Blog Archives

Topic Archives

Holger Danske

Holger Danske

Contact
This form does not yet contain any fields.
    Tuesday
    Apr062010

    Four Peaks Arizona Peach Beer Review

    Four Peaks Arizona Peach

    Four Peaks Arizona Peach

    Fruit Ale 4.0% ABV
    Type 15 Beeriodic Table

    Like all good fruit beers, the bouquet of peach is stronger than the flavor. Not as sweet and heavy as a lambic, but rather light and refreshing. The Magistra had this beer on tap at the Four Peaks Brewery in Tempe. She drank it with a portabella sandwich on the patio. Clear, with a short head.

    Rating

    The rating is a bit difficult on the six-pack scale. This beer is really good, but you want to drink lots and lots of it, especially on a warm spring afternoon. Thus, I declare it a two.

    My other beer reviews

    Saturday
    Apr032010

    Technical Career Advice Section

    I have been working a Technical Career Advice section for a couple of months now. The initial impetus for this section was my participation in the Career Advice section of the PhysicsForums. The discussion in that section was strongly oriented towards scientific and academic jobs, and I felt that someone needed to represent the industrial and engineering side of things. In the past nine months, I have learned a great deal about the technical job market that complements my work in recruitment and career mentoring at my company.

    Anyone who would like to learn a great deal about the technical job market should check out the Career Advice section. However, one would do well to remember that free advice is worth what you pay for it. One needs to keep one's own judgment about one's own career, and the information given on that forum is not guaranteed to be right. With that caveat, I can recommend it strongly.

    I have been collecting information about careers and job performance, and I will be placing this information in my own career advice section. Some things will be general, but most of my advice will focus on careers in Science, Technology, Engineering, and Mathematics [STEM], since that is where my passion and my experience are.

    I enabled the first page today, Places to Look for Jobs, because I have enjoyed the two sites I placed there so much. The best two job sites I know are Indeed and Glassdoor. Both have provided me with good information, so I can recommend them. Glassdoor requires you to provide information in order to see more than a preview, but the quality of the information is high. It is worth remarking that the company reviews are provided by people with a reason to review a company, which often means disgruntled employees, but if you keep that in mind and look for the general gist of the reviews, you can get an idea of what a company is like. The salary information is also quite good on Glassdoor, and it gives you the ability to compare salaries across companies and locations. It is all self-reported, but it seems to match up with other salary information I can find, so I think it is basically reliable. Indeed is a job site aggregator, and it gives you access to more jobs than any other site I have found, with a robust, Google-based search functionality.

    Wednesday
    Mar312010

    CrossFit 2010-03-31

    As Many Rounds As Possible
    20 minutes
    • 70 m lunge lap
    • 10 burpees
    • 10 pullups
    Monday
    Mar292010

    CrossFit 2010-03-29

    Miracle Mile Mix

    • 1/4 mile run, 3/4 mile bike, 500m row [row]
    • 20 kipping pullups
    • 1/4 mile run, 3/4 mile bike, 500m row [row]
    • 25 kettlebell swings [20 kg]
    • 1/4 mile run, 3/4 mile bike, 500m row [row]
    • 15 burpees
    • 1/4 mile run, 3/4 mile bike, 500m row [row]

    Time 17:21

    Sunday
    Mar282010

    The Reversal of the Great Risk Shift

    John Reilly has an excellent post up about the recent federal health care bill.

    The enactment of the new federal health-insurance system this week marks the beginning of the reversal of the Great Risk Shift that has characterized American society since the late 1970s. There is now reason to hope that American society can return to something like its historical condition, in which it was possible to change jobs and geographical location without taking your life into your hands. Wages may again be able to rise, since all new money that becomes available for personnel costs will not automatically be eaten by insurance-premium increases. More generally, the subtle loss of morale generated by a system that, for the first time in American history, made a physical necessity problematical will begin to abate, with a consequent rise in personal initiative and national cohesion.

