HNHacker News
TopNewBestAskShowJobs

salmonellaeater

910 karma · joined October 16, 2012

submissionscomments
salmonellaeater··on Lessons from NYC’s improperly anonymized taxi logs
These are old lessons: in 2006 AOL [1][2] and Netflix [1][3] both released datasets that were supposed to be anonymized but were easily de-anonymized. There are older examples based on Census data[4]. It's difficult if not impossible to release a dataset that is both useful and truly anonymized; in Schneier's words:

The obvious countermeasures for this are, sadly, inadequate. Netflix could have randomized its dataset by removing a subset of the data, changing the timestamps or adding deliberate errors into the unique ID numbers it used to replace the names. It turns out, though, that this only makes the problem slightly harder. Narayanan's and Shmatikov's de-anonymization algorithm is surprisingly robust, and works with partial data, data that has been perturbed, even data with errors in it.

[1] https://www.schneier.com/blog/archives/2007/12/anonymity_and...

[2] http://www.securityfocus.com/brief/286

[3] http://www.securityfocus.com/news/11497

[4] http://crypto.stanford.edu/~pgolle/papers/census.pdf

salmonellaeater··on Types Are The Truth
You could certainly handle that kind of error using a type system. There could be a scoping rule where certain kinds of operation (i.e. close file) put the reference to the object out of scope. The way Haskell chains I/O operations using monads is another way: the result of the 'close file' operation is basically a world that doesn't contain a reference to the file.
salmonellaeater··on Data Modeling in Graph Databases: Interview with Jim Webber and Ian Robinson
"The problem with a join is that you never know what intermediate set will be produced, meaning you never quite know the memory use or latency of a query with a join. Multiplying that out with several joins means you have enormous potential for queries to run slowly while consuming lots of (scarce) resources."

Anyone with even a meagre understanding of databases will put indexes on join columns. If the data model is complete, then the joined columns will be modelled as foreign keys (conceptually the same as a relationship in a graph DB) which force indexes. I think they are talking more about problems with ORMs, where the ORM might construct unexpected queries that don't hit indexes. This is an ORM problem, not a relational DB problem.

One of the major promises of relational DBs was that you could write code describing what you wanted, and the DBMS would figure out how to efficiently find it for you. This promise was derailed by the push to merge relational models with object-oriented models (i.e. ORMs), but it's not dead. What we need is a more powerful SQL, one that doesn't require boilerplate and can do things like recursion (making every database a graph database). We need a SQL that makes the application code seem like boilerplate. We need the equivalent of type inference for joins; let me say A join D and the DBMS infers I mean A->B->C->D and figures out the cardinality of the result. We need result sets that are graphs instead of one-list-fits-all. These are all things that can be modelled in an RDBMS without losing its expressiveness.

salmonellaeater··on Why Online Communities Decay Over Time
Game theory is just an explanation; people don't use it explicitly to make decisions, they just follow their gut, and evolution has caused our guts to produce rational decisions most of the time. It's the same as the theories of attractiveness based on symmetry and secondary sex characteristics: you don't look at a potential partner and think, "He/she looks so symmetric, I bet if I were to have kids with him/her they'd have a higher chance of survival." You just intuitively find the person attractive.
salmonellaeater··on Working on The Witness, Part 1 (2012)
This is called a boustrophedon decomposition[1] (the name comes from the movement an ox makes when plowing a field). It's used for minesweeping robots among other things. Constructing the decomposition simultaneously covers the space being explored, so a robot can explore and mow grass at the same time. Applied as a coverage algorithm, it's much more efficient than a random walk.

The robot is usually modelled as having a non-zero footprint since coverage is often a goal, but a radius of zero can be used for exploration. To explore a space with minimum obstacle radius R and robot radius r, each sweep can be 2(R+r) units wide.

[1] http://planning.cs.uiuc.edu/node352.html

salmonellaeater··on How Did Dread Pirate Roberts Acquire and Protect His Bitcoin Wealth? [pdf]
mavhc's joke[1] is apparently a little too close to reality.

