Shredding hard drives is not pointless, and neither are the other steps taken.
It may seem redundant to first wipe the drives, then shred them, but if you think about it both steps are necessary. Wiping them is the best method to ensure that no data is recoverable, but remember that drives are pulled from service when they're failing. Can you trust a failing drive to successfully zero itself? Even verifying that you can successfully read all of the zeros from the disk after writing them doesn't prove that all of the data is gone, because a failing sector may have been "relocated" to a spare sector -- but that doesn't mean the failing sector is necessarily completely unreadable. And, of course, many failing drives won't operate well enough for the wipe/verify process to complete successfully.
So, given that after the wipe operation data may still remain, it's necessary to physically destroy the data. So why not just shred the drives, rather than bothering with the wipe? Because a small piece of a hard drive platter may contain a significant amount of data. Actually recovering that data is probably quite difficult, but it is possible.
I'm somewhat surprised they don't go a step further and use an acid bath, but I suppose they figure after wiping has gotten rid of most of the data and shredding has made whatever's left very hard to retrieve, the drive fragments are shipped to a trustworthy recycling operation and any remaining data will quickly be destroyed by the recycler. Of course, the recycler could do the shredding, but since it has to be done anyway Google may as well do it up front.
Plus some customers undoubtedly have security policies that require wiping and shredding.
If you want to argue that the shredding is unnecessary, what you should point to is the fact that all of the customer data on the drives is already encrypted, with keys that are not on the drives. That arguably makes all of the destruction steps unnecessary.
(Full disclosure: I work for Google but have nothing whatsoever to do with data centers or hard drive lifecycle management, and no information about how this stuff is done beyond what I saw in the video.)
I like having the geo-location metadata embedded in my photos. Services like Facebook and Twitter really should sanitize photos by removing most, if not all, of the EXIF data by default, though.
That is complete rubbish. It just has to be public. Now, fact is, the USPTO will often not check anything but prior patents (if that), and just trust the patent submitter's list of other prior art (no really), but in principle (and what matters in court) prior art that is anything publically released.
The problem is the climate and the attitude. Too many desperate job seekers, too few jobs. Employers are simply too picky and arbitrary, putting on this big act as if even a rather ordinary job requires such specialized and honed skills that only 0.1% of job seekers can handle.
In the case of Google and other companies like it, I don't think it's a matter of a "big act"... smart technology companies understand that their employees are their single most significant success factor. If they can get great talent and empower it, they'll do well. So it's not a matter of "this job requires these skills", it's a matter of "we can't know what skills matter because we can't know where we should go". So, the solution is to hire the best people available. For companies with the ability to attract top talent, there's really no reason for them to accept less.
In Google's case, I think they have actually lowered their standards a little over the years. I think that because my perception is that a few years ago they wouldn't have hired me, and now they did. I think it also simply because they've increased their hiring rate so much that it seems unlikely that they can really be as selective as they once were. But, still, as long as they can set the bar relatively high, and yet still succeed in hiring enough people to fill their openings, why shouldn't they?
I suspect Google weeds out a lot of good people. There are more good people out there than employers will admit.
Google absolutely does pass over a lot of good people, and the company is fully aware of that, and sees it as a problem to be solved. Unfortunately, no one really knows how to solve it without also increasing the number of less-qualified and unqualified people that get hired. Hiring is a really hard problem.
We all know industry complaints about not enough STEM workers is a self-serving lie, used as a reason for increasing H1Bs among others. I believe Google is above that sort of thing.
Google's approach is to create lots of local offices around the world so they can better hire the available talent, rather than trying to find ways to bring them to the US. But Google's overseas hiring isn't about hiring cheap talent; it's about trying to find the best people they can, knowing that there are lots of capable engineers outside of the US, and about having engineers that understand local culture around the world, because Google's customer base is global.
If there really was a healthy demand for STEM workers, we'd see pay going up, and last I heard, we aren't seeing that.
