> One of the advantages of UTF-16 is knowing right away it's UTF-16 as opposed to deciding if it's UTF-8/ASCII/other encoding. Sure, for transmission it's a waste of space (still, text for today's computer capabilities is a non issue even if using UTF-32)
First of all, if you don't know the encoding, then you don't know the encoding, and you will need to figure out if it's UTF-8, UTF-16, ISO-8859-1, etc. If you happen to know that it's UTF-16, you still need to figure out if it's UTF-16BE or LE.
> "It's not fixed width" But for most text, it is.
This is a dangerous way of thinking. One of the big problems with UTF-16 is that for most text, it is fixed width; so many people make that assumption, and you never notice the problem until someone tries to use an obscure script or an emoji character. This means that bugs can easily be lurking under the surface; while with UTF-8, anything besides straight ASCII will break if you assume fixed width, making it much more obvious.
> Sure, you can do UTF-32 and it may not be a bad idea (today)
UTF-32 isn't really meaningfully fixed width either. Sure, each code point is represented in a fixed number of bytes, but code points are not necessarily the interesting unit you want to index by. A glyph could be composed of several code points. Most of the time, you actually want to deal with text in longer units such as words or tokens, which are going to be variable width anyhow. The actual width of individual code points is only really of interest to low-level text processing libraries, not most applications.
"First of all, if you don't know the encoding, then you don't know the encoding"
True. But as you said, you have to know if it's BE or LE on UTF16. And there are ways to determine that automatically. (or it's on the same platform so it doesn't matter). With "ASCII compatible" codes, you can't.
I guess the main issue to me is that UTF-16 is not "ASCII compatible" so you know it's a different beast altogether.
And don't worry, I'm not assuming UTF-16 is fixed width. One should use the libraries and not try to solve this 'manually'.
About UTF-32 think: CPU registers and operations. Working with bytes is inefficient (even with the benefit of smaller size).
> Statistics show that most software developers are out of the field by age 40.
Really? What statistics? And where do these statistics show the developers going? People don't just dry up and blow away at 40. They move to managerial positions. They start their own companies. They stay in jobs longer, so while you might see fewer of them being hired, that doesn't mean they're not still working in the field.
And there are simply not many older developers to begin with. How much of this effect is the fact that there are simply many more qualified developers under 35 than there are over 40, as the field wasn't as big and not as many people were getting training 20 years ago?
This article is high on anecdotal doom and gloom and low on actual evidence. "Statistics show" is not a very convincing argument.
I think you touched on the core of the issue here. How much has the market grown on the past 20 years? Quite a lot actually. As the market grows, more people gravitate towards it as a career choice, and they are mostly young people looking for career choices. 20 years from now, those people will be seniors in a highly competitive market where solutions grow in technical complexity day after day.
The only thing certain is that things will have changed by them from what they are currently.
No variables, no multiple classes. Now, I'm not saying that this is better than variables, or that variables won't make it easier to manage. But you don't have to repeat attributes or use many classes for each element currently in CSS, you just have to organize it a bit differently than you might expect at first.
And that's a lousy way to go, considering how spread out your properties can get. It's okay in small amounts when applied to a group of adjacent elements, but going any further makes no sense (unless you like using Ctrl+F a lot).
I need an editor/IDE with a dropdown that displays the properties of any element in one panel, perhaps showing me where the actual stylesheet changes are made in a side column. Like editing styles in Firebug, only smarter. Anyone got a recommendation?
Character delay is absolutely a real issue over mobile network connections. I will occasionally try to get some work done on my laptop during a car or bus ride, tethered to my phone.
And while tmux/screen do keep the state server side, so you can reconnect, you have to do it manually. I usually connect to the wired network at my desk (to get lower latencies using Synergy), but may unplug my laptop to go to a meeting or debug an issue at another desk. Having to re-establish all of my connections when I do that is a pain.
And finally, tmux/screen do nothing about the UTF-8 incompatibility situation, and in fact may make it worse by adding an extra layer in between in which encoding information can be lost (I haven't played with this very extensively, but I have been frustrated in the past trying to get UTF-8 text entry to work over SSH).
Character delay is an issue over a sufficiently long distance network cable, forget mobile connections. Good thing I have tramp, and it doesn't ship it's own crypto....
Well, I just didn't want to go too far into making the debate about the OP personally rather than just evaluating the merits of his claim. It's possible that Google killed his dog and that his anti-Google argument has merit.
