HNHacker News
TopNewBestAskShowJobs

kgrin

1,443 karma · joined March 23, 2009

kevin at activefrequency.com et al

[ my public key: https://keybase.io/kgrinberg; my proof: https://keybase.io/kgrinberg/sigs/WEI_d3HGjUIUBsOtsfBAyLlxPP2GxVWvgMAg5oxxwJE ]

submissionscomments
kgrin··on [dead]
True enough RE: importance of saving the financial system, though there's plenty of room for debate about the best way to go about it. For instance: what should the government (read: taxpayers) get in return for their guarantees of these companies' debt? (As any other lender would!)

I'm all for saving the financial system - but to the extent that we're socializing losses, it's importance to balance it by socializing some of the future gains as well (via stock, for example) - otherwise, we get into the sort of crazy loop where there are effectively unlimited guarantees for screwing up.

kgrin··on [dead]
Sure, but to the extent that the author is comparing two numbers to generate shock value, the fact that he's really comparing an amount of apples spent to a number non-inflation-adjusted oranges lent makes the whole thing a bit misleading.
kgrin··on [dead]
That 300 million isn't inflation-adjusted, is it?

Also, most of the $4-trillion-range figures are of guarantees, not actual dollars spent or committed - a good deal of that money has been or is going to be repaid. If the banks default on literally all of it, we're going to have much bigger problems; certainly, some of that money is truly at risk and is likely to never be repaid - but hardly all of it.

Now, that doesn't mean any of it is a great idea - a government guarantee of your debt is certainly a valuable give-away! - but it's not as if "we the taxpayers" wrote a $4 trillion check to Citibank. Distinguishing between a line of credit and a "gift" is pretty important (even though the LOC itself undoubtedly has value).

Edit: just to be clear, I actually agree with the broader point that our priorities are pretty screwed up and that there's a whole host of problems with the way these "bailouts" and whatnot have been handled. That said, exaggerating or making "truthy" comparisons just makes the whole thing less credible, which is unfortunate because the broader point is valid. Call it "the Michael Moore effect."

kgrin··on Django for Rails devs
Seriously, ModelForms are pretty much the greatest thing since sliced bread.
kgrin··on Acquisitions: cull the managers
The author makes a telling comment about how "worker-bee" programmers can be reassigned that displays the same contempt for developers as the managers who call them "resources". Call it a pet peeve, I guess.

One other factor to consider is that if a company were to do multiple acquisitions and consistently behave this way (immediately lay off every manager of the "losing" internal product), it would likely create a not-very-healthy culture of fear and opposition to acquisitions.

That's not to say, of course, that underperforming managers should get a break "just because". Rather, it seems reasonable to say that their performance and future roles should be re-evaluated along with every other affected party's - and yes, they probably shouldn't end up leading the newly acquired team a la Collabra.

(Keep in mind that sometimes the internal product's lack of success is quite often due to senior management's decisions... the same senior management that makes the decision to acquire).

kgrin··on Your Race Affects Whether People Write You Back (WRT Online Dating)
If I misunderstood your point, I apologize. Care to correct?

Here's what I understood to be the debate:

Original comment you quoted: > A lot of dating and who a person would consider for relationships is based on social, cultural and economic considerations, and it’s a simple and unfortunate fact that generally speaking whites are at the higher end of the social and economic spectrums.

Your response (paraphrased, I hope fairly): Vietnamese entrepreneurs and Jewish scientists did quite well, and that's not unfortunate at all. Inequality isn't inherently bad, and we should focus on bringing people up rather than down (or for people to bring themselves up, etc.)

Either your response is a bit of a non-sequitur, or the logical inference is that the fact "whites are at the higher end of the socio-economic spectrum" isn't 'unfortunate' (which is sort of a squishy word, but that's just the OP's language). I understood that to be the point you were responding to - if that's wrong, then I apologize and please do set me straight.

My point was that the OP calling better socio-economic 'outcomes' for whites "unfortunate" is pretty reasonable, as there's a good deal of evidence to suggest that it's not just the result of benign factors (see: history of slavery and segregation; evidence of present-day racism, as in the article I linked).

