When I need to do some serious thinking, perhaps about the design of a new system I'm building, or about some deep technical problem, I need things to be Just Right. I think you do, too. But we don't usually spend a lot of time thinking about exactly what "Just Right" entails. I have, lately, and it's starting to come together. That is, I'm getting close to the point where I can articulate exactly the set of constraints that must be met in order for me to innovate at my maximum rate, at a sustained pace.

So I'm writing an essay about it. Why would I start the essay before I know the answer? As Dr. Graham pointed out recently, "essay" means "try", and writing an essay is one of the best ways to pull your ideas together into a coherent and useful form.

Today's essay is, at least for now, about innovation. My essays don't always end up being "about" the topic I started with; sometimes they just don't work out and I keep them around as reference material but never finish them, and other times they'll start spawning sub-topics, until one of the mutant sub-topics grows large enough to merit its own essay. That might happen today. We'll see.

Keep in mind that it's not bad when that happens. One of the first keys to innovation is not to force it. Douglas Hofstadter has written a great deal about the process of innovation in his "Metamagical Themas" column, which he eventually collected and published in a very nice book by the same name. He points out quite rightly that you don't walk around thinking: "I need to have a great idea!" Innovation is something you practice every day; it's a skill that you can get better at, and one of the key elements to it is to let your mind wander at the right times. I think that writing essays, the "trying" kind, is one of those times.

Paper is RAM

The biggest reason I write blog essays is to help me think. In fact, relatively few of the essays I write are ever published; they languish in varying stages of completion, and sometimes I decide to split them up, rewrite one completely, throw one away, whatever. Sort of like the tools and scripts lying around in your personal /bin directory. Some are throwaways, some might be worth keeping if you polished them up a bit, and a few are genuinely useful tools that you might even share with other people. That's how writing is, for me: I do it just like I approach programming. Maybe a bricklayer would approach writing like laying bricks. Who knows.

When you're thinking your way through a problem, you're actually generating an idea-tree and searching it on the fly. Usually you're doing this in your head, although sometimes you're doing it while "thinking out loud" (not always a great idea), or when you're throwing up ideas on a whiteboard. But you're definitely doing it when you're writing an essay.

It's similar to doing a mathematical proof. You want to get from some known point A to a possibly vague point B, following some set of logical steps. This is what groups do in group discussions, e.g. in a hiring meeting. Point A is a set of data points from the interviewers about the candidate, and Point B is a decision about what to do -- offer, decline, re-interview, whatever. Because the path from A to B isn't always obvious, the group can sometimes get sidetracked down some mutant side-path, such as "why the heck don't we do pre-briefs?" or "how do you get features added to our recruiting tool, anyway?"

Side topics. They're important. Sometimes folks in a hiring meeting or debrief, trying to decide whether to hire a candidate, will suddenly reach the conclusion that they're all going to go off and add one interview question each to a particular Wiki page. Innovation has occurred, like a bolt from the blue. That's how it happens.

Here's the problem with thinking through problems in your head: you don't have enough short-term memory to do it very well. When you're building out an idea tree, you're basically walking down to some node in the tree, expanding out the child branches, and then choosing some order in which to visit them. You may be doing a BFS or a DFS, or iterative deepening; you might even be doing random jumps when you feel like you've gotten too far down some thought path that isn't productive. Heck, it's kind of like you're an artificial intelligence, isn't it? Funny how that works.

To expand out the nodes of an idea tree, you have to think them up (deductively, inductively, any way you can think of them), and then store them in your short-term memory so you can visit them in turn. You really don't have all that much short-term memory to work with. And your short-term memory is more like DRAM than SRAM: you need to refresh it periodically or it fades. That means you have to stop "thinking" fairly frequently, and walk your way along the tree again, from Point A to the intermediate node you've just reached.

