On Prompt Engineering

I have a visceral reaction to people talking about "prompt engineering". This is an attempt to reflect on why this phrase elicits such a strong negative emotional response in me. There are a lot of things to criticize about the current AI hype and its impact on the web, the environment, the economy and society at large. That’s not my focus here. I want to instead evaluate what I understand good (software) engineering to mean and contrast that to the activity of writing prompts.

What are some properties you want a software system to exhibit? I’d argue the list should include at least:

• Predictable. My mental model should be aligned with the software so that I can with reasonable certainty predict the outcome of calling any one function in the system. This is very much not the case for LLM prompts. Replacing a word with a synonym, fixing a spelling mistake or changing the order of statements can all have dramatic and unpredictable effects on the result.
• Deterministic / Repeatable. The software should be deterministic. Providing the same inputs to the system should reliably produce the same outputs. LLMs have built-in randomness euphemistically referred to as “temperature”. Execute the same prompt multiple times, get a different result each time.
• Inspectable / Debuggable. I should be able to pop open the hood, look at the code, and figure out what it is doing. If I experience bugs or unexpected outputs I should be able to trace back through the code to understand where and why the logic went wrong or differed from my expectations. None of that is true for LLMs. If I get an unexpected result all I can do is permute my prompt and hope for the best. I can’t inspect the inner workings of the network to figure out which part of the input was responsible for the deviation.
• Composable. Software systems should be building blocks I can combine to build something bigger than the parts. LLMs are all or nothing. Their interface is effectively all of human language as input and output. Multi modal models go even further to include other modalities like images, video and audio. This is too wide and deep an interface. By its very design, an LLM is not only a monolith but also perfect spaghetti. Everything potentially affects everything else. There’s no isolation in the network, no re-use of just parts of it is possible.
• Stable. Software should be stable across versions. LLMs change in random unpredictable ways from one release to the next. A prompt that provides good results now may become entirely useless with the next generation of the model. Numeric embeddings produced with one version are meaningless and incomparable to ones produced with a different version. Great fun to start over each time.
• Testable. Software should be (unit-)testable with fairly high coverage. LLMs offer a single entry point for a huge set of functionality. The input and output space is enormous. It is utterly hopeless to achieve anything approaching good test coverage. All you can do is shine a flashlight in a vast dark ocean and hope that your tiny collection of training examples covers all cases relevant to your problem.
• Efficient. Software should be resource efficient. The principle of least power applies. LLMs are massive resource hogs and you activate large fractions of them no matter how simple or complex your prompt is.
• Fast. Most software counts QPS - queries per second and latency in milliseconds. LLMs count QPM - queries per minute and latency in seconds. Maybe fast enough for an interactive chat bot. Painfully slow for working with large data sets. Especially if conjuring up prompts is so random that responsible development involves a ton of repeated experimentation and testing on large validation sets.
• Precise. Software interfaces should be precise. Human language is not that. It is inherently ambiguous, redundant and open to interpretation. There’s no way to ensure the LLM will choose any one particular interpretation of a statement.
• Secure. Software should be robust against injection attacks, leaking data and other safety concerns. Given the properties listed so far, it seems highly doubtful that this can ever be guaranteed for prompting LLMs. How can we pretend to secure a system we don’t understand and can’t test comprehensively?
• Useful. This is arguably the one point where LLMs make up for not ticking even a single one of the other boxes. They do solve problems where we don’t currently have any alternative approach for a solution.

Prompt engineering is hard for all the wrong reasons. Don’t take my word for it. Take a recent (December 2024) paper The Prompt Report: A Systematic Survey of Prompting Techniques which surveys more than 1500 other papers to compile a taxonomy of prompting techniques. Everyone is just blindly pushing and prodding the machinery, hoping to tickle it in just the right ways to provide useful results. Consider choice quotes like “exemplar order can cause accuracy to vary from sub-50% to 90%+” or “providing models with exemplars with incorrect labels may not negatively diminish performance. However, under certain settings, there is significant performance impact” or the summary “prompt engineering is fundamentally different from other ways of getting a computer to behave the way you want it to: these systems are being cajoled, not programmed, and, in addition to being quite sensitive to the specific LLM being used, they can be incredibly sensitive to specific details in prompts without there being any obvious reason those details should matter”. This is utterly ridiculous and shows our fundamental lack of understanding of cause and effect with these things.