Likewise (and I sort of hate to use this example), a monopolist's market dominance may be the result of a superior product, or an unfair/illegal abuse of [whatever]. In the latter case, calling that monopolist's position "unfortunate" doesn't seem out of bounds - even if the best remedy is to bring the other market actors up, or even do nothing.

You called the OP out on calling white advantages "unfortunate" - I'm saying, it is.

kgrin··on Your Race Affects Whether People Write You Back (WRT Online Dating)
The "unfortunate" comment seemed to be directed at the relative inequality, not the fact that whites are "succeeding". I think it's pretty tough to make the case that economic and educational inequality in the United States along racial lines is intrinsically a good thing, particularly given the historical context.

Inequality may be the least-bad outcome given the menu of potential remedies ("Harrison Bergeron" and all that), but it's not a great thing in and of itself. Particularly given the historical and occasional modern-day evidence of straight-up racism (see: http://findarticles.com/p/articles/mi_m0DXK/is_9_20/ai_10452... ), I think it's reasonable to at least scrutinize situations of high inequality to figure out whether they are, in fact, just caused by some people working harder or being more lucky or what have you, or whether there are actual problems.

On the merits, though, I think OKCupid's matching algorithm (the questions) theoretically accounts for the social, economic and cultural differences - what their data shows ("suggests" is probably more appropriate) is that people react to race even after controlling for the other factors, which is at least interesting.

Edit: just to clarify and follow up - here's my claim: if the relative dominance of Ashkenazi Jews among American scientists, or Vietnamese women among nail-salon entrepreneurs, were the result of, say, discrimination, or government subsidies at others' expense, or some sort of other external factor that made the "competition" unfair, then it would be fair to categorize that inequality as "unfortunate". As far as I'm aware, that's not the case for either of those examples (and in fact is somewhat the opposite). But if it were (and the author of the original comment we're talking about seems to make that assumption about whites being better off on many socioeconomic measures), then calling that inequality "unfortunate" wouldn't be out of line.

kgrin··on Fred Wilson: Startup Visas
As someone pointed out recently (sorry, can't find the original blog post) this sort of plan really skews the power dynamic between the entrepreneur and VC/angel. If you really make this visa contingent on funding, then in addition to having the power of the purse, the VC will have the power of immigration.

One solution, of course, is to make it independent of funding, but then distinguishing between "founders" and "people who want to move to the US and might one day start a company" gets pretty hard, policy-wise. Now, I have no particular aversion to the latter group either, but plenty of Americans apparently do.

The real answer, as many have suggested, is to fix immigration policy rather than carving out exceptions for our preferred niche.

Edit: found original blog post referenced above - http://kirkwylie.blogspot.com/2009/09/founder-visa-ignores-i...

kgrin··on Obama mandates FCC to support Network Neutrality
Technically it's not Obama, but an Obama appointee who's doing the mandating (and wrote the linked blog post): Julius Genachowski (FCC Chairman). Obviously the proverbial buck stops with Obama, and it's fair for him to be credited, but it's important to recognize the distinction.
kgrin··on Study Gives High Marks To Retailers' Clinics
Generally speaking, PAs and NPs must technically be "supervised" by a physician. I think the degree of supervision varies by state, but it's typically not case-by-case supervision - rather more of a general sign-off sort of thing, with availability for the "tough cases". That sort of sounds like what you're describing, though obviously there's a wide spectrum of how involved the doctor can get.
kgrin··on Is The Microsoft Stack Really More Expensive?
Scaling horizontally is considerably more expensive on the MS stack, as the StackOverflow folks discovered recently ("discovered" is a poor word choice maybe). I distinctly remember a blog post or podcast where Jeff basically lays out the numbers and says that the software costs make scaling up more enticing than scaling out.
kgrin··on Poverty Kills
And what happens when, faced with the rational person's ultimatum ("learn to fish or you get nothing"), the man chooses nothing? Surely it's his own fault, but suffering still happens. Of course, that's a small-scale, contrived example, but there are plenty of areas in public policy where many or most people's behavior isn't rational.