    Like Reilly, I too am sanguine about the effects of the passage of this bill, and I see it as an ultimately positive development. This may seem a little strange, because by political temperament, I ought to be bitterly opposed to this enterprise. However, John Reilly was instrumental in convincing me otherwise. He is one of the most reasonable men I know, and he exemplifies the virtue of hope, which makes for a critical difference in perspective.

    Reilly has been making the argument that a reform of the health care system in the United States will be a liberation. His reasons for saying so are fundamentally conservative, which annoys pretty much everyone involved, but I believe he is on to something. He has been arguing that the current system limits our free market system by tying people to their jobs which offer good benefits, and making businesses bear the brunt of funding health care, when our foreign competitors do not need to do so. He also insists that health care is not a right, and cannot be. It is rather a matter of public order and entirely practical. Rights-talk is far too absolute to be productive.

    One of the chief difficulties with the American system is its opacity. The quality of the health care is really quite good, and it is a vicious calumny to say that the uninsured have no health care in America, rather, they get all their care at the Emergency Department, which is both expensive and inefficient. The rub here is that the people who don't have insurance often wouldn't go to the doctor until they are really sick anyhow, so there is only marginal improvement to be made here. From my time working in the local hospital, I know that the truly poor often simply don't pay their medical bills, which means the service is subsidized by those of us who do pay. It would be better if we did this up front, instead of persisting in sending overdue notices to people with no means or intent to pay anything.

    The pricing is the really bizarre thing. If you look at a bill for any medical service, you see a ridiculous value that is invariably cut by half or more if you have insurance, but it is nearly impossible to find out what this value is beforehand. The actual cost of medicine in the US is practically a trade secret. Billing codes are actually copyrighted by the American Medical Association, and guarded fiercely. The Magistra needed hand surgery recently, which was covered well by my excellent insurance, but we actually were unable to find out beforehand how much it would cost.

    No business can run this way. Not only can you not determine the cost beforehand, the matter can take 3-6 months to be paid in full. Costs would go down in American health care if only the accounts receivable were smaller! One of the best experiences I ever had was when I went to the ED and forgot my insurance card. When I got the bill I forwarded it to my insurance company, and then I paid the negotiated price directly and got reimbursed. I carry a large savings balance, so this kind of thing is easy for me, but if this is the best to offer the system needs work.

    Since I work in the medical device industry, I know that the prices that the manufacturer ultimately sells for in different countries are not all that different. The US is the single biggest medical market in the world [followed closely by Japan], but medical devices still make plenty of profit in the various and sundry health care systems in the rest of the world. I have to laugh when the claim is made that switching to some kind of national health care system in the US will inevitably stifle creativity and innovation. It is certainly possible, depending on how stupid the system is, but all medical device manufacturers compete globally, with a majority of our customers being in allegedly stifling systems already.

    The pharmaceutical companies may be in a different boat. I've been of the conclusion that Big Pharma is in trouble for a while now. Actual innovation there is slim, most recent drugs seem to have worse side effects than the things they treat, and R&D costs are ridiculous for new drugs. Don't mistake me, R&D on a medical device can easily cost hundreds of millions of dollars, but I think the results are actually better, and you don't have to worry about generics stealing your market once the patent runs out.

    There is a lot of room for cost-improvement in the American health care system, but the really interesting thing is that this political development will not result in creeping socialism, but rather presages the return of a more traditional society. The Age of Autonomy, and both right- and left-libertarianism is ending.

    Sunday
    Mar282010

    Mathematics and Computer Science ebooks

    Friday
    Mar262010

    The 7 Year Itch

    Friday
    Mar262010

    Master Bladesmith

    This reminds of the knives of Frank Richtig.

    From: 37signals

    h/t Tom

    Thursday
    Mar252010

    First Things Twofer

    Wednesday
    Mar242010

    Corn Syrup and Rats

    Steve Kamb at Nerdfitness posted a link on Facebook about a study that supposedly proves that high-fructose corn syrup makes you fat. I've always been pretty dubious about the bad rap that HFCS gets, and Archer-Daniels-Midland didn't even pay to have me say that. I'm not saying no mind you. Anyhow, I was roused to action, so I went and got the paper so I could read it for myself. I found it, interesting. More about that in a moment. First I want to talk about the press release.