Prompt engineering is performing alchemy in a chemistry lab. It is an insult to the chemists.

So yeah. Devising good prompts is hard and occasionally even useful. But the current practice consists primarily of cargo culting and blind experimentation. It is lacking the rigor and explicit trade-offs made in other engineering disciplines.

Gemini 1.5 Flash: "Draw a brightly colored cheerful cartoon robot with an idiotic grin on its face and bulging bloodshot eyes. It should swing a hammer and wrench in dangerous and clumsy ways."

Civilization VI - it could have been great

Queen Victoria on my Western border has just offered me a joint declaration of friendship. Trade between our empires, England and Germany, is flourishing. Peter the Great of Russia to the East is a different matter. He keeps denouncing me for being behind in both technology and culture. He's a wonder building machine and keeps churning them out one after the other. My greatest worry though is Norway across the great sea. Harald Hardrada is leading the world's most technologically advanced nation and is already starting to work on space race projects before anybody else has even unlocked the secrets of rocketry.

The tensions with Russia escalate and war breaks out. Peter commands his unique units, cossacks, a frightening force in a land war. I take advantage of a strategic blunder and shell Moscow with field cannons and artillery. Once the capital falls the rest of the country is soon to follow. Peter admits defeat and my empire has just doubled in size. The spoils of war include all of the world wonders Russia built and a treasure trove of its cultural great works of art, music and writing.

Russia sending all its troops on a sailing trip while I get ready to invade the capital.

Meanwhile Norway has become dangerously close to winning a science space race victory while America under Teddy Roosevelt is working towards a cultural victory. They already lead the rest of the world in tourism by a huge margin. I send all of my spies to Norway to infiltrate their major cities and sabotage science and spaceport construction. Trying whatever I can to delay progress on their mission to Mars. While James Bond is performing the subtle art of sabotage for me I'm ramping up bomber production at home. I have discovered a source of uranium and the Manhattan project is well underway. My plan is to Blitzkrieg Norway and nuke them back into the stone age.

My major port city of Hamburg has constructed the Venetian Arsenal world wonder, producing two fleets for the price of one. This allows me to quickly grow my armada and get ready to cross the ocean to invade Norway. At this point Harald is launching one more rocket away from victory. I'm still decades behind in technology and stand no chance of catching up. My aircraft carriers are loaded with long range bombers and, protected by my unique U-Boot units, race to reach the coast of Norway.

Nuked Norway. The tool tip on the right shows how close they came to victory.

I declare a surprise war and in a single turn nuke the major cities of Ålesund, Nidaros and Sarpsborg while at the same time bombarding all coastal cities from battleships. This war is brutal and swift. Within just a few years I have conquered all of Norway's cities fending off a premature end of the game just in the nick of time. But there's no time to rest, my empire is spread thin financially, having to support the new Viking colonies across the ocean. America is creeping ever closer to a cultural victory and the entire world now thinks of me as a villain and warmonger.

It doesn't take long before half the world declares war on me and the other half refuses to trade or even talk to me. This is bad - already strapped for cash I'm going bankrupt without international trade. My military units desert and disband and my cities revolt against my rule, spawning rogue units roaming around the country, pillaging my infrastructure. I become even more ruthless. I have since upgraded from regular nukes to thermonuclear devices and don't hold back on using them. The major English metropolitan areas of London, Manchester and Birmingham all get reduced to radioactive rubble. I follow up by sending ground troops in and razing them to the ground, irrespective of the fact that my own troops are also dying from the radiation.

My bankrupted empire after the war. Loosing 400 bucks a turn and spawning all these red insurgent units everywhere.

America is out of reach for my military so I keep sending my spies to perform heists of great works of art. This should postpone their cultural victory for a while yet. My own path to victory is to either burn the entire frickin' world to the ground and be the last man standing or build the spaceship myself, finishing what Norway started. In the end I build the spaceship and fly to Mars, winning the most exciting game of Civilization VI yet.

Fuck this planet. Let's go to Mars instead.

This is Civilization VI at the best of times. According to Steam I've spent 59 hours playing it. I've won the game by Science Victory, Religious Victory, Culture Victory and Domination Victory. I've won the game playing as six different nations. I've won the game on Prince, King and Emperor difficulty. I've won the game on Small, Standard and Large maps. I've won the game on Continents, Inland Sea, Island Plates, Shuffle and Pangea maps. I think I have covered what it has to offer pretty well. And still I hate it. It is fatally flawed from a user interface, technical and AI point of view.