Lest this discussion become too abstract, think of some time in the past few days when you were thinking through something like so: "Let's see... I know the person wants blah, and they wanted to help with blah, which is too hard, so I offered blah, but they couldn't take that because of blah, so I asked if blah was OK, which it was, and that means either blah or blah, so if it's the first blah, then..." Imagine that you can substitute anything you like for each instance of the word "blah" in that partially-instantiated thought chain. You get the idea. You occasionally need to go through and remind yourself of how you got to where you are in your thought process. And it's because your short-term memory sucks: like capacitors, the charge leaks out if you don't refresh it frequently.

Your short-term memory is also rather limited in storage capacity. You have lots of long-term memory, lots and lots, but it requires a tremendous amount of repetition in order to store anything in it. That's how your brain's neurons work. They gradually build up a sort of "charge" after being electrochemically stimulated enough times; it's a mechanism to keep you from remembering stuff that doesn't matter very much, presumably either for survival, or to keep you from running permanently out of room. Dunno. In any case, your long-term memory is more like a hard disk: it can hold a lot more, but it takes a big effort to update it. And it has the same problem as your short-term memory: you need to refresh it frequently or it will gradually fade, the way disks gradually accumulate magnetic errors.

Paper to the rescue. Paper, or Emacs. They both have their uses. When I'm in Emacs, I can type fast, especially with all my shortcuts, but I'm mostly stuck with text: language, that is. It's not as easy to draw pictures. Sometimes when you're thinking through a problem, doodling on paper or a whiteboard is easier. But if I want to express my thoughts in a language (usually English or a programming language), then Emacs is the best bet, since I write like a second-grader, except slower and messier. That's what happens when you type for a living.

In either case, I'll refer to the external persistence as "paper". It could be a whiteboard, a text file, a pen and paper, even another person that you're bouncing ideas off of, hoping some of them will stick. In a sense, group discussions (up to a certain size group) actually benefit from having slightly more temp RAM available.

Hmmm... I'm at a sticking point in the essay, which has been flowing quite nicely so far. I want to talk about two separate but related things, which runs the risk of derailing me, since THAT (getting your topic split) is yet a third topic that I'd love to write about. I think I'll just stick with the paper-as-RAM topic for now, and perhaps return to group-think later.

I'll get back to the main thesis (innovation) in a bit. Part of the reason the essay is flowing so smoothly is that my situation right now is Just Right. More on that later, I promise.

In any case, if you have paper available, you can use it as extra short-term memory. It's even better than short-term memory, because it will still be around the next day, after the current contents of your short-term memory have gone to /dev/null.

It helps to be able to type fast. Really, really helps. If you want to use external persistence to help you think, it's not much help if you only type 20 or 30 words per minute. You need to be comfortable doing at least 60 or 70, preferably better than that. I think I've hit a pretty fair pace of about 90 wpm for the past few paragraphs. Helps. You can learn to do this; it doesn't take that much practice. Half an hour a day for a few weeks. Try it!

When you're using paper as RAM, you've increased your capacity to think. You've given yourself a hardware upgrade. Pretty cool, eh? Someday they'll be able to give you the real thing. Sign me up; I'll be the first to try it. Big memory chip in the back of my neck, possibly with a thin glass "inhibitor" sticking out that keeps the mechanical arms from going nuts and turning me into a monster.

Anyway, if you want to innovate, you need to increase your capacity to think through difficult problems. That means you need to learn to write as you think, and not have it get in your way. It's a tool that you need to practice with.

Why essays? I used to do all my thinking-writing in outline form, since it helped me structure things; I could sort of see the idea tree forming that way. But I don't do that as often anymore; I prefer narrative, and I think it's because it forces you to be more disciplined. To write an essay or narrative, you need to turn some wildly nonlinear tangle of thoughts -- a tree, graph, or perhaps a disjoint set -- into a linear progression. That means choosing a particular path through the idea tree and walking through it methodically.

I'm not entirely sure why that's better. Maybe it's because the end result is easier for other people to digest, and innovation is rarely useful unless you can get other people on board with your ideas. Having the end-result of your thought process be a readable essay is certainly preferable to handing people a digital-camera snapshot of your whiteboard, which you wrote all in red dry-erase marker, illuminated by our dim red epilepsy-causing cheapo fluorescent lights, then filtered by your camera's redeye-elimination mechanisms. I've tried it: whiteboard pics don't come out too well. You want the essay.

