| I reply to the Internet |
[Aug. 19th, 2010|06:19 pm] |
Some guy named John Cook begged the question! On the Internet!
I, of course, did my civic duty. I commented.
You are begging the question by defining “minimalism” as necessarily thoughtless. So, of course, minimalists are!
As you mentioned in the comments, “minimal” is a superlative meaning (roughly) “no less of X is possible.” You appear to define ‘X’ as “things necessary for life without conforming to required social norms.” But your more-minimalist-than-thou minimalists could just as easily have a different definition, e.g., their statement could be “Buy my book. I have only 39 things, while still maintaining appropriate social relationships with my friends!”
Indeed, it is good you did not link to the man in question because context because or facts could undermine your point; your hypothetical man, by definition, does give you things you want in exchange for the hypothetical things he needs from you. With context, he could have been a brilliant mathematician who trades knowledge, co-authorship, and bragging rights for a warm place to sleep. I'd take that over 4 eggs and a cup of flour any day.
Then again, maybe he’s just douché. I guess it depends on the minimalist instead of the minimalism.
|
|
|
| Economic quiz for the day |
[Jul. 6th, 2010|10:55 am] |
Suppose you own and operate a Jimmy Johns franchise. You normally have two counters producing sandwiches, but since the 2008 recession you've been getting fewer customers, so you laid off half your staff and only operate one counter. The other one sits idle and unused.
Given that you already meet the reduced demand for lunches, what will motivate you to rehire your staff and open the second counter?
(a) A sack full of cash
(b) A loan from the bank
(c) Less government spending
(d) More customers
Bonus question: The sandwich counter manufacturer wants to sell you another counter, which of the above will cause you to buy a THIRD counter?
This is the mental exercise you should do whenever you hear a politician talk about tax breaks vs. stimulus spending. Tax breaks are (a), stimulus spending is (d). |
|
|
| Why Amazon's spot pricing is at least not good (and maybe bad) for you |
[Dec. 15th, 2009|12:47 am] |
Yesterday, Amazon released Spot Prices for their EC2 instances. Really, it is quite ingenious; like Mechanical Turk, they've taken a task they normally would have to spend costly internal resources on and outsourced it to thousands of people that now externally do it for far less. In this case, finding the profit maximizing price for EC2 resources is now done by EC2 customers. While MT pays typically a pittance for each task done, the EC2 consumers are in fact paying for the privilege of figuring out exactly how to be charged the most for their resources.
Don't get me wrong, providing a spot price and allowing users to make contingent purchases is a brilliant business tactic. It certainly boosts AWS revenues, but more importantly it provides them immensely valuable data. Data you can't buy; data that would make most economists drool uncontrollably. Amazon collects this incredibly valuable data for free, their users clamor to share it, to give it to Amazon to lock up in its databases and mine to their hearts content. They get the demand curve.
The most important element to realize is that Amazon's spot pricing system is not a functional market. There is only one supplier of goods, so none of the standard economic theory applies. Well, at least none of the theory about competitive markets.
The graph above is the classic depiction of a monopoly market (from wikipedia). The Y-axis is Unit price of an EC2 instance, the X-axis the number of instances supplied/demanded. The line labeled MC depicts Amazon's unit cost for supplying another EC2 instance (the marginal cost). The line slopes upwards because when Amazon is not running at full capacity, they will start up their least expensive servers first (e.g. the most power efficient, newer ones), and only run their more expensive servers if they have to. The line labeled demand depicts the number of instances people are willing to buy for a given price. It slopes downward because with each reduction in price, additional folks who were priced out of the market come in and will buy more instances.
In a perfectly competitive market, the price and supply of goods stabilize where the MC and demand curves intersect. At that point, it is more expensive to turn on the next instance than the next buyer is willing to pay, so no more instances are powered up. Because Amazon is the sole supplier of instances, however, they can withhold them from the market; intentionally leave machines powered down to drive up the price. Since everyone pays the same price, instead of supplying Qc instances, they only run Qm instances. That increases the price to Pm. This will increase their total profits (the yellow shaded area). In all cases, the monopoly price will be greater than the natural market price; it never works to the consumer's advantage.
Spot pricing is Amazon's way of getting their users to put this graph together. Spot price bids collectively produce the demand curve. By supplying the data, users help Amazon increase their prices to the optimum limit. And only Amazon gets to do this calculus, their users have tipped their hands regarding prices they'll pay, but Amazon has shared nothing about prices they'll accept. Goods in competitive markets are priced proportional to the cost of producing them. Goods sold by a monopoly power are priced at a higher level, proportional to the value they provide to the customer.
Furthermore, by being setting the spot price Amazon can engage in lucrative price discrimination. Right now, any user can get a small EC2 instance for $0.085/hr. This is a terribly crude way to sell instances. If I'm a (hypothetical) Tech Crunched startup desperate for more capacity on opening day, I might be willing to pay $1/hr while if I'm still eating ramen in my garage (somewhat not hypothetical), I'm only willing to pay the published $0.085/hr. EC2 pricing is uniform so Amazon can't charge the TC rockstars a different rate than they charge me. Our counterparts in other industries have figured out some workarounds.