User interface:

• The interface has race conditions where you have a unit selected and give it an order but the auto unit cycle jerks the unit away from you just then and auto selects a different one so your order goes completely haywire. Hard to describe but infuriating in practice.
• The user interface ergonomics are terrible. You are constantly made to travel huge distances with your mouse for no good reason. An example is placing spies: you click end of turn at the bottom right corner of the screen, then select a spy mission at the top left, followed by bottom left and top left again. Why?!
• There's a lot of mindless busywork like placing defensive spies, trade routes and spreading religion. These work OK at first, but once your empire grows it's mind numbingly repetitious. Spies and trade routes "expire" after some number of turns and you have to set them active again. Why not just allow me to give them a mission and stick with it until it's either no longer possible/valid or I change my mind and assign a new one? Why do I have to remind the stupid spy every N turns that yes, you are still supposed to defend the capital and yes, that trader there should still bring production to my city.
• There are no usable overview screens. Not for my own empire (Really? We don't maintain a spreadsheet of how many U-Boots we have and where they are?!). Nor for the state of the world (Who's at war with whom? Who has what to trade?). The information is available, but it requires a lot of work to get to. I miss the clear charts of Civilization IV.
• The game looks gorgeous, but all that detail often obscures the most important information. I have forgotten about a battalion of tanks in the middle of a war simply because they are so hard to spot in all that's going on on the map.
• There's a complete lack of keyboard shortcuts. I can't move units on the map without using the mouse. Bombing runs require way too many clicks.
• You are forced to sit through useless audiences with the AI players all the time. The game pauses and loads the animated AI character just to deliver a single sentence like: "Your army is weak!". Why can't I decide whether to receive a delegation? Ironically the AI can refuse to see me, but not the other way around.
• Lack of feedback for critical events. Why am I suddenly at peace with this city state? Ah yes, it's because the Suzerain changed.
• Outright bugs. An AI talks to me and during the negotiations we declare war. Once I leave the diplomacy screen it's peace again because we weren't actually able to declare a war due to some existing contract.

Technical:

• Loading times are insanely long. I usually have my cell phone ready and browse Facebook or read news while the game is booting up. I always play ironman, not ever reloading during the game (a leftover from playing Civilization exclusively as multiplayer) - I don't want to imagine how frustrating the game must be if you save game cheat and reload frequently. In fact, that's the reason I didn't put more screenshots in this post: I can't be arsed to wait for the game to load.
• Turn processing takes an eternity. I complete my turns faster than the AI does. Which is especially bad, because:

AI

• This is where the game falls completely flat. The AI is just terrible. Firaxis has designed a game that its own AI doesn't know how to play. It feels as if a checkers algorithm was competing at chess. It's utterly hopeless. In all of my games I have never even seen the AI conquer a non-city state city. They are constantly at war with one another or the player but most of the time you don't even notice because it gets all of its units stuck in a choke point and won't even make it to the battleground. Completely ineffective. My war with Russia in the description above was symptomatic: Russia had stronger units, the larger military and a huge bonus to unit strengths from the difficulty level I was playing at. Yet, while I was getting my artillery lined up to shell its capital, Russia decided to send all its land units out to see. Embarked land units can be picked off one by one with battleships and that's what I did while the units were just floating there for no reason whatsoever and with no destination to go. Same for Norway. Norway was technologically way ahead of everybody else and I was expecting strong resistance. Yet when I arrived at the coast they didn't have a single defensive unit. The only reason they made it so far ahead in research was for the advantage the AI gets over the player. The cities weren't even improved - only the most rudimentary infrastructure in terms of farms and mines and such. Yet every single village had a spaceport, nevermind you can only ever use three at the same time or that a rural village of a handful of farmers will take centuries to build a rocket. The AI is a clusterfuck of epic proportions and fails at every single step of the way.

This is a tragedy because I can see the potential. Many innovations in this iteration of the series are really nice. I like city districts and how they make the terrain more relevant. I like that they got rid of the dedicated city screen and made the map the central UI element. Policy cards offer a detailed and nuanced way for setting your governmental policies. The way traders automatically construct roads takes away a lot of micromanagement. It looks gorgeous. And yet... sigh.