OK, I think this section has gone on long enough. You get it, though, right? Paper is RAM. It's a useful trick. Let's move on.

Just-Rightness

I'm talking about "just right" in the Goldilocks + Three Bears sense. Not too hot, not too cold. Just right.

Innovation requires thinking. Yes, you need to be able to have those rare and often highly productive 2- and sometimes 3-person brainstorm sessions. But you can't innovate purely with group brainstorming. People get tired, confused, sidetracked; you can only go for so long in a group discussion before someone poops out. And it might be you.

It's because of another weird problem with the way our memory works: you can only think and/or absorb so much before you have to get some sleep; that's when your brain is organizing stuff, compressing stuff, indexing it, whatever magic it does so that you can fetch astonishingly minute facts on demand from your gigantic memory store. Like the name of that actress, the one who played Queen Somebody, Mrs. Brown something, the one who was the Air Elemental in the Riddick movie. Yeah, her. No, I can't remember her name either. Isn't your memory just the most awesome thing?

If you're too hot or cold, it's hard to think. If you're hungry or thirsty, or have to go to the bathroom, or your foot hurts really bad from accidentally smashing your pinkie-toe opening the door this morning, it's hard to think. It's not impossible, of course, but each little distraction occupies just a little bit of memory and processing time that you can't devote to the problem at hand.

So you have to be comfortable.

Your short-term memory is an odd beast. For one thing, it acts like a cache, but it's not really an LRU cache. It appears to discard things randomly in order to make space. Let's say you're having dinner with a friend, and you start explaining some interesting topic to the friend. Maybe you're explaining how much you like the show "The Apprentice", and you're using one of ol' Donald's techniques as an example of the way we should be running the company. (Hypothetical example, of course. Heh.)

As you build your way along the path to some conclusion (e.g. "and THAT is why we should bring half the team to the boardroom each month and fire someone", depending on how much you've had to drink), it's possible to blow your stack. Your path-length gets too long, and you get too carried away running headlong down some expansion of the idea-tree, and BOOM!

You forget what you were talking about. What was it again?

How do you blow your stack like that? We do it all the time. Some people recover from it better than others, and are masters at pretending that their most recent landing spot was, in fact, the intended destination. Other people get all red-faced and announce "uh, I forgot what I was talking about" in front of 50 people. We all have our moments.

I think you blow it by overextending your path -- that is, the proof-chain of ideas taking you from Point A to Point B, and at some point your garbage collector came along and tossed out one or more links in the chain, to make room for the exciting new sensory input that's telling you loudly that your boss appears to be fixedly staring at the zit on your forehead. That one bit of information has caused you to discard one of the steps along your idea-chain.

*If* you can remember what Point A was, then you may be able to quickly reconstruct your steps. But if your garbage collector discarded a link relatively close to where you just were, you may have no idea what Point A was, at least for the moment. Your friend might not remember either, especially if the distraction was mutual, e.g. the waiter coming by with a plate and saying "Who ordered the PORK LOIN?" The stack may be mostly obliterated by that kind of interruption.

External interruptions are bad for innovation. More on that later.

Here's the weird thing -- both you and your friend may have had different links discarded from the mutual thought-chain. But you both probably have a pretty good idea about how long the chain was.

And it is mutual, although they're not identical; your friend is busy trying to understand what you're saying, which involves listening to you, watching your body language, picturing what you're saying, humming "laaaaaambert, the sheeeepish liiiion" in the background because they can't get it out of their head, and wondering if the pork loin will be here soon. But for the most part, assuming your friend is trying to follow you, the two of you will have a fairly similar representation of the conversation so far stuffed into short-term memory. When the waiter with the pork loin comes by, it's fairly random which bits will be blasted from each of your storage areas.