But besides that, LWN has a feature to allow sharing subscriber-only links for a reason. They want people to share those links and discuss them occasionally, as long as someone doesn't create an aggregator that posts all subscriber-only links. And all of the articles become public after a week, anyhow.
They are not trying to prevent people from discussing or reading articles; they are merely trying to encourage people to subscribe if they want to be able to see all of the subscriber-only content as soon as it's posted.
I presume he actually kept a copy or draft himself (so the "only copy" part was not strictly true), which he published seven years later in his autobiography, figuring that was long enough after not to be taken for opportunism but mere historical interest.
Great idea for a product, but the privacy policy here is seriously inadequate.
In order to use this, I would have to give a lot of personally identifiable information. The privacy policy basically gives Leaky the right to do whatever they want with my personally identifiable information, such as providing an unlimited set of third parties for "internal business purposes," an unlimited number of present or future affiliates which they will "seek to require" to adhere to the privacy policy, to anyone who may purchase the company or its assets, or for a fairly extensive list of fraud and law enforcement reasons, to disclose it to an unspecified set of entities.
On top of that, there is no limitation on how long Leaky will hold such personally identifiable information, nor any information about how to get Leaky to remove it. So basically, once I enter information into Leaky, they may hold onto it forever and sell it to whoever they like for quite a wide variety of reasons, with no recourse for the user to protect their information.
I don't think that I would ever want to put such personal information in a service with such a weak privacy policy. And given the new model, in which this is just an implementation of the complex pricing structures used by auto insurance agencies, I wonder why this couldn't be made into an application that runs locally and never exposes my personal information to anyone else. Yes, you would probably have to go with a paid application, rather than an ad-supported web site, with all of the problems that entails, but for a product like this with such important privacy concerns and such little value (to the customer) to putting the data online or sharing it with people, it really seems like it would make more sense as an offline application, not a web app.
However, what exactly is the sensitive data here? Your first name, your zip code, your car make/model/year, and your profession.
99.99% of people would happily give you that data for a free pen, just for the asking.
Also, while the Leaky guys (being in YC) are likely to actually be better than their privacy policy, what happens when you click over to the actual insurance company and give them your full details (SSN, credit info, vehicle accident history, license info, etc.). I'm more concerned with how AIG will use my data than how a small startup will, and the insurance company is the one collecting vastly more -- and their privacy policy allows them to use it for all sorts of things which are opposed to the user.
So, I haven't actually tried the service (due to my privacy concerns, as well as just time). I was assuming they would use more information than that (credit info, vehicle accident history, etc) if they were going to do a good job of modeling how car insurers do pricing.
The amount of information you mention is not as troubling (though you missed your email address), but I would still like to know that they will delete that information after they are done with it, not retain it forever while maintaining the right to give it to pretty much anyone they want as long as they are "affiliated" with Leaky.
And yes, a YC startup may be more trustworthy than an insurance company, but on the other hand, the insurance company needs to collect that information, and is much more heavily regulated, while Leaky, as a private company under not much regulation, can do pretty much anything it wants.
I haven't read your comments (too long, no time) but I think you should fix them. They seem like they might be critiques of a product you haven't even tried.
I tried it and found that it underestimated my insurance cost by exactly $500 with both Geico and Allstate but interestingly the spread between the two was correct (about $100)
Auto insurers will definitely give you a discount for certain fields of employment. "Engineer" drops it $100 or so every time I've tried. My business cards sometimes say "Software Engineer", but I dropped out of a math program, so it's pretty bogus.
Haha, yes but... Consultant. I hope somebody in Hartford, CT (or equivalent) at least finds the notion of giving a discount for having a title which is almost entirely devoid of meaning.
Leaky guys -- I'll try to ring Geico and tell them I'm a consultant so I'd like my $150 discount. Will let you know how it goes.
sakai -- We take our accuracy very seriously; email us at [email protected] and we'll look into it.
If you're in California (the only state that we officially support), we're within 1% in about 50% of cases and within 3% in about 80% of cases - and we're always getting better. However, there are still occasional outliers that we are always trying to identify and fix.
If you're outside California, well - the prices will be different. However, the magnitude changes (eg for switching cars) will be similar to what you would see in your state.
As for the occupation discounts - those come straight from the insurers' filed documents.
> Nowadays, it is illegal in the U.S. to give IQ tests to job applicants
Citation needed.
