’s break it down.

 

USB Modes. This is all about data rate. This has nothing to do with audio.

 

1.0: The earliest standard. So slow I forgot what it was. Not used for audio.

1.1: Transmits data up to 12Mbps. Can be used to transmit audio up to 24/96.

2.0: Transmits data up to 480Mbps. Can be used to transmit audio up to insane torture-the-cats-and-hard-drives rates like 32/768 and such.

3.0: Transmits data up to 5Gbps (ha.) No USB Audio 3.0 standard. No USB Audio 3.0 receivers. However, USB 3.0 ports are backwards-compatible with USB 2.0, so they can be used with anything using USB Audio 2.0 standard.*

3.1: New reversible fantasy USB spec created out of Apple envy and support for people too dumb to insert a cable the right way. No products out yet. Looking forward to all the confusion coming our way because of this.

 

USB Audio Standards. These are standards (hello, Microsoft) created to enable the transfer of audio data over USB. They are not USB modes.

 

USB Audio Standard 1.0. Supported by everyone. Plug a DAC using USB Audio Standard 1.0, like Modi, into any Mac or PC, many Linux systems, some phones, etc. and it will be recognized and play audio with no drivers required, up to 24/96.*

 

USB Audio Standard 2.0. Hello, Microsoft. There is this new standard called USB Audio Standard 2.0. And you really should support it. Because you look really dumb when anyone can plug in a DAC using USB Audio 2.0 Standard to any Mac, many Linux machines, phones, etc and expect it to work without drivers,* while you still rely on kludgy workarounds that require drivers, ASIO setup or WASAPI setup, etc. Please please please please include this in Windows 8.2 or Windows 9 or Windows Apology for Metro, or whatever you’re calling the next version.

 

*So what is it with all these asterisks? Well, it’s because, in Microsoft and Apple’s infinite wisdom, they’ve decided to save us from the threat of extreme power dissipation through USB ports with a new innovation called “port power management.” What this means is that the USB port, rather than delivering the full 500mA, or 1A, of power, as required by the USB standard, can be throttled down to use less power. Which plays merry hell with some DACs. Yes, including ours. Which means we get to educate everyone about how to turn off port power management, or, in some cases, ask them to go out and buy an externally powered USB hub to completely mitigate it. Maybe they should label the ports as “really full power/real USB spec,” and “battery-lifetime-promoting, save-the-planet USB port that doesn’t really provide full power.” No, wait, that wouldn’t fit. Never mind.

 

And, you know what? After writing that, I agree with Mike. USB sucks.**

 

**About 80% of our customer support is helping resolve Windows USB issues, from driver installation to port power management, so I’m biased. But now you know why Modi is locked down to driverless USB Audio 1.0 Standard operation.

 

 

USB 2.0, 24/192, and Beyond (and the Silliness of it All)

 

For Bifrost, though, locking it down to 24/96 wasn’t an option. 24/96 was becoming the de facto entry level for digital audio via USB. I wanted to start with 24/192 capability from Day 1. And I got that.

 

But USB still had some oddities. The earlier C-Media USB receiver chip didn’t work at 24/176.4, even though it did 24/192. Why? No idea. But at the time it wasn’t a huge consideration.

 

Fun fact: the CM6631A can easily do up to 32/384. Why don’t we enable it? Two reasons: (a) There is no 32-bit music, and never will be.*** (b) There is no 384k PCM audio for sale that we know of. Sure, there’s DSD 2X at 352.8k, but that’s a whole ‘nother discussion.

 

***There’s a famous napkin-scribble by a famous analog designer floating around out there on the internet somewhere, regarding the noise and precision of analog circuitry necessary for different digital resolutions. I can’t find it at the moment, but it went something like this:

 

14-15 bits: standard parts and layout

16-17 bits: attention to power supply noise, premium parts, careful layout

18-19 bits: extreme measures taken with low-noise parts, multi-layer boards, and exceptionally fine layout

20-21 bits: God’s domain

 

Fact is, 24 bits is 144dB dynamic range, which is about the limit of our Stanford analyzers. The best DACs, to date, manage 19.5-20 Equivalent Number of Bits (ENOB), even if they are “24 bit” or “32 bit” spec’d. 32 bits is 192dB dynamic range, which ain’t gonna happen, no way, no how, not even in temperature-controlled circuits sitting within 2 feet of solid lead shielding. Consider that a stun grenade is 170-180dB, and you’ll see how crazy this is.

 

Perhaps it’s a matter of capability. With SPDIF, we had some finite, and rather low, limits to amount of data we could transmit reliably in the past, especially if you were talking Toslink optical. That’s why Theta went to AT&T glass-fiber optical to get more bandwidth. Now, Toslink is better, but it’s a rare Toslink that can do 24/192 reliably.

 

But with USB 2.0, and even more so, 3.0, we have no such restrictions. How big a data rate do you want? How many bits? No problem. We can make up silly numbers all day. But don’t think it’ll be meaningful in musical terms, if, say, we can transmit 64/1.544Mbps bit depths and sample rates.