If neither of you can remember, it's possible that your friend might say: "I think we were talking about Donald Trump." Let's say that Point A was, oh, hiring practices, and Point B was going to be that we should run our performance program like a "reality show." And Donald Trump was somewhere in the middle there, let's call it Point A'. Your friend may remember you were saying something about Donald Trump, which may reconstruct part of the path for you, but if neither of you remembers what Point A was, things become momentarily awkward.

Because you both know there was a Point A, and it was before Donald Trump came up.

You may both agree that yes, we were talking about Donald Trump and Reality Shows, but the fact is, you both remember that there was some long-ish chain of thought before Donald came up. You just can't remember what it was.

That implies that you've got some sort of camera, some extra thread, that's recording meta-information about your thought processes. It may remember that at one point you both laughed out loud, or that one of you said something really loud or controversial, or that at one point you were interrupted repeatedly by other diners. And at all times, it appears to be keeping track of how much time you've spent talking.

Isn't that weird? And you're not controlling it directly; it's running more or less automatically for you. But it's the process that knew that when you blew your train of thought, and your friend reminded you of Point A' halfway along, that you were only halfway along.

Hopefully we've established that you really don't want to rely too much on your short-term memory when you're trying to think (or talk) through a problem. Or to put it another way, you want to minimize interruptions, so you run the least possible risk of losing some crucial path through your idea-tree, and forgetting what you were thinking about. You might lose that idea for months.

So you want to be comfortable, you want your surroundings to be quiet and interruption-free, and you want to have some way to persist your ideas, to let your brain focus more on computation and less on simple DRAM-refresh activities. OK. What else goes into being Just Right?

Well, being comfortable involves more than just not being hungry or too hot or cold. You need to be physically comfortable -- the more comfortable, the better. In the ideal case, you wouldn't have a body at all; you'd just temporarily be a brain floating around in thought-ether, solving problems without a care in the world. But that's sort of unlikely, so at least you could ask for a comfy couch or a squashy armchair. With some pillows, maybe. If you can get comfortable enough, it's almost like being a brain in a jar, right? For all we know, it might even be better.

Speaking of not having a care in the world, it's also hard to think when you're nervous, or anxious, or stressed, or distracted. Those things are harder to control, and are often caused by non-work-related issues. But I think work is just as often the culprit. Work can be a source of stress. Part of a good manager's job is to remove as many stresses and other distractions as possible, to free up your mind and give it the resources it needs for innovation.

If you have a sev-1 or sev-2 pager emergency to deal with, you've pretty much violated just about every tenet of Just Rightness. It's an interruption, and it stresses you out, and it hurts, and it can often take precedence over hunger and calls of nature, at least for a time. That's not to say that we shouldn't respond to emergencies, but people should recognize that the first thing to go down the toilet during a sev-1 is innovation.

Finding JustrightLand

So where is the best place to innovate? I'll tell you: right this moment as I'm writing, on a scale of 1 to 10, my Just Right factor is sitting at about a 9.0; i.e., pretty darn good. I'm at home, lying on a couch with my laptop in a comfy position. It's 4:04am, so it's incredibly quiet, and the lights are mostly off, except for a few candles. My laptop is bright, fully charged, and has wireless internet access -- a small point, perhaps, but it was useful when I went to look up the name of that new Johnny Depp Peter-Pan movie on IMDB just now. I had a nice dinner earlier, and the temperature now is neither too hot nor too cold.

It's Just Right.

I could no more have written this essay at work than I could have solved Fermat's Last Theorem. Work is a constant distraction, especially for managers, who operate in permanent Interrupt Mode. Managers don't innovate, or if they do, it's in the shower before work, or perhaps during the boring commute to work, or late at night after the kids have gone to sleep. But not at work. Not much, anyway. The environment just isn't right for it.

I often wonder what the ideal workplace would be like. I tend to operate under the assumption that if the workplace were closer to Just Right, then people would be more effective. I've read about studies that showed this to be true, e.g. in Peopleware and similar books. But it's so hard to measure that all you can really do is go on anecdotal evidence and experience. But I believe that a workplace that is Just Right for innovation is likely to be a more productive one.