It is legal (at least based on the guidelines provided by the US Department of Labor: http://www.onetcenter.org/dl_files/empTestAsse.pdf) to give people all kinds of tests as long as (a) they do not select against certain protected traditionally disadvantaged classes more than it does for the average applicant or (b) if it does, there is a demonstrable link between the test and ability to perform job functions (so, for instance, a test of strength in a job that involves lots of lifting may select against women, but if strength is highly correlated with job performance, that is OK).
IQ tests are actually a fairly poor test for many jobs, and they can test many things besides just intelligence (for instance, knowledge of the English language or cultural trivia). For a blue-collar job, an IQ test may disadvantage people from a different cultural background or who don't speak English as a native language, without actually being particularly relevant to job performance. One of the cases that has come up many times is in fire departments, in which written tests are given and used to screen out applicants who score below a certain level, when much of the knowledge tested is not actually all that important to the job.
These days, for a programming job, you would give a more specific test of programming ability. At the time, given that there probably weren't a lot of people with training or experience programming who could be hired, an IQ test was probably a reasonably good generic stand in for the aptitude to learn and think critically.
We still give IQ tests, they are just IQ tests that have a cultural background required. When we ask an applicant to solve a problem we do not expect them to have seen before, we are testing their intelligence, even if we then proceed to expect them to turn this into code.
Even then, plenty of the Google interview questions aren't really any different from the test she received. IQ tests have sections that it doesn't sound like that test included, such as analogy tests or word memorization. The test she received was tailored to the logical, computational thinking required for programming, not generically designed to measure "intelligence".
The major difference between then and now is that they were willing to train people with no prior experience.
Even in this story, I note they were screening a very limited pool of applicants. I wonder what would happen if we went into high schools and conducted universal tests of potential: could we find (and possibly hire) large numbers of people with enormous aptitude who had never considered programming?
"Good intent or absence of discriminatory intent does not redeem employment procedures or testing mechanisms that operate as 'built-in headwinds' for minority groups" - Warren Burger, Chief Justice.
Yes, and in that ruling, as in my comment, the crucial part is that the IQ test is not necessary for the business. "The touchstone is business necessity. If an employment practice which operates to exclude Negroes cannot be shown to be related to job performance, the practice is prohibited."
IQ tests in general are not banned. IQ tests are not allowed when they disproportionately impact certain protected groups, and are not shown to be necessary for screening applicants. Sure, that means that IQ tests will have a lot of problems these days, because they are not very specific for particular business needs.
But if today, you were to invent an entirely new field of work, and needed to hire people who had an aptitude for grasping something new and solving abstract problems without being able to depend in prior training or experience, then an IQ test might very well be relevant.
The key is whether there is a business necessity for this test or other requirement. It sounds like general intelligence test was necessary in hiring people in a completely new, and intellectually demanding field, where there were no other educational, training, or experience qualifications you could rely on. It is not necessary when hiring firefighters or workers at a power plant. From the article you link to: "It was found that White people who had been working at the firm for some time, but met neither of the requirements, performed their jobs as well as those that did meet the requirements."
It really depends on the job doesn't it? Google doesn't get sued for making logic puzzles part of the hiring process just as Hooters is free to exclusively hire buxom young women for their floor staff. Like many of the complicated regulations in the US there is no clear line and many loopholes to be exploited.
This story about a guy who was denied employment as a police offer because he scored too high on an IQ test is a sufficient counterexample to disprove the assertion that IQ tests are strictly illegal (the decision was upheld in court): http://abcnews.go.com/US/story?id=95836#.T3V6K3lRyxN
I've edited it to read "effectively illegal", but the essential point remains. In practice, any company using a straight-up IQ test in job interviews is practically begging to be sued.
It is legal (at least based on the guidelines provided by the US Department of Labor: http://www.onetcenter.org/dl_files/empTestAsse.pdf) to give people all kinds of tests as long as (a) they do not select against certain protected traditionally disadvantaged classes more than it does for the average applicant or (b) if it does, there is a demonstrable link between the test and ability to perform job functions (so, for instance, a test of strength in a job that involves lots of lifting may select against women, but if strength is highly correlated with job performance, that is OK).
Imagine trying to apply these hilariously vague "guidelines" as a hiring manager. Then, imagine being sued for applying them wrong. Then, you'll realize why almost no one uses IQ tests for employment screening.
Oddly enough, the main exception I can think of is the NFL:
First of all, you have to ask yourself how important is this. Is it important for you to get a lot of reputation? Or is it just important to help people out? There are a lot of things that you can do that help people out that don't gain you a lot of reputation.