[1] http://princessbride.wikia.com/wiki/Dread_Pirate_Roberts

salmonellaeater··on The one statistic you never hear about drugs
It does distinguish, at least between smoking and alcohol:

While many fewer Americans are killed from alcohol than from tobacco, the absolute percentages are somewhat deceiving; whereas the lifespan of current smokers is about 12 years shorter compared to those who have never smoked, the average death from alcohol comes a whopping 29 years prematurely.

salmonellaeater··on Circle of fifths and roots of two
"In classical music from Western culture, a fifth is the interval from the first to the last of five consecutive notes in a diatonic scale."[1]

CDEFG -> CG is a fifth

[1] http://en.wikipedia.org/wiki/Perfect_fifth

salmonellaeater··on Brazilians welcome genetically-modified mosquito to help fight dengue fever
One effect of parasites and diseases is to increase diversity. If a species becomes too common, a disease or parasite targeting it can flourish, cutting back its numbers and allowing other species to take its place. Eradicating mosquitoes might have an unexpected effect like making the number of rats or some other vermin to explode.
salmonellaeater··on Functional Programming in JavaScript is Garbage
The author is correct in saying that JS was not designed for functional programming. The issues they raise are real issues, and a different language design would make functional style much easier to write.

The issue with immutable data is not that Javascript's model is confusing, it's that there is no way to write "this array is immutable." Without that guarantee, you can't allow multiple data structures to reference the same copy of data because the data might get modified unexpectedly.

There is a similar issue with tail calls. The compiler could replace calls in tail position with jumps, but it's important that this be part of the language spec. Without a guarantee in the language spec, code that runs fine on one system may blow up with a stack overflow on another.

It's impressive that JS works as well as it does when written in a style different from what it was designed for. Many languages make other styles so difficult that it's not even worth trying (looking at you, Java).

salmonellaeater··on A Fervent Defense of Front-running HFTs
> If we had no price discrimination, perhaps because of a regulation requiring market makers to leave their quotes out in the market for 60 seconds, then ... [t]he net result is that Big Joe would pay more to trade and Little Billie would pay less.

The debate about HFT is largely about who gets to benefit from new information. Suppose someone does a lot of research and discovers that a firm is undervalued. Should they get to profit from the information, at the expense of current stockholders? Or is it better for stockholders to quickly figure out that their holdings are worth more? HFT helps the stockholders in this case, because it moves the price faster to its eventual higher equilibrium before the researcher can buy up many shares at the (low) current price. The researcher may have expended a lot of time and effort to get this information, and with the existence of HFT their work may not be sustainable.

Contrast this with someone who accidentally discovers public information that could be profitable, or someone who engages in insider trading. In both these cases it's hard to argue that we as a society should create rules that let them profit at others' expense.

In other spheres of life (software patents, copyrights), the law over-protects dubious discoveries and creations. Is information related to stock pricing different? Are we under-protecting it, such that we should let someone with new information collect more profit than they can today?

salmonellaeater··on The Disruption of Cable Television Has Arrived
Don't underestimate the power of bundling. Even if content becomes decoupled from the cable companies, bundling is still a powerful way for content companies to get a small amount of revenue for content they know you don't care much about.

The way it works is this: consider customer A, who isn't subscribed to any TV service. Customer B is subscribed to HBO only. Which customer is likely to pay more for a subscription to a different channel, say AMC? The content provider wants to charge less to the customer who already has HBO, since they know that customer gets less marginal value for an additional channel and is willing to pay less. The best way to figure out which customers already have HBO is to bundle the services.

This is also why dreams of splitting the 100-channel packages into separate deals for a dollar a channel are so misguided. If you're trying to buy a single channel, the content provider knows you really want that channel, so they'll charge you $40 for it. If you already have 99 channels and you're buying channel 100, they know you don't want it at all and might charge an amortized $0.05.

salmonellaeater··on Virtual Spaces, Real Data
> FB has a well documented habit of arbitrarily changing those [privacy] settings without the user knowing.