Actually, we are. There have been some/. articles about rising salaries.
The hiring decision could be made in 15 minutes, or less, with good information. But there isn't good information, or is there? Seems that most of the time needed in Google's interview process is for gathering information, not evaluating it. Another poster complained that most of the information the interviewer gets is biased, and therefore of little value.
Actually, that was me saying that the interviewee can provide biased or inaccurate information (or maybe someone else said it, too). The sucky thing about that situation is that most candidates do provide good information. But if it were relied upon, getting hired would be a simple matter of typing up an outstanding resume.
I'm sorry, but employers haven't deserved better. They play games too. [many real and bothersome problems elided].
Yep. I've seen all that and more. The one good thing is that the employers who play those crappy games the hardest are also the ones you really don't want to work for anyway. I'm more than happy to let them exclude me. But, then, it's easy for me to feel that way, since I've been fully employed by good companies for most of my 20-year career. It's tougher for people who
Thanks for the compliment (calling me a bright kid -- I have enough gray hair that I haven't been called a kid for quite some time, and I'm far less bright than I once was, though I do think that experience makes up the difference).
As for the other part, I'm going to assume you're not trolling, and spell it out.
First, "very, very few bad hires" does not equate to "no bad hires". Assuming making bad hires were the only possible flaw, "very, very few bad hires" would still equate to "not perfect"... because perfection would require that there be NO bad hires. As long as there are any bad hires, there is a need to repair the process.
However, as I explained in my previous post, there are other relevant criteria to consider. One is cost. Google's process is time-consuming and therefore expensive. Another is burden on the candidates. A five-hour interview marathon is exhausting and consumes a whole day that could be used for other things. Another is whether or not the process rejects good candidates. Google's process does, a fact that is very well understood.
So, the process is imperfect in many ways. It has false positives (though relatively few). It has false negatives (likely a substantial number). It's also not clear how even to measure the error rate on either side. Ideally, calibrated interview scoring should provide a strong predictor of job performance (as measured by job performance reviews). Does it? I don't know; though I'm sure that someone at Google is tracking the statistics. Until interview scoring is strongly correlated with job performance, the interview process can be improved. And, of course, it's expensive for Google and time-consuming and intimidating for candidates. I have a couple of friends I think would do a great job at Google, but they're frightened off by the duration and intensity of the interviews.
There is no contradiction between the statement "the process is imperfect" and the statement "there are very few bad hires".
I wonder how many female human beings do you happen to have in your special, white, boy-scout Stanford' grades.....
I think women comprise the majority of the Stanford studentbody, like they do at most universities. I didn't attend Stanford (or any other top-tier university), so perhaps someone who knows will speak up.
If you're asking about Google, I also don't know. My own team is about 10% female, which is a higher ratio than anywhere else I've worked.
I'd really like to have a low-battery-impact location stream for my phone, but my iPhone 4 apparently has this logging feature turned off, because the CELL_LOCATION table on my device is empty.
I think your line of thinking does a disservice to blacks and other minorities, because assuming the standard for hiring is an accurate reflection of job performance, lowering the bar for hiring will reinforce the perception that minorities are inherently less capable of performing well. If you systematically lower the requirements for any identifiable group, you will cause that group's membership in the company to perform more poorly, on average, than other groups. I think that's exactly the opposite of the effect you're looking for.
Note that this is my personal opinion, and not Google's.
I've seen a lot of different approaches but, yeah, Google's is pretty typical of competent, high-tech companies.
Where you get really different processes is when you start talking to companies who aren't really tech/software companies, but have their own engineering teams tucked over in a corner. Their approaches are all over the map, and may vary significantly depending on who happens to be in the office that day.
So you are saying that during this 5h long test, i have to solve the global peace without even....google-ing?
You're not graded on your ability to perfectly remember details of library APIs, language syntax, etc. Or to remember every algorithm you ever learned about. You are graded on your ability to solve problems and write tight, well-structured code, so googling really wouldn't be helpful.