For example, those badly worded and incomplete questions. You will almost never get much rep from answering those; but you can help that person out a lot, by providing an answer that describes how to formulate their question better, asks them for the extra information they need, shows them how to provide a minimal example of the problem at hand. Then as they edit their question to post this extra information, you can try and use that to flesh out your answer, until you have answered their question for them. By the end, you will probably only get one or two upvotes and an accepted answer (at best, sometimes they just give up), but you will have helped explain to them how better to ask questions, and how better to solve the problem on their own.
Beyond that, sometimes you just have to answer questions quickly and get lucky. I spent some time "playing" StackOverflow like an MMO, and learned some tricks for getting rep (while still doing a good job of answering peoples questions). To do well on questions on popular tags, find ones with no answers yet (or only one or two answers) that you can answer. If it can be answered in a sentence, do so. Then edit your answer. Flesh it out. Provide links to the documentation. Write a sample program demonstrating how it works.
That way, you got in quick with the one-sentence answer. If that is what they need, they may accept it and you're done. You may also get early upvotes. The edit, to have a more in-depth answer, will make your answer a lot more useful than all of the other quick and dirty answers. So if they needed something more in-depth, they will appreciate the extra information. People always appreciate links to docs, and short and sweet example code. Some people appreciate an answer that expands a little bit, describes more generally how something works to put the answer in context and allow them to figure that out in general.
Here's an example of that strategy at work: http://stackoverflow.com/posts/793867/revisions . In fact, I believe there was another edit that's not shown (if you do an edit within a certain window, it gets merged into the last edit). I think that I just started out with the first two lines; the most basic answer. I then expanded a little bit, to cover some other possible cases (as I couldn't tell precisely what they needed from the question). Then I Googled for the documentation, and provided links to it, to help them find other information they might want to know.
Another strategy is to find a hard problem, and spend the time and effort running it to ground. These problems don't come up often, but when they do, they can be really good ones to solve. They can substantially help people out. They are usually much more interesting, and you can learn new things while trying to solve them.
For example, here's one where I didn't know the answer when I first saw the question: http://stackoverflow.com/questions/1990464/efficiency-of-pur... . In fact, I saw the question, thought it was interesting, and wanted to find out the answer; several people had "answered" it, but their logic was so bad that I couldn't let those answers stand (there are a number of answers on that question that have since been deleted by their authors, which you can only see if you have a high reputation, but the answer that's at -15 should give you an idea of the quality of many of the earlier answers). Luckily, I had a bood that I knew would probably answer the question, or give me the starting point. I pulled out that book, found the appropriate references, Googled them, found the papers in question, skimmed through those, and summarized it all. It took me a half an hour to an hour to do so, and then I read the papers in more detail and posted a substantial edit that went into a lot more detail a few hours later. That took some serious effort, but it has been one of my highest voted answers.
So, in summary, doing a good job on the writeup can be important. Provide context, links, references, sample code. Quick throwaway answers will sometimes answer the question, but people like a little extra information (as long as it isn't tl;dr like this comment). If you can't solve easy problems quickly enough, either solve more obscure problems, solve harder problems, or just do a better job answering those easy problems. Don't worry too much about your reputation; it can be a serious time sink to get a lot of reputation. I have spent time climbing the reputation ladder as fast as possible, and I spent way too much time doing so.
First of all, if you don't know the encoding, then you don't know the encoding, and you will need to figure out if it's UTF-8, UTF-16, ISO-8859-1, etc. If you happen to know that it's UTF-16, you still need to figure out if it's UTF-16BE or LE.
> "It's not fixed width" But for most text, it is.
This is a dangerous way of thinking. One of the big problems with UTF-16 is that for most text, it is fixed width; so many people make that assumption, and you never notice the problem until someone tries to use an obscure script or an emoji character. This means that bugs can easily be lurking under the surface; while with UTF-8, anything besides straight ASCII will break if you assume fixed width, making it much more obvious.
> Sure, you can do UTF-32 and it may not be a bad idea (today)
UTF-32 isn't really meaningfully fixed width either. Sure, each code point is represented in a fixed number of bytes, but code points are not necessarily the interesting unit you want to index by. A glyph could be composed of several code points. Most of the time, you actually want to deal with text in longer units such as words or tokens, which are going to be variable width anyhow. The actual width of individual code points is only really of interest to low-level text processing libraries, not most applications.