What would we add to a workplace in order to make it Just Right? I'm assuming that everything that applies to individual innovation also applies in the workplace -- you want quiet, comfortable, distraction-free surroundings, with lots of paper and whiteboards for taking notes as you think. You'd just want to add in having some comfortable meeting rooms where people can meet as needed. And then arrange things so that people, for the most part, worked quietly as individuals.

Interestingly, most office working environments are in many ways the complete antithesis of what I idly envision as being "Just Right" for innovation; i.e. having the set of properties that are most conducive to innovative thought, good design, deep thinking, solid engineering, and so on: anything that involves thinking, which, we've established, only happens if you can actually free your brain from all other distractions.

In most office environments, you're surrounded by other people, and unless you're a manager, you have no door to close when you really need to get work done. Ironically, it's the managers that need the closed door the least, except for having private 1:1s, because managers have the most fragmented workdays.

The temperature and/or ventilation are often messed up, particularly at night and on weekends, which is exactly when innovators are lurking about the office, because that's when it's quiet and distraction-free. So you have to trade one kind of distraction (noise) for another kind (atmospheric discomfort). For what it's worth, you can at least find the ventilation overrides if you know where to look. Nice to be able to breathe.

The furniture is uniformly uncomfortable. It's almost impossible to find a comfortable location to work in today's office environments. Certainly not in your office, unless you've hauled your own dingy old couch in, but not many people do that. (It's worth the effort; I can tell you that much.) By "comfortable", I specifically mean "squashy furniture" -- furniture that doesn't protrude sharp angles into your body, or threaten to sever an artery on some metallic joist. Remember: Brain in a Jar. Sometimes you need to be able to lie down. Possibly even close your eyes -- vision itself can be a serious distraction when you need to think really hard about something.

It's hard to find meeting rooms. It's hard even to go offsite with a friend to have a design discussion without running into a bunch of coworkers. People are typically compressed into their workspaces like cattle, and last I checked, cows don't innovate much.

And let's not get me started on how hard it is to go to the bathroom. The steady state of restroom stalls, like hard disks, is full.

It's not just at Amazon; it's most of Corporate America. Although to be perfectly fair, Amazon's pretty far below average; folks in the movie Office Space appeared to have more creature-comforts than Amazon's facilities offer.

If you knew you had 18 months to build the world's greatest XYZ, and you could hire a world-class team of scientists and engineers to build it, what kind of facility would you set up? To cite a real use-case, let's say you're the U.S. Government, and it's 1943, and you need to build an atomic bomb. So you entice (not draft; entice -- they're all far too old to be drafted) the world's leading scientists to come work on it. How would you make them comfortable?

I've often idly wondered what the ideal innovation environment is. I know it's not at home -- staying up all night just to get some free time for innovation doesn't scale well when you have a family and a day job. Home has its share of distractions. And I know it's not in the office, at least not the space we're in today.

Every time I let my thoughts wander on this subject, the same vision comes to mind. I'm sitting in a comfortable armchair, in a large room or atrium that's lit primarily by daylight coming through large windows. There are nice green plants everywhere, including a large fern-ish looking tree (apologies; botany has never been my strong subject) next to me. There are people coming and going nearby, but their noise is white noise, and it's not terribly distracting. I've been in this place a hundred times, in a hundred different locations.

It's a hotel lobby.

Much of the best work I've ever done, the best writing or design work or whatever, has been done in hotels on business trips, usually down in the lobby, or near the pool if they have one. Sometimes it'll be in the atrium of a nice airport, when my plane is delayed for several hours. But it's always a big, comfortable, temperature-controlled, anonymous place, with daylight, plants, clean and highly available restrooms, some food vendors, and wireless internet access.

I've thought about this on and off for years, and I always come back to the same sorts of locations. I've worked in hundreds of venues, in and out of the office and home, and hotels are the best. They win, hands-down.

If I were a CEO pulling together the world's top scientists and engineers to build something incredible, something that required the very best innovation and work from all of them, I would put them in a hotel. Or something an awful lot like a hotel -- you could obviously build a corporate campus that had many of the same amenities. But for a one-shot deal, you can't beat a hotel for convenience, since it's already built.