Health care is actually a great example of this: classical theory and basic common sense suggests that if you make people pay for their health care more directly, they'll have more of an interest to negotiate a better deal, or at least comparison-shop. This is occasionally true, but far more often, people aren't actually comfortable bargaining with the surgeon who's going to be handling their innards.

Likewise, with health care, people often make "irrational" decisions and go into $500K debt to prolong an 89-year-old grandparents' life for 6 days, because life, death and health just seems to be one of those areas that we're not wired to be rational about.

Health care aside, there are all sorts of other examples of irrationality - for instance, I remember seeing multiple studies in which people preferred a lower reward as long as everyone got the same reward (basically they valued relative well-being and parity over absolute well-being).

The point of all this is that while rationality is a great tool for "solving problems" generally speaking, when you deal with people you really must take into account the fact that most people (and I suspect even most people on HN, which is a pretty atypical group!) aren't all that rational in many, many areas.

kgrin··on Why I'm Quitting Gmail
We have an app for sale on the Android Market (which uses Google Checkout for transaction processing), and it seems like every week we have a support email claiming we charged them twice.

We always apologize (even though it's not our fault), explain how the authorization charge works, assure them it'll be gone in a day, and point them to Google's help (bottom of): http://market.android.com/support/bin/answer.py?hl=en&an...

I do wonder why it's done that way on Checkout and not other payment systems, or why it seems more noticeable there; it seems to cause an awful lot of confusion.

kgrin··on Dirty Coding tricks used in Production Video Games
This vaguely reminds me of how some people set clocks 5 minutes ahead to make themselves think they're late.

Of course, people also tend to get used to that pretty quickly and compensate accordingly, which doesn't sound like a problem here.

kgrin··on Tell HN : JetBlue All-You-Can-Jet Pass Sale Ends Today
Yes, you have to call: 1-800-JETBLUE (538-2583), prompt 4.

I believe you also need to register for their frequent flyer program (free?), though I'm not sure if you can just do that when you call.

kgrin··on Ask HN: Web programming for a C++ server developer?
In general, Django is considerably more explicit about what it's doing (particularly after the "magic-removal" branch two years ago). The implicit trade-off is that some things that Rails does automatically you have to declare explicitly in Django, but it hits the sweet spot for me.

My understanding is that Rails has trended that way lately, but since I don't use it day-to-day I can't say whether that's the case.

I can't compare the Django source to Rails, having never really read the latter, but on its own it's pretty good. Like any large-enough pile of code, there are ups and downs, but whenever I've had to dig in and understand what's going on, it's been fairly straightforward to follow.

kgrin··on Judge Throws Out Red Light Camera Tickets As Program Declared Illegal
From the actual article linked to by the submitted post:

"Schwartz saw the error in the announcement timing as more than a technicality, citing the false assertion of Police Chief Paul Walters that motorists would be given 4.4 seconds of yellow time at enforced intersections. Records show that seventeen of the city’s eighteen camera intersections had yellow times of 4.0 seconds or less. The vast majority of red light tickets in the city were mailed to vehicle accused of entering an intersection less than half-a-second after the light had turned red."

And again, yellow doesn't mean stop. It means "stop if it's safe". If yellow meant unequivocally stop, then everyone would be slamming on the brakes - a recipe for accidents.

kgrin··on Judge Throws Out Red Light Camera Tickets As Program Declared Illegal
That's not really sound logic - if you see a yellow light and there's enough time to safely stop, you should stop. That's an important qualifier. Indeed, that's why the duration of yellow lights differs based on the expected speed that people will approach them - because it's not, in fact, always safe to stop when you see yellow. (As a thought experiment, ask yourself why we have yellow lights at all, and why they're not instantaneous).

What the city did in this particular case was deliberately shorten the yellow light to a lower time than is actually legal or safe - just to make a few extra bucks.