    Staged Lab PhotoI really liked the picture that went with the article. You should click on the picture for the full version. It is very clearly staged, but I'm not blaming the PI or his students for that. Science is tedium occasionally punctuated by excitement that turns out to be either a complete failure or the best thing that ever happened. This makes for bad photos.

    The Magistra has modeled for science lab photos several times for NAU, so I have an inside scoop on what you do during photo ops. What is done has no relation to what actually goes on in the lab, but it looks much better on glossy paper. The number one thing the photographers insist on is colored liquids, when almost everything you ever use in a lab is clear. There is a lot of standing about trying to look busy or transferring liquids in improbable fashions.

    On to the study itself. I don't really think this is a masterpiece of science. The sample sizes were small, the DOE was unbalanced, and the graphs blow. I copied the results table to share with those of you who can't easily access journals. This nicely illustrates my complaint about the design of this experiment. The three experiments all have different sets of conditions, and a bad side effect of this is the sample size is really small for each grouping. If they had consolidated into fewer conditions, the sample size would have been much better [20 instead of 10 for Experiment 1 with just 24 hour corn syrup and controls].

    Here are the main graphs from the article. I suppose the graphs really are not that bad, but if you look at the legend, you see that the graph shows "means±SEM". SEM, WTF is that? I guessed it meant standard error of the mean, but I'm used to seeing that called SE or SE(x). That is one of the perils of statistics, that there is no real standard symbology or nomenclature. But, I still thought it odd to graph that, so I did a Google search, and the first result confirmed my suspicions.

     

    The problem here is that we understate the variability by using the standard error of the mean, which is really nothing but the standard deviation divided by the square root of the sample size. You would think I would be for this, because it punishes you for small sample sizes, but really I care more about individuals than means in most cases. Think of it this way, if I gave you one of these fat rats, could you tell me which group it came from by weight alone?

    Here is a graph to illustrate what I mean. I took the end point body weight data from Experiment 1 and created distributions. I assumed a normal distribution, which is really not a bad assumption for weights, and I graphed them by the stated mean and calculating the standard deviation from the standard error of the mean. Red is the 24 hour corn syrup group, blue the 12 hour corn syrup group, green the sucrose group, and black the control group.

    For all but the 12 hour corn syrup group, you probably couldn't tell a random rat apart. Even with the 12 hour group, you have only about a 50-50 chance of getting it right, unless the rat is really, really fat. This is what I mean by understating the variability. If we saw a graph of the raw data, I bet we would see a whole lot of nothing. Unfortunately, scientific protocol does not usually offer us the raw data.

    I think this is bullshit. These protocols are from an age when data was hard to come by. It is no longer. I provide the raw data with every technical report I file, so I feel that this is not unreasonable for scientists to do either. I usually have more data points too. The 96 data points this study depends on would not break the server the journal is hosted on. Since we have no raw data in the journal article, I decided to simulate it. I did four simulation runs and printed the results. The simulation data is 10 points drawn from the distributions graphed above for each group, and then the groups are compared by an ANOVA as stated in the paper.

     

    They do pretty well in the ANOVA. A real simulation would need at least 1000 runs, but I'm busy, so here are four. In three of the four we get the same significance results. This is good. However, they do less well in the pair-wise comparisons, which are really more important here. Only one out of four showed clear groupings, the reason being the sample sizes were too small for this kind of comparison.

    Really, the authors just tried to do too much with too little. But, I suppose that is the real story here. Even though this was funded by taxpayers [me], there wasn't enough money for enough rats to do a good job. Now, I know good and well that animal research is surprisingly expensive, but what is going on here is this research is just not important enough to justify doing it right, which tells you all you need to know.