Sometimes I think Zoolander might be something of a documentary, but actually fashion folks know how to sell, they know how to sell well, and we can learn from them. They figured out how to bucketize their customers charging them each a different price and in doing so increase their profits. They get a shipment of hot trendy clothes and put them out at a steep markup. Wealthy, price insensitive customers rush to the store to buy the latest outfits and pay exorbitant prices. Then once all of the wealthy folks have filtered through and sales start to drop off, they lower the price, advertise a holiday blowout, and a new wave of slightly more price sensitive folks come to gobble up inventory. Then, anything that's left is marked "clearance" and sold at a meager profit to those deal-seekers that patronize The Rack. Each group pays price tailored to their disposable income rather than the uniform and lower price predicted by a perfectly competitive marketplace. The tradeoff is that the lower on the price scale the consumer, the less reliable their supply of goods. Now Amazon naturally wants to do the same thing with their EC2 instances. They want to sell to all their price-insensitive customers at high markup, published prices, while still having something to offer at a lower price to all the cheapskates.
In an actual, functional market this would not be possible. The spot bids by consumers and spot asks by suppliers would be public knowledge. Everyone could see that Amazon is willing to sell an effectively unlimited supply of small instances at $0.085/hr, so as the sophisticated buyer, I realize I can also set my bid to $0.085. Then, in the worst case, I'll only pay the full retail price, and most of the time I'll pay the lower market rate. This, of course, will never work with Amazon's spot price system.
Amazon sees all the bids and is the only "ask". They know exactly how many folks are willing to pay the retail price for an instance and once that number is high enough, they will withhold all inventory from the spot market. They'll have the servers; they'll just turn them off for an hour. It will ensure that the folks for whom an instance is truly worth that $0.085 experience an avoidable, unexpected outage that incents them cough up the full price 24x7 and not get identical service at the cheaper price offered to the low end customers. Because everyone communicates all their bids to Amazon, they know exactly when to pull the trigger to maximize their own profits. Meanwhile, the bargain basement folks will already be taking frequent outages as the price fluctuates, so they'll be prepared for an hour of downtime anyways. Thus, the EC2 community is segmented into two groups and Amazon gets to charge each of them at the limit of what they're willing to pay.
The important thing to remember (and this is where most journalists get it wrong) is that spot pricing does not constitute a (functional) market. Amazon has simply launched a tool to help them extract more profits from you. Compared to only paying full on-demand prices, yes, this is better as computing resources will sometimes be available for less. Compared to an actual competitive market for compute resources, EC2 spot pricing is in fact terrible. You are at a significant information disadvantage, availability is necessarily unreliable, and you subject to higher monopoly pricing. Your participation only helps to stack the deck against you.
As a business, Amazon is being incredibly innovative. They're figuring out exactly how to squeeze every last dollar of profit out of their platform. As the consumer, that means they're squeezing every last dollar out of you. That's good for Amazon, but probably not good for you. Eventually, there will be many VM providers and there will be a real spot market that has all the positive characteristics we'd expect. This is a step in the right direction, but just for Amazon. |
|
|
| Gardening, Week 10 |
[Jul. 4th, 2009|11:58 am] |
|
All things considered, the garden has been pretty successful. (Almost) everything is still alive and growing pretty well. The casualties so far have only been two broccoli plants, one of which I'm pretty sure was uprooted by Amélie (the other died for reasons unkown). It looked like the squash wasn't going to make it, but an increase in its waterings nursed it back to health.
Speaking of which, when you read about people over-watering online, do not believe them. It is a trap. As near as I can tell, the Internet and I have very different opinions on what constitutes "wet" soil, so watering only when the soil is dry does not work. My plants all get a healthy amount of direct sunlight and through trial and error (mostly error), I've found they need a generous amount of water every day.
Squash
 Squash, it turns out, cannot be over watered. This one's soil always felt quite damp, so I only watered it every few days. It grew fine for the first few weeks but once it started to bear fruit the leaves started yellowing and dying. The Internet gave conflicting advice on the watering of squash, with many sites suggesting it should only be watered every two or three days. False. This guy needs plenty of water every day. After I started giving it ~.2 gal/day, its leaves greened up and it is now growing significantly faster than ever before. About 50% of the leaves in the photo have developed in the last two weeks. Squash are supposed to sprawl out to ~10ft, but given the late start and that this one is potted I'm sure it will be much smaller.