I love the series and I really want to like this game. Alas, while it is a step up from the equally flawed Civilization V, it is no match for the masterwork that was Civilization IV. I spent literally hundreds of hours on that one (luckily this was pre-Steam and I don't have exact counts ;-)). It got so many things right from a game design perspective. I used to play that with a chronically ill friend in Germany. He has since passed away from the mucoviscidosis he was suffering from, but while he was still alive, Civilization was our way of keeping in contact despite being in different countries. We usually played cooperative games against the AI at difficulty levels that saw us lose more often than we'd win. Making the rare victory feel so much sweeter and deserved for executing a flawless strategy over many hours. I remember us beating the game in a domination victory one turn ahead of the AI reaching a space race victory. It's a turn based, slow paced game, but it can still get your palms sweaty and your heart rate up. Triple so in a competitive multiplayer setting ;-) Firaxis, please
Make Civilization great again!

2015

2015 was one of my best years yet. Anita and I got engaged; I mostly recovered from my accident; I left my old team at Google with my strongest performance rating yet; the new team and project are incredibly fun; we went for a week of skiing and wellness in Austria; a week of climbing on Crete; a week of hiking through the Swiss alps; a beach vacation on La Gomera; an extended weekend in Vienna; skiing in Engelberg, Laax, Andermatt and Flims; Rock Climbing Rivella, Sunneplättli, Engi, Weesen, Roggenflue, Siggenthaler Flue, Sobrio, Roggenhuser Steinbruch, Balmflue, Leiteregg, Lungern and Brüggler many of them with more than one visit; I completed my fourth Rigi Marsch; I went hiking and mountaineering to Lagh de Cama, Arena Sardona, Ulrichshorn, Säntis, Nesthorn, Gross Furkahorn, Balmhorn, Krönten, Bristen, Gulderstock, Montalin, Brisen and Buochserhorn; we swam across Lake Zürich and bashed our bums sledding. All of that in the company of great friends and family.

Here's to an even more awesome 2016!

Why?

Mountain climbing is dangerous. It requires suffering through pain, cold and exhaustion. It serves no purpose. Why do it then?

Many people more qualified than me have tried to answer this question. This is my rambling attempt at a more personal answer. Why do I go to the mountains?

I love being outdoors. I love the connection to nature, the rawness and wildness of it. I love throwing myself into the wind and shouting of joy. The mountains and the sea are two of the few places left in our civilized and tamed world that are still mostly wild. Mountains are raw energy. Inconceivable amounts of energy created them. You need to be strong to climb them. Everything there is charged with energy. Everything there has the potential to kill you if the energy is discharged too quickly.

Mountains allow me to grow and learn things about myself. It is tremendously satisfying to experience getting stronger. You start with small strolls. They get longer and longer until you can comfortably hike the entire day without thinking much of it. There is a natural progression ramping up the challenge and climbing harder and harder routes. Learning the exact boundaries of ones physical limits and extending them gradually.

I know things about myself that few people know about themselves. I know, don't imagine, how I react to extreme stress and fear of death. I know that I won't panic and that I'm still able to control myself and act rationally and constructively. Knowing where my limits are gives me a lot of power and self assurance in everyday situations. The coping mechanisms and mental strength you train while climbing are applicable elsewhere.

Mountains offer a well defined, unambiguous goal. Either you reach the summit or you don't. This is in refreshing contrast to the ambiguities of our modern world. Everyday life rarely offers such easy black and white goals, everything is muddy. At the same time the pay-off is proportional to your investment. The more effort you expend to reach a goal the more gratifying the reward. Give it all your strength, invest all your smarts. The ultimate stake is your life. You can't put more on the line than that.

Taking risks is freedom. I choose to do it. The default choice in our modern world is to always minimize risk. But if you accept that, then there is no choice. You are not free. You don't decide. It gets exciting once you make a conscious decision: how much risk am I willing to accept for the experience? I crave the independence of it. Nobody tells me what to do, I am the master of my own decisions and carry the weight of the responsibility.

I love the feeling of being self-sufficient, autonomous. I carry everything I need in my pack. If I want, I can stop right here and bivvy for the night. This was an eye opening experience on my bicycle trip of half a year. You don't need much, you can be self reliant and make a comfortable home almost anywhere. That is a powerful realization.