Innovation Heaven

When I refer to "innovation" in this essay, I'm mostly talking about programming and software design, and to a lesser extent writing. I have no idea what a chemist's ideal venue for innovation is -- probably somewhere with concrete walls, and workbenches that don't burn or corrode easily. But for gold-collar innovation, the kind that involves a keyboard, I know what Heaven is like, because sometimes I get to work there.

In Heaven, also known as a Four-Star Hotel in a good city, you get your own office, and not only does it have a door, it comes with a little sign you can hang on the door that says "Do Not Disturb." Just typing that gives the programmer side of me weird pangs of longing.

In Programmer Heaven, your office has a window. On good days, it looks over tall mountains by the water; on bad days, you can see a brick wall 5 feet from your window and nothing else. But there is daylight during the day, when you need it, and when you don't, there are thick, heavy curtains you can draw to cover the window. You have absolute control over the lighting level.

Your office has a bed and a squashy armchair. When you're thinking about something that feels nearly impossible to solve, and you've thought until you practically can't stand it, you can grab a technical book, lie down, and read until you fall asleep. When you wake up, you have your answer. That's a nice office.

Your office has a door that leads directly into your own personal bathroom, and there's never anybody in it. And it's clean. 'nuff said.

In Programmer Heaven, your office comes equipped with a high-speed internet connection -- a wireless one, and a company-issued Macintosh Powerbook laptop connected securely to your corporate network via the Wireless virtual private network. There are no cheap HP-issue Windows laptops in Heaven.

There are meeting rooms galore, and your co-workers are located all around you, in their own private offices just like yours. Although it's considered gauche, you can call them and they can call you. Mostly, though, you communicate via the computer -- email or IRC, typically -- and via occasional meetings you have during the week.

Programmer Heaven has a Concierge, and Room Service, and a TV that you can plug your GameCube, PS/2 or XBox into, for those times when your brain needs you to go away and do something mind-numbing while it indexes your day's data.

In Programmer Heaven, you'd rather be at work than at home. You work all the time, in fact. Your innovation stretches are no longer measured in minutes or hours; they're measured in days, during which time you tell your family: Sorry, but this is important. I'll be home for the weekend.

You work constantly, and your mind becomes so involved with the problems you're working on that even when you're soaking in the jacuzzi, or running on the treadmill in the gym, you're still thinking about the work. You may have your most surprising innovative insights when you're nowhere near your computer, because you're in the Zone.

Incidentally, you heard right -- Innovation Heaven also has hot tubs, a gym, a pool, and a haircut place. And dry-cleaning. All of your day-to-day needs are attended to. Minimum distraction, maximum productivity. Brain in a jar.

The extent to which a company can achieve this kind of environment is potentially the most significant constraint on how innovative the company can be. Certainly the environment drives recruiting -- you could attract some serious talent, the kind that doesn't need to work for a living, if you could set up Programmer Heaven and give your world-class brains-in-jars something meaty to work on.

Even if you can't (or more accurately, won't) set up Programmer Heaven, you can at least try to avoid setting up Programmer Hell: an environment that's so easy to imagine, you don't even have to close your eyes.

I don't know how long Programmer Heaven could actually last. I suspect "weird" stuff would start to happen if your programming staff was in a Four-Star hotel for more than about 18 to 24 months -- families moving in, extra distractions, I'm not sure. It would have to be monitored, certainly. But if you've hired a world-class engineer and you're putting them in a world-class environment, maybe you really are justified in doing a regularly-scheduled Donald Trump-style firing, just to keep people from becoming complacent or feeling a sense of entitlement.

In any case, I'm fairly convinced that if you want a Los Alamos-style project, and you need the best people in the world to build the best "thing" in the world, in some thing-category you care about, then a 4-star hotel with wireless VPN is just about the best possible environment you could possibly put them in.

