This is a cool game that has it’s players guess whcih google query generated the the images shown in a montage. My immediate thought was that it was talent scraping to collect data for a new kind of captcha. The author says nope. The esp game guys have data that could serve the same function.
Traffic on my sites is highly skewed. Hour by hour I don’t get much traffic; but just occationally a flash-crowd shows up at my door. The distribution would be power-law; except I truncate them. By the way; “truncate the flash-crowd” is a nice way to say the machine crashes. Don’t say “Damn the machine crashed!” instead you say “Oh boy! A flash crowd!”
I’d get around 50% more traffic if I could deal with the flash-crowds successfully. But I’m a cheap bastard and I’m not willing to buy a bigger pipe, a bigger web server, and spend the time to stress test my site regularly to assure it can deal with these very rare very huge bursts of traffic.
Last time the flash crowd showed up I rebooted the poor machine and used the Coral Distribution Network to solve the problem. Coral’s cool. It’s research project built on the planet net distributed computing research platform. So they let people use it for free. Which is very nice of them; though it makes me feel like a freeloader.
That all leads to two desires (or ideas).
First off I wish I had a means to configure my web server so it would automatically shed load to Coral when the flash crowd show up. I’d prefer not freeload during the 95% of the time when my site is doing it’s usual thing out on the heavy tail of the power-law distribution. This might be easy with one of the Apache modules that does load monitoring and shaping; but I haven’t found the right mix of stuff to make it easy for me to set up.
Secondly, it would be way cool to architect a means that sites with similar load distributions could form cooperative peering relationships and then when one of them get’s struck by a flash crowd the other members of the consortium would step up to deal. This one would make a sweet topic for somebodies doctoral thesis (particularly the design of the goverance and it’s enforcement). Or, the other kind of somebody could ‘just’ would make a weekend hack our of these two ideas.
Ok, I admit it, I’m trying to bait the lazy web here. I forecast sunny weather!
The other day I said to somebody that “it’s a political process” and he replied “you mean he has a hidden agenda.”
I think the word politics names the process by which many parties with widely divergent world views attempt to find common ground so they can get something done. That search is the key. So when I say “political process” I mean that the parties are still searching; but that they are still in the same room trying.
Lots of people hate that room for quite a laundry list of reasons. First off it appears that nothing gets done in there. Attempting to find common cause is what gets done in there. It’s risky hard work. There is huge risk that you spend resources and fail to find common cause. There is risk that spend resources, find what you think is common cause, and then latter you discover it wasn’t strong enough to stand the buffetting of future work. Always there is the risk that the guy in the room won’t be able to deliver his constituency. All that raise the risk that the game your playing is a short term rather than a long term one; which only makes the possiblity of defections higher.
If people all see the world in the same way then you wouldn’t have a problem you wouldn’t need to go into a room and try to find some common cause. In such rooms the puzzle is to figure out what the other guy can see that you can’t. In retrospect I find it amusing to realise that the difficulty of that work makes it always appear that the other guy is hiding something. It is rare that people know what they know. It’s even more rare that they know what you need to know.
If you rub two surfaces against each other they tend to smooth each other out. But, if you rub them every which way for long enough they don’t get flat. To get a flat surface you need to rub three surfaces against each other in assorted combinations. You can make optically flat surfaces this way. My father was an optics guy. He taught me that, but I forgot until recently.
Since being reminded about that I’ve been thinking about how interfaces boundaries rub against each other and how they tend to smooth out over time.
In industrial standards work we spend a lot of time proactively creating specifications whose intent is to assure smooth efficient exchange on interfaces. It’s not uncommon to fly brilliant engineers and implementors to other continents so they can do interopt testing to assure that the resulting systems are conformant to the spec and work smoothly with each other. The cost of such coordinated efforts is extremely high; often fatally high.
Rubbing isn’t a very sophisticated approach to gettting a smooth surface. It is what I was taught to call a strong method, a method that works in all cases. Rubbing, in optics, is always the last method. When you make a lens you attempt to get it right; but latter you always use rubbing to get it right.
I distill two points out of all that.
Simple standards tend to win because they have fewer rough edges that need to be worn down when you get to the rubbing stage.