And just one more comment, you are saying two things that are kind of illogic:
1.Is the system perfect? Clearly not, and Google recognizes that and is constantly looking for ways to improve it.
2.my experience as a Google employee is that there are very, very few bad hires
So, please, enlighten me, which one is the correct sentence?
There is no logical inconsistency.
The interview process takes too long, costs too much, is too burdensome on candidates and tends to reject too many good people. But it doesn't often hire weak candidates.
Where is the racial modifier? Does Google not value diversity in hiring? A strict "meritocracy" is just a codeword for "whites only".
First, I have to disagree -- a truly strict meritocracy is race, gender and culture-blind. However, it has to be acknowledged that when you have predominantly white men evaluating candidates for merit, the result is a white male bias. The bias may be purely unintentional, but that doesn't mean it doesn't exist. Google recognizes the problem of bias, whether intentional or not, and does try to address it -- but without lowering the hiring standards. Google has done internal research which shows that diverse teams create better, more innovative solutions, so the company has a strong motivation to increase diversity.
The minority modifier in the process comes in two ways, from what I can see.
First, interviewers are put through diversity training, and taught to err on the positive side when dealing with people of significantly different backgrounds than themselves, as well as to avoid saying or doing things that might make minority candidates uncomfortable.
Second, diversity is an important goal of the second-level hiring committee, and suspicions of exclusion based on cultural/racial/gender differences are a major motivator for overridden no-hire decisions.
There's actually a lot more that Google is doing around this issue, including a lot of detailed analysis to identify biases so it can find ways to work around them, but I'm not sure how much of it I can share. In general, that's a problem for Googlers... there's so much openness and transparency within the company that it's hard to know what's confidential and what's not.
Thanks for clarifying your own closed-minded, data-averse and moronic attitude, as well as your complete inability to state a cogent and well-supported opinion. I can now feel confident that there is nothing of value in your posts.
"... lean towards rejection", yes indeed! Second, I get just a little tired of claims of exceptionalism. This says, not in words but in this action, that Google thinks themselves very special indeed, whatever anyone may say about their efforts to eschew arrogance.
I don't think it says that Google "thinks themselves very special" it means that Google wants to hire excellent candidates -- significantly above the average -- and doesn't want to have to deal with figuring out how to get rid of employees who turn out not to be able to do the job to the level Google wants. Google may eventually find that this approach results in not being able to hire enough people, at which point a different approach will have to be taken.
Third, it's very expensive to conduct such long interviews.
Indeed it is. Especially when you consider that many such expensive interviews result in a "no-hire" decision. I think the actual hiring rate is confidential, but you can multiply those five engineer-hours by a substantial number to get the engineer-hours-per-hire.
But, the cost of hiring a person who doesn't work out is orders of magnitude higher. This is perhaps more true at Google than other companies. Google has a tremendous amount of homegrown technology (somewhat due to NIH syndrome, but much of it because there simply isn't anything on the market that will do the job -- or at least wasn't when the technology was created), so the learning curve for a new engineer at Google is both steep and long. That means it takes months to really determine that a new engineer isn't going to work out, plus all of the time and effort required to push someone out.
As much as companies try to externalize costs, this shows that when it come to hiring decisions, Google at least has utterly failed at that. I have suggested before that a college degree (of some level from bachelors to PhD, depending on the job), which should be checkable in about 2 seconds, ought to be enough.
That's a much worse way to evaluate candidates. Not only do plenty of people who can't do the job get degrees, but there are plenty of people who can do the job who don't get a degree. Even supposing you don't mind excluding the talented non-degreed individuals, how do you choose among all the people with degrees? Just hire anyone who walks through the door with a degree? Decide which schools do a good job of filtering out the chaff and only hire graduates of those schools? Base your decisions on GPA?