 

But, you know what? If you have a Bifrost, you don’t have to worry. If aliens from the planet Zebtron land on our world tomorrow, bringing physics-defying technology that enables 64/1.544Mbps audio transmission over USB, we’ll have a USB Gen X card soon enough to handle it.

 

But I wouldn’t hold my breath.

 

 

The Problem with Not Invented Here

 

Finally, let’s talk business. Thanks to Mike’s insistence on being different, we ended up with a truly unique DAC. And sometimes it takes that stubborn insistence, that rejection of everything “not invented here,” to make something great.

 

But “not invented here” can bite you in the butt, too.

 

It’s something to watch for, if you’re going to start your own business. Too much “not invented here” hubris can delay products, reduce efficiency, and interrupt operations. Sometimes the right answer is only an internet search away. Or a great idea might be just a small tweak to a similar product.

 

“But wait,” you say. “Are you saying…steal from other companies? Plagiarize?”

 

No, not at all. It’s a balancing act. You should be aware of what your competition is doing, how other people have solved problems like yours, and what the basic industry benchmarks are. At the same time, you should have your own ideas.

 

And—here’s the hard part—you have to have a feel for whether or not your ideas are better than the prevailing wisdom, and also if your ideas are realistic to implement.

 

“So how the heck do you do that?” you ask.

 

Believe me, I wish I had a formula. Some companies will spend tons of time benchmarking against their competitors and running focus groups to try to determine if they’re going to be successful, but I believe this is more likely to result in mediocrity rather than brilliance. You can’t assume your competition has all the right ideas, and you can’t assume a focus group is a microcosm of your entire prospect base. What’s more, you can’t assume that a truly great idea will make it through a focus group, because they’re more likely to be confused about something that’s truly unique, and has no point of reference.

 

Case in point: until the original iPhone was announced, everyone was wondering what kind of keypad and stylus it would use. Nobody guessed it would have neither one. It was simply insanity to consider it, at the time. Love or hate Apple, they changed the game.

 

I think the best way to decide on when to stick to your guns on new ideas—to be stubborn, and pound the table, and insist on “if it’s not invented here, it’s not for us,” comes down to weighing the risks and rewards.

 

Here’s an example, using Bifrost.

 

Rewards of doing what everyone else was doing:

 

  • Faster product introduction: If we’d done what everyone else was doing—using USB 1.1 and upsampling everything with ASRC, in a non-upgradable platform—we would have had a product out much sooner.

 

  • Easier development: we wouldn’t need code, integration, multiple boards, custom connectors, etc.

 

  • Easier support: no explanations about why the DAC has to be reset, resulting in the famous “clicking)

 

Risks of doing it like everyone else:

 

  • Prospect disappointment: people were already expecting great/different/innovative products from us, not a rehash of what everyone else was doing, so Bifrost might have flopped if it had been like a lot of others.

 

  • Undershooting the bar: if someone else introduces a game-changer right before your me-too product, you’re gonna be in a world of hurt.

 

  • No story: going with the crowd means you have no ideas of your own. No position. If you have no ideas of your own, what is your value? Sometimes you have to stand up and say, “This is what we believe in.”

 

In the end, we decided to stand up and say, “We believe in this, and we’ll take the pain to make it right.”

 

Which is good. Because it was a pretty painful path getting to the Bifrost introduction, which was several months late. But the anecdotal results are that we did something right. Have you noticed that most of the better DACs these days are avoiding sample rate conversion, or allowing you to turn it off? And how about all those inexpensive upgradable DACs that assure your new purchase isn’t going into the trash can after a few months…no, wait, that hasn’t happened yet…

 

And now, on to the next chapter. Our most humbling experience. And the closest we came to throwing in the towel.

post #702 of 14744

Whew ! Nice primer on USBs (for me at least). Jason, you're starting to write like Dan Brown :biggrin:.

 

Edit: In addition, I must say that the latest USB implementations on the Bifrost Uber and the Gungnir sound fabulous and i stopped using my USB-SPDIF converter a while ago.


Edited by kothganesh - 4/9/14 at 8:48am
post #703 of 14744

So, does the Bifrost use a custom digital filter, and can the actual DAC chip be upgraded (or is that included in the analog output board)?

post #704 of 14744
Quote:
Originally Posted by Currawong View Post
 

The health and safety stuff brings back memories. The warehouse I was in was a pretty tight shop when it came to rules (no daisy-chaining power cords even -- though in every other respect the place was total anarchy) but the importer across the park that handled incoming Dell and Apple shipments (among a crapload else) had a reputation for forklift madness, especially as they had quite a few gas-powered ones and a lot of gung-ho workers, one of whom got skewered... 

 

That brings up a famous story though:  An OHSE (or rather the Australian equivalent) officer was pulled over by a cop for speeding and given a $160 ticket. As the cop went back to his car, the worker pulled out his camera and took a picture, then went and fined the cop $900 for not wearing his fluorescent safety vest.


I work in manufacturing myself.  Stories like this make me laugh and cry at the same time.

post #705 of 14744

Another nice read - pure marketing genius. 

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Jason Stoddard
Head-Fi.org › Forums › Head-Fi Special Forums › Head-Fi Bloggers › Jason Stoddard › Schiit Happened: The Story of the World's Most Improbable Start-Up