So I costed it out, just for fun. You could easily negotiate a deal where you get a room per engineer, some dedicated conference rooms, free parking and access to the standard hotel-guest amenities (e.g. gym, pool) for about $36k per year per engineer. Probably less; that's based on a rate of $100/night, which I'm sure you could improve on. You'd still have to pay the up-front cost of equipping the engineers with the Powerbooks and other office materials, but much of the standard "overhead" (desks, chairs, etc.) is supplied along with the room. Good hotels cater to people who do their work on computers.

Presumably such a team would have a hardware-free facility, with all data servers being hosted by an ISP or ASP, so really all the engineers need is rooms, laptops, a network, and conference rooms. Compared to the lease rates for office space in the Seattle area, the hotel is in the same ballpark, and well worth it if you're really bringing in the best people in the world.

All you budding young investors and entrepreneurs out there, take note. Programmer Heaven is relatively inexpensive, and might give you amazing results.

As for all the rest of us: well, one can always dream. Speaking of which, it's 5:16 am. I b'lieve I'll go do some dreaming right now.

(Published Dec 07 2004)


Comments

>>> In Heaven, also known as a Four-Star Hotel in a good city,

It's hard to fault your logic, but I've always found the rooms and lobbies of even expensive hotels to be pretty sterile and often noisy (lobbies, and hallways outside rooms).

Wouldn't you rather have the tranquility of a room in a quiet country inn, in say the Cotswolds, or Shropshire? I've stayed with friends in a small cottage just outside this village: http://www.leintwardine.info and I got a lot of thinking done, _and_ came away relaxed and happy.

You probably won't get broadband yet, but I'd trade bandwidth for quality-without-a-name beauty any day. If you're really lucky, there won't be a TV either.

That would be my ideal for deep thinking on a known problem. But what about pure innovation? Surely the most important factor there is who is working nearby, and what they are working on? e.g.

http://www.amazon.com/exec/obidos/asin/0887309895

And from http://www.research.ibm.com/journal/sj/384/weiser.pdf...

We placed a ParcTab on the CSL coffeepot. Whenever anyone made a new pot of coffee we pushed the reset button on the ParcTab, which sent an infrared signal to the computer network. A message popped up on our computer screens letting everyone know that there was a fresh brew. This caused an instantaneous gathering around the coffeepot, and as a result generated lots of fresh hallway discussion which is one of the best ways to create new research ideas.

Posted by: Chris N. at December 10, 2004 08:03 PM


"Peopleware" talks about a manager sending the entire dev team to a beach house for a week, to create the architechture and design for a new product. The manager stayed at the office to handle phone calls.

Posted by: Derek U. at December 11, 2004 01:00 AM


Beach houses, country homes -- I can certainly see those venues, as long as (a) the surrounding entertainment (hikes, kayaking, etc.) isn't too distracting, and (b) you can actually get errands done when you need to. Being in some downtown area in a hotel seemed like the best solution to all the constraints I had in mind. But there are all kinds of places that could work, and I agree with the Quality Without a Name (QWAN) being really important.

I worked for about 18 months on a project at HP Singapore, and they flew us out frequently to work with the engineers there. They always put us up in the Westin or in the Shangri-La. Both hotels were amazing, more like resorts than hotels. We used to think HP were total fools for spending all that money, but we eventually realized that it made us *want* to go out there and help. We didn't mind staying for 2 or 3 months at a time, although after that even the Shangri-La could wear on you. But in retrospect, HP was pretty smart to put us in 5-star hotels; we would have been leaving after the first week or two if we'd been in anything but heavenly conditions. I doubt we'd have launched the projects on time (and we did) if they hadn't been able to lure us out there for so many weeks and months at a time.

Shangri-La Singapore has QWAN.

I hope everyone realizes this blog entry (other than the paper-as-RAM part) is just idle fantasizing, and it would probably never happen in real life. But it can be fun to imagine!

Posted by: Steve Yegge at December 11, 2004 01:21 AM


John Carmack is known to take a sabbatical right around when the latest id software shooter comes out. He hunkers down in a hotel far away, laptop in hand, and just hacks and tinkers and research/prototypes new ideas.

Posted by: Andrew W. at January 19, 2005 10:50 PM