Climbing requires full concentration. A day of scrambling on a ridge is not only physically exhausting but also mentally draining. Staying one hundred percent focused in the here and now requires discipline. You cannot afford a single lapse. As a consequence you train an iron will. It has meditative qualities. You don't think of anything else. Nothing clears my mind of work like climbing a steep wall.

I like the exclusiveness of it. I like being part of an elite club. The group of people that don't need to read an article like this, because they know. Because they are driven by the same passion. Who don't shake their heads at the "crazy risks", because they are already scheming their own next epic.

I'm a fiercely competitive person if I set my mind to it. I enjoy being stronger and faster than the average guy, the next guy. I enjoy the tension this creates. Mountaineering is an intensely shared experience. You need a rope partner. You literally trust your partner with your life and rely on her for your safety. So climbing is also fundamentally cooperative. You reach the summit together or not at all. An epic struggle shared with a partner creates a strong bond.

Whenever I'm in wild places I imagine what it must have been like to be the first human to stand there. To lay eyes on this landscape. To explore and discover. Being in the mountains where few people can go allows me to live this fantasy at least a little bit. White spots on our maps have become very rare, so I go looking for them in the vertical world. True adventures have almost become extinct.

At the same time I like ticking off places on my own, personal map. It is tremendously satisfying to see a map of the entire country and have it covered in a spiderweb of GPS tracks of my own hikes. Experience that you can explore so much solely under your own power. I can walk across an entire country in a few days - imagine that!

Mountaineering has beautiful toys. I like playing with the highly specialized, high quality gear. Maybe childish, but then again, most hobbies have equipment fetishists. Becoming proficient in, and eventually mastering, its use is a reward all by itself. Like a carpenter with her tools.

Climbing, and even hiking, is intellectually satisfying. You marvel at a particularly beautiful line, elegant route or an efficient move. Something as seemingly trivial as a foot placement on a downhill run can be quite stimulating. Split second decisions. Either you need to waste a lot of energy or you smoothly float over the rocks.

You are playing in magnificent surroundings. I'm an atheist, but if I ever thought there was something more to our existence it has been in the majestic arena of nature. I've had some very emotional moments in the mountains. Taking it all in has caused me to well up and my voice to break. I have shouted of joy against the wind when there was no one to hear. I have run pirouettes down a glacier to celebrate existence. Life is great!

Mountaineering then is fun because it satisfies deeply human urges as few other endeavors can.

Probably my favorite mountaineering photo: Vince Anderson, photographed by Steve House, after climbing the 4100 meter Rupal Face to the summit of the 8126 meter Nanga Parbat, a superhuman effort.

Civilization V is a huge disappointment

I'm a great fan of the Civilization series of games. In fact, after the original UFO Enemy Unknown, I think Civilization IV is the best game of all time. I have played the entire series extensively, including the latest installment, Civilization V. Although it breaks with the mechanics in very significant ways I started out with an open mind and welcomed the changes. Using a hex grid instead of a rectangular one; a single unit per tile instead of stacks of units and the introduction of independent city states are just some of the changes I really liked on paper. However, the game turns out to be deeply and fundamentally flawed.

Civilization games traditionally are complex and long games with an insane amount of strategic depth. The latest installment streamlined the interface and the complexities to the point where very little is left. While the original games were very carefully designed to offer meaningful choices every step of the way the latest game runs mostly on auto pilot after the initial set of decisions have been made. This is all the more frustrating as the bulk of the game is still to be played after that point and just churns on for hours. Especially in multiplayer sessions this has the effect of reducing some or all players to "end turn" clicking robots.

To test this theory I've played a single player game on "King" difficulty. This is the first difficulty level where the AI gets an unfair advantage against the player and used to be quite challenging if you didn't cheat (by reloading on bad outcomes or setting up the map to your advantage). I built just three cities, pre-determined the entire technology research tree, automated all of my workers and the end of turn and let go. Every couple of minutes I had to decide on what building or unit to produce next or answer a diplomacy request by the AI (defaulting to the least offensive option every time). I won the game by a huge margin.

This is ridiculous. My decisions/hour ratio hovered around 6, or one every ten minutes. That's a huge fucking waste of time and also deadly boring.