Hiring anyone with a degree is clearly a non-starter. Hiring only from top schools not only misses a lot of capable people who went to lower-tier schools, but also tends to create a monoculture. Google actually struggles with that, to some degree -- too many people from Stanford, MIT, CMU, etc., and so the company tries hard to broaden the pool. Basing hiring decisions on GPA just means that you get the people who play the game well in school; many of the brightest and most talented people don't do well in school.
what's wrong with at least considering something like the results from appropriate subject tests of the GRE?
Because those generally test knowledge, rather than problem-solving ability. Tests of problem-solving ability could be constructed, of course, but a written test gives you very little information about how the candidate solved the problem. Having them work it out, out-loud, in front of a smart interviewer provides much deeper insight into their abilities.
The on-the-spot challenge does, of course, exclude some people who don't do well in that sort of a pressure situation (though Google interviewers try hard to avoid applying pressure, there's no getting around the fact that candidates who want the job feel pressured). That's unfortunate but I, at least, don't know what would work better. The ability to perform under pressure also isn't completely unrelated to the ability to do the j
but on the issue of markets, this person is clearly a grade AAA certified moron
Right, because he disagrees with you.
My opinion is that the situation with respect to markets and regulation isn't so clear-cut. I see huge problems that result from having an inconsistently-regulated market (the primary cause of the recent crash was the assumption of investors that they could rely on regulation to protect them, when in fact the protections had been removed), and there's clearly a problem with markets in which information about is unreliable. In fact, arguably, it was unreliability of information, caused by regulation which wasn't doing the job that everyone expected it was, that caused the sub-prime mortgage bubble and collapse. Had the regulation not existed at all, investors would have been more cautious and would not have purchased mortgage-backed securities, limiting the banks' ability to fund ever-more-foolish loans. Had the regulation been doing what it was supposed to (note that I think there's room for serious debate as to whether that's even possible, but let's suppose it is), then the mortgage-backed securities would have been banned, or at least downgraded, with the same effect.
I think regulation which serves only to ensure the completeness and accuracy of information provided, with no real limitation on action, would provide the most efficient market, and that the more regulation steps in to actually control the behavior of financial markets, the less effective and efficient they will be. At the extreme, the degree of massive regulation you seem to favor will stifle economic progress and centralize economic power to a frightening degree.
But, then, I'm sure I'm a grade-AAA certified moron, too, because I fail to agree with you.
if you never agree to that, you're just another among millions of morons, who speak freely and judgmentally on subject matters they don't even understand
People often evaluate the intelligence of others based on the degree to which those others agree with their own opinions. Rarely, however, do I see someone who is willing to state it so plainly.
Hiring is hard, trust me, I've done it for years. However, I claim that silly little tests and so on are just that: silly. I have turned down senior management positions at companies that think they have some sort of Golden Test that candidates need to pass.
Google doesn't have a "Golden Test". I'll describe Google's process below.
There are really only two things to evaluate: (1) Is the candidate smart? and (2) Can the candidate be effective in the position?
Yes, those are the things you need to figure out, but no, neither of them is very easy to evaluate. Especially not when you set your standards as high as Google does. Your approach to evaluating intelligence is especially flawed -- there are lots of people who can talk the talk, but can't perform when given problems to solve.
Here's Google's interview process (the engineering interview process; I don't know about sales, etc.):
The first step is optional, and depends on various things. It's a phone screen. Usually about one hour long, it involves a series of computer science/algorithms questions, and sometimes includes some coding as well, using a shared editor over the web. This screen has nothing to do with the hire/no-hire decision, it's just a filter to verify that it's not a waste of time to bring the candidate on-site.
The on-site interview takes five hours, each hour an interview by another engineer. One of the five "interviews" is lunch, and it has no effect on the hire/no-hire decision; it's mostly an opportunity for the candidates to ask questions and to talk about Google culture.