Tomatoes
 Early on the tomatoes exhibited "leaf curl," which apparently is a common symptom of over-watering. Maybe. Apparently many varieties of garden tomatoes just have a tendency to curl, and it doesn't materially affect the plants. But, at first I backed off the watering since the leaves were curling, which was again a bad idea. The plants themselves haven't grown very fast (relative to the Sungold plant on the next block), and many of the fruits are growing very, very slowly. I've upped their watering and they seem to be doing well, though I won't know for sure if that has helped until later in the season.
Despite having far fewer fruits, the Green Zebra (pictured) is on track to produce first. The Sungold, however, has 3-4x the flowers (50+), so it will probably produce the larger yield. I also didn't get tomato cages until just a few weeks ago. Those are much easier to put in the pot when the plant is small; whoops!
Sunflowers
The sunflowers are getting huge. The biggest are around 4ft tall at this point and they grow ~1"/day. They've been the hardest to mess up. They seem nearly impossible to kill (as long as I give them some water), only growing faster or slower based on the amount of care they get.
I planted them in 3 groups of ~6. A few in each group failed to take root after sprouting so I had to sow new seeds in their place. These new plants were growing a few weeks behind its siblings and thus generally crowded for light. Interestingly, they've raced taller, faster, at the expense of a thiner and weaker stem. A few of them needed steaks to stay upright, but as they've got more sun they've grown stronger.
Broccoli, Spinach, and Lettuce
 The side garden has worked out okay. The spinach bolted really early (~week 6) and was eaten; it really should be in a place that doesn't get too much direct sun (or perhaps pulled and re-sown frequently?). The lettuce is ready to be eaten, just as soon as my housemates or I actually make a salad at home. It turns out they grow really big and need to be quite far apart, I'd say 24" or more. The broccoli seemed to flower early and produce small heads, but Mike said they were good. They're very delicate plants, I'm not sure I'd grow them again without an, e.g. flower bed where they would be more protected.
Corn
 The corn is definitely the most anticipated crop of the yard. I read online that they need plenty of water, at least 1"/week. False. They need more. I carefully gave them 8gal/day (which works out to 1"/week) and they remained healthy but many were growing very slowly. The ones in sunken portions of the garden (where water tended to pool) grew noticeably faster than the others. I've at least doubled their water (now using the hose, so not sure of the exact amount), and they are all growing much faster. The bigger ones are now growing between 0.5-1"/day. I think the corn will come in later than the early Aug I was anticipating.
A ridiculous number of poppies grew naturally with the corn (the orange flowers). I've left them in because they look nice, but periodically thinning them to keep them from overcrowding the corn. Over three thinning sessions, I've probably pulled 100lbs of poppies. They grow like weeds, but at least they look nice. The bees really love them. I water before I leave for work and will routinely see ~50 bees hopping from flower to flower.
|
|
|
| May need a 4th roommate |
[Jun. 5th, 2009|10:04 pm] |
|
Looks like one of our roommates is moving in with her b/f at the end of July. Barring us moving out, we'll be looking for a new (preferably female) 4th roommate. Give me a shout if you or someone you know would be interested in being a roomie in our house. |
|
|
| DNS poisoning FAIL |
[May. 18th, 2009|11:05 am] |
Today when I noticed some web sites were rendering as blank pages or other odd behaviors. When slashdot.org suddenly exhibited the same symptoms, I started to look into it.
( Code follows here... ) |
|
|
| Adventures in Gardening |
[May. 3rd, 2009|12:07 pm] |
My housemates and I started a garden in the front yard last week. We tore up sod along this stretch of the front sidewalk and planted corn and sunflowers. It very clearly used to be a garden before as we found probably 30+ onions growing the in grass we pulled up.
Our neighbor has sunflower plants in their yard, so I'm fairly confident they'll sprout, but I'm a little concerned about the corn. Apparently corn doesn't like to germinate in soil temperatures below 65℉ so we may just end up with a fallow patch of dirt. :( So we'll need a backup plan in case this corn patch looks the same next weekend.
This Saturday I dug up a section around the tree in the front yard to plant more edibles. It too wasn't always unweeded grass, as evidenced by another 20 or so onions (with much bigger bulbs!) I managed to pull out of the grass clods. That probably gave me another 30 sqft or so to fill with plants; fortunately Erin was working at City People's yesterday so I could go bother for recommendations. :) We picked out a few starter vegetables and now I've planted all the scavenged onions (we'll see if they survive, I think I tore up their roots), mixed salad greens, spinach, and broccoli. I also picked up two tomato vines and a squash plant for planting elsewhere.
I'm a little concerned about the Broccoli though. It apparently needs full sunlight and as you can see from the photo, the neighbor's tree is still shading the whole area by noon. I may have to move them to a more hospitable location. If the corn hasn't shown by next Saturday, I think I'll move them into the patch along the sidewalk.
 |
|
|
| navigation |
| [ |
viewing |
| |
most recent entries |
] |
| [ |
go |
| |
earlier |
] |
| |
|
|
![[info]](http://l-stat.livejournal.com/img/userinfo.gif?v=91.6){kind=small}
loganb's journal