Some thoughts on the major design mistakes that broke the game for me:

• A lot of the game mechanics have been changed from qualitative choices to quantitative ones. This leads to a complete and utter lock in and a 100% predictable game. Once you've decided your social policies you can never change them again. Where you used to have a revolution and change to a police state in order to go to war you now have to stick to your Pacifist ideals forever. There are no sliders for controlling how you want to spend your tax money. Where you used to be able to change from a research focused nation into using all your tax dollars for the war effort you are now stuck.
• Worst consequence of the first point: no technology trading! This is huge. You can only sign "research agreements", giving both parties a boost to their research. However, there's no way you can ever catch up once you've fallen behind or back stab one of your friends in a multiplayer game by giving his enemy the secrets of gun powder.
• It doesn't really matter whether you're at war or peace. Your citizens don't care at all. Long time civ players know that managing war-weariness used to be a big issue. No more. Also the AI is so stupid as to not pose a threat at all. The new, more tactical combat system, with a single unit per tile, introduces things like choke points, ranged attacks and unit formations. None of which the AI understands. And to add insult to injury cities have the new ability of defending themselves. So no more surprise attacks. You can only start building defenses once the enemy has already crossed your border and don't need to worry about it.
• Despite this brain dead AI it still takes ages to process every turn. Even on a recent, high powered gaming rig the AI on a moderately sized map will take longer for its turns than I do for mine. This is ridiculous.
• City states, while a novel and interesting idea, offer an easy way to victory if you want to end a game early. It's so easy it almost feels like cheating. Buy their loyalty, have them vote for you, done. Speaking of which: money is grossly overpowered. This is the only mechanism by which you can attempt sudden strategic pivots, but it's also the most boring one. Need units? Buy them. Need to expand your borders? Buy some culture. Need a wall to fend of an attacker (as if)? Buy it.

I could go on and on, but I think I'm already boring you. You get the idea. The game is a huge tragedy - it had so much potential and some really innovative ideas but tanked horribly.

Contrast this to love letters to Civ IV like this beautiful piece: Civilization: The Good Kind of Addicted. Now Jon Schafer (the designer of Civ V) seems to be a smart guy and understands most of the shortcomings of Civ V. His new project, At the Gates, looks quite promising. And in any case, it seems that few people are still making the kind of deep, complex, turn-based games I love, so I guess I should cheer on and encourage any attempt.

The Mindset of an Engineer

Google I/O 2013 is over and a lot has been written about it. What struck me is how negative and cynical some of the reactions to Larry Page's surprise appearance and Q&A were. This article in Slate is a prime example. Now I understand perfectly well how people are distrusting and jaded with marketing speech from big companies. But that's sort of the point. The live Q&A format was set up in a way that Larry could be his geeky self and way more candid than the typical polished prepared speeches you get at events like this. Gizmodo and Techcrunch got it I think.

Anyway, an older article in Wired nailed his character perfectly, which was evident in the keynote speech as well:

"Teller imagines wheeling a Dr. Who time machine into Page’s office. He plugs it in and—it works! But instead of being bowled over, Page asks why it needs a plug. Wouldn’t it be better if it didn’t use power at all? “It’s not because he’s not excited about time machines or he’s ungrateful that we built it,” Teller says. “It’s just core to who he is. There’s always more to do, and his focus is on where the next 10X will come from.”"

This! A 1000x times this. My younger brother once remarked on the difference between the two of us. When confronted with something new his first impulse is to understand why it is the way it is, the history and background of the thing. My first reaction is to find the flaws, bulldoze the place and come up with an improved design. So in this sense I can very much sympathize with Larry's mindset. As I suspect most engineers do.

"We should be building great things that don't exist!"

An argument in favor of just winging it

Last weekend's planning fiasco made me ponder my attitude to planning and life in general. There seem to be two very different personality types, the planner and the improviser. I'm very much the latter.

I don't want my life to feel like ticking off a grocery shopping list. Executing a never ending todo list, comparing each new experience with my expectations for it. Better to be surprised every day and enjoy it. I saw a terribly depressing drawing once: rows of little boxes, 12 to a row, 80 rows. Those represent the months in your life. Cross off the first bunch according to your age. A very sobering countdown... Better stick to schedule, get your diploma, get married quick and get it over with! Hurry!