Now, if your argument was more that "well, the city's wrong, but most of the people who ran the lights are still guilty", then the argument is really similar to the one about the exclusionary rule (evidence in criminal cases can't be used if obtained through an illegal search). Without some way to punish the city, there's no disincentive for them not to keep screwing over people - in this case, in a way that actually increased the accident rate. There's got to be some pain felt for malfeasance, and it sounds like that's exactly what's going on here.

kgrin··on Javascript and DOM Load Time Tester with jQuery
Can someone explain why this is better than a real profiler, or something like YSlow?

I'm not criticizing - just trying to figure out what I'm missing.

kgrin··on Is US Chief Information Officer Vivek Kundra a Phony?
The details in both Daschle and Geithner's cases are fairly weird:

Daschle's mistake was in not reporting a free car/driver as a gift (and hence not paying taxes on the value). There's a whole other conversation to be had about the culture of ex-Senators getting free stuff from rich friends "out of the goodness of their hearts", but it's not like the guy just decided not to pay a bunch of taxes.

Geithner screwed up by not paying his own SS/Medicare taxes while he worked at the IMF. The wrinkle is that unlike most private-sector or government jobs (which is where he was before and after his time at the IMF), international organizations (or at least those) don't do automatic payroll deduction of those specific taxes, so you have to deal with it yourself. Again, not a super-common situation.

So, yeah, these guys screwed up, were in positions where they should have known better, and it's fair to hold them accountable. But it's a little unfair to suggest that they were acting in bad faith when I'm not sure that I at least would know the arcana of those rules off the top of my head (though it makes a fabulous argument for simplifying the tax code...)

To look at it another way: Daschle underpaid by $140K while making millions, knowing full well that he may want to get back into government (which would involve vetting). Is it more likely that he: a) was knowingly risking his future career by not reporting the car service that he knew was a reportable gift; or b) being blissfully ignorant of how those rules worked?

kgrin··on [dead]
Fair enough; though insofar as we talk about removing the barriers to starting a business, "how will I provide for my [or my employees'] health care?" is a big one.

Personally, I'd love to see a study of the impact on entrepreneurship/small business starts of MA's health reform.

I'm walking anecdotal proof: I quit a "real" job to start a startup in no small part because I could buy into a reasonably priced individual pool and not worry about losing my (former) employer's group coverage. But I'd love to see some hard numbers, if anyone knows of a good study.

kgrin··on Lies, Maths, And Health Insurance
The latter point isn't entirely true - at its core, buying insurance isn't necessarily about using more than you put in over time, but rather mitigating risks (hedging). That's not necessarily how everyone treats the current system, or for that matter how it's set up, but there's certainly a tension between the "insurance" model and the "health savings account" model (and the current system sort of tries to do both).

In terms of insurance companies being run as non-profits, let me ask two pointed (and I assure you, not-meant-to-be-trollish) questions:

1) If insurance companies should be non-profits with set salaries and so forth, why not just have the government do it at that point?

2) Two recently prominent examples of non-profits set up to further socially useful goals without actually being government agencies are Sallie Mae and Freddie Mac. Those didn't turn out so hot. How would you set up these non-profits to avoid the same sorts of problems?

Actually, I should throw in a third: does anyone know of any examples (successful or not) of a non-profit insurance entity or cooperative?

kgrin··on Lies, Maths, And Health Insurance
There's a pretty clear medical consensus (based on tons of data) that diet and exercise affects the development or progression of diabetes (well, type II). There's also a wide consensus that professional counseling and monitoring helps (though I believe it's harder to get really "clean" studies to support this, since you'd need a control group of people who you monitor closely enough to get good comparison data without actually talking to them about what's going on).

The follow-up question, of course, is whether (or how much) the cost of checkups/counseling affects whether/how much people seek it out, or whether that behavior is more idiosyncratic (i.e. "some people just don't care about their health, so even if checkups were free they wouldn't bother").