Rubbing to get things smooth and interoperable is always part of the story.
Standards that are many to many smooth out better. I.e. one of the challenges in B2B standards compared to other internet standards is the way that it’s less common to find a firm that is doing B2B exchange with a huge number of partners in a way that’s similar to the huge number of sites that web browser visits.
Sam add some case studies to coordination design problem space.
I’m 
having a lot of fun over at my 2nd blog: Gibbon’s Garage.
I didn’t eat it. I would have eatten it. The Internet saved me. This posting saved me. I’m quite grateful.
…it’s so perfect, it makes you want to believe in god. …
I’ve started getting a lot of that I think of as Jesus spam. If I train my Bayesian filters to recognize this as spam will Jesus’ email get thru? Could it be that one of the other Gods is sending this mail. Is taking a God’s names in vain a form of identity theft?
How to increase your tip (I bet these work for increasing your bonus too).
| 17% | Wearing a Flower in Hair |
| 53% | Introducing Self by Name |
| 20% | Waiter Squatting Down Next to Table |
| 25% | Waitress Squatting Down Next to Table |
| 100% | Repeat Order Back to Customer |
| 140% | Smiling |
| 23% | Suggestive Selling (aka upselling) |
| 42% | Touching Customer, Study 1 |
| 27% | Touching Customer, Study 2 |
| 22% | Touching Customer, Study 3 |
| 28% | Touching Customer, Study 4 |
| 40% | Tell a Joke (to entertain customer) |
| 18% | Give a Puzzle (to entertain customer) |
| 18% | Forecast Good Weather |
| 13% | Writing “Thank You” on Check |
| – | Waiter drawing smiley face on Check |
| 18% | Waitress drawing smiley face on Check |
| 37% | Bartender drawing sun on Check |
| 25% | Restaurant, Using Tip Trays w/ Credit Card Insignia |
| 22% | Cafe, Using Tip Trays w/ Credit Card Insignia |
| 18% | Give Customer Candy, Study 1 |
| 21% | Give Customer Candy, Study 1 |
| 10% | Call Customer by Name |
Providing great service is not on this list because studies show measures of service quality as reported by customers is not particularly coorolated with the size of the tip left. Don’t be fooled, tips do not create a feedback loop that improves service.
Tips are an odd epilog to a transaction. The buyer gives a gift to to the seller, or is it the seller’s agent he gives a gift to? There are experts out there on every aspect of commercial transactions. The expert on tipping is Michael Lynn a Cornell. That table is gleaned from his pamphlet MegaTips. I read that 21 Billion dollars of tips are given every year in the US. So a huge proportion of the income at the low end of the income ladder are these gifts.
Like all gift scenarios it’s hard to be sure who’s getting what from the transaction. But it appears that the buyers are buying something with their tips, i.e. appear to be buying a relationship with the server. It maybe they are trying to weaken this agent’s loyalty to his employer. What is clear is that if the buyer is convinced that the server likes them then the buyer will tip well.
I doubt you will be surprised to learn there are efforts to change the “standards” about tipping. It’s a great example of how many players get involved when ever you try to shaping an exchange standard. Restaurant owners and service personel would prefer that the “standard tip” be higher. I was taught as a child that the standard tip is 15%, but the industry is working to let it be known that the standard tip is 15% to 20%. Some dead beat segments of the population known to be lousy tippers. The industry appears to be working on that.
While owners and servers both want higher tips their solidarity around this part of the transaction falls apart moments later. Owners would like minimum wage laws to include tips, servers of course don’t. Recently Rick Santorium tried to stick a clause into the bankruptcy bill to force that change upon states, shortly after getting a big tip from a resturant chain.
I liked this paper that tries to draw connections between attributes of national character and variablity in tipping across nations. The conclusions strike me as tenuous, but they certainly are fun. Anxiety, status seeking, and masculine personality in your nation increases the level of tipping.
Personally I find tipping, and certain kinds of bonuses, to be very corrosive to professionalism.
I’m not aware of any software exchange protocols that include a tip phase. DNS has a way that the server can throw in some free extra answers to questions that he suspects that the client will desire. There must be some senarios where it would be useful for the client to toss in some extra gift as he’s closing the connection.