This is exactly the reason why I am dismayed at the acquisition. FB has demonstrated repeatedly that it has a very cavalier attitude toward users' privacy, and wants to insert itself as an intermediary in everyone's communications so it can keep amassing data. FB also doesn't like people keeping separate personas, despite the fact that everyone presents a different version of themselves to their families, their coworkers, their Quake clans, their political parties, and the public. FB is like the friend who has no discretion about what secrets they share with whom; you can't tell them anything truly private, so they can never be a friend at all. Inviting them to watch everything you do on the Oculus Rift is a bad idea.

salmonellaeater··on How Animals See the World
Although cameras have built-in filters that block UV and IR light, the IR portion is often weak and allows some light through. By adding a filter[1] that blocks visible light, there is often enough IR left to expose a photo[2]. With some caveats[3], it can be a cheap way to see the world in a different part of the spectrum.

[1] Hoya R72 for example

[2] A tripod is usually necessary since only a fraction of the light makes it through the filters. Even in bright sunlight a long exposure time may be required.

[3] Not all lenses work well in IR. Some have internal reflections that lead to "hot spots" or white patches, and some have bad optics for IR that causes blur. Similarly, some cameras' IR cut filters are too strong to be useable.

salmonellaeater··on Apple's Buttonless Mouse Came From Steve Jobs' Misunderstanding
I was assuming American English, since the site is based in New York. Even so, it turns out there is a lot of nuance to this grammatical question. The Chicago Manual of Style (which is for book editing) prescribes 's, while the AP Stylebook and Libel Manual (for press, naturally) prescribes ' only.

So since it's an American news site and follows AP guidelines, the title is perfectly normal. Carry on.

salmonellaeater··on Apple's Buttonless Mouse Came From Steve Jobs' Misunderstanding
The article's title makes a daring statement. The possessive of a name ending in s usually uses 's:

James's Forbes's

Omitting the s after an apostrophe is usually reserved for historical or Biblical names (because historically they were pronounced the same as the bare word, as opposed to current pronunciation with the extra -ez sound):

Socrates' Moses' Jesus' Jobs' (?)

salmonellaeater··on How are humans going to become extinct?
Even an AI with innocuous goals will destroy humanity[1]. It would have to have explicitly pro-human goals in order to be anywhere near 'safe.'

[1] http://wiki.lesswrong.com/wiki/Paperclip_maximizer

salmonellaeater··on Prison Switcharoo
Presh Talwalkar posts weekly logic puzzles[1]. His blog is a good source of interesting applications of economics and game theory to the real world.

[1] http://mindyourdecisions.com/blog/

salmonellaeater··on 30 Things I Learned by Spending One Month in Thailand
I left a top tech company two years ago and moved to a similar country (Taiwan), and I have a few counterpoints.

- If the cost of living is low, odds are the wages are low too. If you're living within your means and saving 30% of your income, you're better off saving 30% of $150k than 30% of $40k. The cheap cost of living only applies if you're living above your means, vacationing, or retiring.

- Cheap labor enables different ways of doing things. This includes transit like tuk-tuks and taxies, but Taiwan has another example: there are no parking meters. You just park, and a police office comes by every half hour or so and leaves a bill on your windshield (linked to your license plate).

- If everyone seems exceptionally friendly, that means you're expected to be friendly too. My friend here has a neighbor who hates him because he doesn't stop by to chat all the time like the other neighbors.

- English names of businesses, like the "Nice Pharmacy" example, are usually targeted to locals who don't know much English. They have to stick to words and phrases that locals know.

- It shouldn't take a trip abroad to figure out your priorities, and traveling won't even necessarily help.

salmonellaeater··on Flappy Space Program
For a real treat, try Desert Bus[1]. The scoring works like soccer: 1 point and you've basically won.