To be fair, I haven't seen any study that conclusively proves the issue (which of course doesn't imply that one isn't out there). However, if we're really going to make the claim that economic incentives don't matter, then we really need to question a lot of other assumptions as well. I'm pretty comfortable saying that at least on some level, to some degree, economic incentives have some impact on people's behavior in seeking care.

In terms of "we'll just make it up on preventative care", I agree, that's not a panacea. At the same time, a lot of the money that's slated to be spent on covering the uninsured is still spent today, just in really stupidly inefficient ways.

For instance, because you generally can't be denied emergency room care, people without insurance often end up using ERs as primary care. Because ERs are primarily meant to handle, well, emergencies, the standard of care is generally more expensive than a clinic or "regular doctor" (this means things like faster and more expensive tests, drugs, and so forth). Many of those people simply can't pay, but the hospital treated them anyway (which, just to be clear, is a GOOD thing - it would be a pretty horrid society where we let people bleed to death or what have you). Many hospitals would have a hard time staying open if this went on indefinitely, so the government reimburses them for this uninsured care.

So, at the end of the day "we the taxpayers" still pay for a good deal of uninsured care... we just do it via an indirect subsidy to hospitals for expensive ER care, rather than paying directly for (much cheaper) clinic care, or something along those lines. As with most things, the less directly we're tackling the problem, the more inefficiency there is in the system.

It would be foolish to claim that all the costs of a big health care/health insurance overhaul would be paid for by money already in the system - clearly, that depends a great deal on the details. But at the same time, it's important to remember that some of the money can come from within the system: at a minimum, if we reduce the number of uninsured people, we can reduce payments to hospitals for uninsured care.

It's not a zero-sum game - electronic medical records and that sort of stuff aside, there's a lot of honest-to-goodness inefficiency and perverse incentives in the system, and that's what (a good) reform ought to tackle.

Edit: just to be clear, reimbursements for uninsured care take many forms, and there's another, more obvious one: higher prices for everybody else. The salient point is that much of this care is being provided anyway, so why not do it more efficiently?

kgrin··on Lies, Maths, And Health Insurance
I don't have the studies in front of me, but it generally depends. In the case of diabetes, for instance, a lifetime of diet monitoring and counseling by a doctor or nurse is still way cheaper than insulin or amputation.
kgrin··on Lies, Maths, And Health Insurance
That's true, though if you're on the hook for a person's catastrophic care, you may very well want to give them economic incentives (i.e. cheap/free doctor's checkups) to invest in preventative care.

Now, this doesn't mean that we ought to necessarily have a big, bad program (government or private) control everything.

But it does mean that the "insurance is for catastrophes, everything else should be out-of-pocket" model isn't necessarily as simple and wonderful as it sounds.

kgrin··on Ask HN: Why is the number of greatest magnitude on the left?
That's true enough (93 = "quatre-vingt-dix trois" = "four twenties and three"), but it's still the case that the more significant digit is listed first - it's just composed by multiplying.
kgrin··on Zed Shaw: Why I (A/L)GPL
Well, in a sense that's what he's doing now - the whole "Open source to open source, corporation to corporation" bit.
kgrin··on Lost suitcases, or how to DoS an airport
Rewound the DVD?
kgrin··on Keeping News of Kidnapping Off Wikipedia
Different circumstances are sometimes, well, different. Sometimes a captive's life is more in danger if the captors consider him or her valuable (like a NYT reporter). Sometimes publicizing the situation helps bring attention in a positive and helpful way (shaming the captors, forcing the government to do something, etc).

And sometimes - often, in fact - you just don't know, and you make the best judgment call you can.

I'm not claiming that the Times always acts correctly in these situations, and it's certainly worthwhile to question what the motivating factors are in whether, when and how to hold a story.

But it's also not really correct to say that all kidnapping situations are exactly the same and must be [not] publicized. They're simply not.

kgrin··on JQuery Tools - The UI library for the Web
Yes; it looks like it's more focused on a lightweight set of core features (obviously opinions as to what constitutes "core" features may vary).

It actually seems more reminiscent of early mootools.

← PreviousPage 7 of 8Next →