Instead I try to preserve the optimism of a child that each day will never end, I'm invincible and everything is still possible. Change the world as you once wanted to.

I don't want to know the weather forecast - I want to go out and have fun regardless.

I love imagining being an explorer, being the first to discover the area I'm in. How it must have been like without any traces of humanity, no footsteps to follow. That experience gets completely destroyed by overloading the travel guide with post-its and slavishly following a tight schedule.

That dreaded interview question: Where do you see yourself in 5 years? I don't. I'm enjoying myself right now, why should I envision myself to be somewhere/somebody else? And no, this does not imply I'm without ambitions ;-)

Retirement planning, better do what you want to do now. If you see an opening, get a chance, seize it. Don't wait for something better or more safe to come along, it might never happen.

I find plans to be incredibly restraining and burdensome.

TL;DR: Live today - you might not be around tomorrow.

Raindrops falling on my head

It has been raining last night, flooding the cozy little creek which is my way to work. It's usually of a size where mothers let their little kids play in the water without having to worry about them. That changed a bit ;-)

Also, don't tell anyone, but these are top secret plans from one of the whiteboards at work:

What's up?

Didn't post any hikes for a while but I have been busy:

I spent a day working at the school for guide dogs for the blind Allschwil. This was part of the Google Serve 2011 engagement where Google will donate a workday and all associated costs for volunteering at charitable community projects. We spent the day shoveling around mulch building a new training path for the dogs. The school and it's pupils were quite impressive. I have never seen such considerate, well trained and well behaved animals. Teaching dogs to understand human constraints and proportions and that a railing the dog can simply walk below presents an obstacle to a human is no small feat. It was impressive in another way too: Educating a single dog costs north of CHF65.000,-. The school is well equipped and staffed (about one trainer for three dogs). Dogs get individualized meals, medicine, training regimes and, later, foster families before they get assigned to their final client (not owner, the dogs remain the property of the school). In short: they get far better education and training than we give our kids. All in a not-for-profit setting. I wonder...

We went to the Google Summer Picnic and competed in a beach volleyball tournament. Didn't make it to the final rounds but had a ton of fun.

Tim and I completed a climbing course in the Kletterzentrum Gaswerk Schlieren - apparently one of the world's largest indoor climbing centers. After three weeks we are now qualified for lead climbing (as opposed to toprope). Getting closer to my goal of doing some serious outdoor climbing ;-)

Anita visited and we worked on the apartment a bit and went swimming in one of the many public swimming spots on Lake Zürich. 30°C air temperature, 20°C water. We both got a bit of a sunburntan. We also had a film night in the company movie room and went for a sightseeing boat trip on Lake Zürich. Cruising along the "gold coast" which my real estate agent introduced to me with the words: "This is where the rich people live - the other side of the lake is for the millionaires".

chill out and chair massage lounge

movie room

I started participating in the Google SPC (Self Powered Commute) program and am cycling about 16km and 100hm every day now. Aside from exercise and fun I get some money for a charity of my choice.

I started a 20% project. I also ran my first job on several thousand cores, completing a calculation that used to take days in less than 10 minutes. Sweet ;-)

My new apartment

No hike this weekend. Anita was here and we went shopping for furniture (we spent a metric shit ton of money and didn't get a tenth of what we'll need). I have finally found an apartment! After applying for a very modest one and being rejected I went to see about two dozen more. Zürich's real estate market is completely nuts. Landlords can demand pretty much whatever they want and still get hundreds of applicants per offer. You hand over basically all of your personal and financial information and they'll actually follow up and call your current and former employer, your current and former landlord and your bank. In any case, after quite a long and exhausting search I finally surrendered and went for one that was well above my initial search criteria. This has the disadvantage that I'm now paying about 5 times of what I did in Münster (Zürich is the world's 7th most expensive city). But it has the non-trivial advantage that it'll easily accomodate Anita and me so we won't have to move again once she moves over here. And of course, it's very nice ;-)

Besides looking for furniture we moved my gear from the temporary apartment to the new home. First trip Anita and me, packed to the limit. For the second trip I went alone, carrying my huge backpack with an additional sack strapped onto it and my desktop computer in my arms. That made for about 65kg of stuff carried uphill over a distance of 1.6km. Don't do that. But I found out that I can actually fit a 19" display into my backpack ;-)