[1] http://www.newyorker.com/online/blogs/elements/2013/07/the-w...

salmonellaeater··on Streetpong [video]
There are... just not in the US. They're actually a mixed bag so it's not necessarily a tragedy that the US doesn't use them:

Based on the operational studies, only the red time countdown device has a positive influence on safety and delay but increases the travel time and decreases the capacity. The pedestrian countdown signal, however, has little improvements for both travel directions.[1]

[1] http://www.ce.siue.edu/faculty/hzhou/publication/What%20do%2...

salmonellaeater··on NC Electric Car Drivers Begin Paying $100 Annual Fee
For half the US population (those in regions that use more renewable power sources), electric vehicles are equivalent to about 70mpg in terms of emissions, so they're a clear win. In the regions that use mostly coal, it's more like 35mpg, so a really efficient gasoline-powered car is a better choice.[1]

However, I'd argue that even a car that gets 35mpg is better than average and its use should be encouraged.

[1] http://www.forbes.com/sites/jamesconca/2013/07/21/are-electr...

salmonellaeater··on Driving Is Going Out of Style
I keep a few disposable plastic bags in my bike bag for this kind of situation. They're almost weightless, and balled up take up very little space.
salmonellaeater··on Carry Your Instrument On The Plane – It's The Law
> As a safeguard, take an all-purpose screwdriver and pack it in the gig bag.

After having security confiscate my small hex wrench set (part of my bike kit) and, on a different occasion, a tiny screwdriver (part of my camera's filter kit), I would recommend getting a CHEAP screwdriver.

salmonellaeater··on An Empirical Evaluation of TCP Performance in Online Games
HN ought to parse "http://www.iis.sinica.edu.tw" as "sinica.edu.tw" rather than "edu.tw".
salmonellaeater··on On the Typography of Flight-Deck Documentation (1992) [pdf]
It's a similar reason for the advantage of lowercase over uppercase: the overall shape of the letter is easier to recognize if it's "bumpier," as in serif fonts.
salmonellaeater··on Why the Tech Industry Needs to Deal With Its Ageism Problem
While I agree with the sentiment, the statistics quoted don't support the author's thesis.

> The average age of a successful entrepreneur in high-growth industries such as computers, health care, and aerospace is 40.

> Twice as many successful entrepreneurs are over 50 as under 25.

> The highest rate of entrepreneurship in America has shifted to the 55–64 age group, with people over 55 almost twice as likely to found successful companies than those between 20 and 34 – in fact, the 20-34 age bracket has the lowest rate of entrepreneurial activity.

If ten times as many people over 50 than people under 25 try to start businesses, then the statistics strongly favor the under-25's. These statistics don't mean anything unless you know the number of people in each age group who start businesses.

> 75% have more than six years of industry experience and 50% have more than 10 years when they create their startup.

What does this even mean? Years worked is not the same as having useful experience.

The article would be much stronger if it showed evidence that older workers were more productive, more likely to succeed at starting a business, or were in some other way undervalued by the current job market.

salmonellaeater··on FBI struggles to seize 600,000 Bitcoins from alleged Silk Road founder
> An FBI spokesperson said to Hill that the “$80m worth”

The FBI gave the number denominated in USD, so the journalist had to estimate the true number of BTC. The 600k is almost certainly calculated from that $80m number.

salmonellaeater··on What is the theoretical limit of information density?
It's worse than that. There are a finite number of states of a message, and you can only reference that number of states of the outside system. If you think of the message as a pointer containing n bits, then you can only reference the first 2^n positions in the dictionary.

If the outside system is a probability distribution over output messages (a more general case of the dictionary you described) then the problem is synonymous with compression.

salmonellaeater··on Captain American in a Turban
The reason some of the photos look unnatural is because the photographer used fill flash [1]. Her crime was not putting a colored filter over the flash to match its color to the ambient light, so the bluish light from the flash stands out as wrong.

[1] http://strobist.blogspot.tw/2006/03/lighting-101-balancing-f...

← PreviousPage 4 of 6Next →