Each of the actual interviewers gets to ask whatever questions they like (though with some guidance from HR about what kinds of questions need to be avoided). However, there are some recommendations: Questions should be focused on technical topics that evaluate candidates' problem-solving and coding skills, and at least one coding problem must be included. Google interviewers pay no attention to what you have done in the past, except maybe to break the ice and perhaps as a source of technical topics to discuss. Mostly, they ask serious CS questions, requiring you to design (and implement) algorithms to solve problems, and to evaluate the real and asymptotic efficiency of your solutions, and to discuss issues related to scaling your solutions to Google scale (meaning really, really huge).
Afterwards, each interviewer writes up their thoughts, complete with the code you wrote. They do comment a bit on cultural fit, but unless you're really just impossible to work with (e.g. extremely arrogant) that's unlikely to be a problem. Mostly they discuss your problem-solving approach and ability and your coding ability. Each interviewer also rates you on a scale from 0 to 4, and gives their hire/no-hire recommendation. Google's process minimizes and discourages communication between the interviewers, because they don't want one interviewer with an excessively negative or positive opinion to affect the other interviewers' opinions.
After all reviewers have submitted their feedback, the data is compiled and delivered to a hiring committee (again a group of engineers, perhaps with a manager or two, but mostly engineers -- and Google managers are all engineers, too). Based on that information they have to come to a consensus decision to hire, reject or request more interviews (the latter is rare). Candidates who are rejected are not allowed to interview again for six months.
In rare cases, the decision of this hiring committee may be overridden by another, higher-level committee.
At all levels, the direction given to interviewers and committee members is to lean towards rejection. False negatives are perceived as less painful to the company than false positives, so the process is negatively biased.
Is the system perfect? Clearly not, and Google recognizes that and is constantly looking for ways to improve it. I'm not sure how muc
I seem to remember hearing that there were a couple programs that could sniff out a TC volume with a pretty good amount of certainty
Cite?
And who fills their unused hard drive space with random data? If you make that claim, it pretty much sets you up as someone who would likely use TrueCrypt anyway.
I thought we were talking about whether or not Dropbox could identify TrueCrypt volumes in order to refuse to store them, not whether or not someone with files of uniformly-distributed bits is "suspicious".
The application gives me an error about being unable to find the file, but if I identify and locate it manually and open it up with sqlite manager I see that I have the database, and the table, but there are 0 rows in it.
FYI, TrueCrypt files don't have any distinctive headers, by design. Given a file full of uniformly-distributed random bits, there's no way to tell if it's a TrueCrypt volume or the output of a random number generator.
Can you provide a sample search that shows this problem? The intent of the change was clearly to do the exact opposite of the effect you're describing. It might be useful for someone at Google to look into why their work has backfired in your case.
Two-factor doesn't have to cost much. Your phone can be the second factor. In the case of smartphones, a one-time password generator can be installed as an app, or you can get even more sophisticated and have the web site display a 2D barcode which a phone app photographs, munges into an auth code and sends via the data network. For traditional phones, the site can SMS a one-time password to the phone.
Of course, this assumes that the phone isn't the device accessing FB in the first place.
Slashdot Mirror
Shredding hard drives is not pointless, and neither are the other steps taken.
It may seem redundant to first wipe the drives, then shred them, but if you think about it both steps are necessary. Wiping them is the best method to ensure that no data is recoverable, but remember that drives are pulled from service when they're failing. Can you trust a failing drive to successfully zero itself? Even verifying that you can successfully read all of the zeros from the disk after writing them doesn't prove that all of the data is gone, because a failing sector may have been "relocated" to a spare sector -- but that doesn't mean the failing sector is necessarily completely unreadable. And, of course, many failing drives won't operate well enough for the wipe/verify process to complete successfully.
So, given that after the wipe operation data may still remain, it's necessary to physically destroy the data. So why not just shred the drives, rather than bothering with the wipe? Because a small piece of a hard drive platter may contain a significant amount of data. Actually recovering that data is probably quite difficult, but it is possible.
I'm somewhat surprised they don't go a step further and use an acid bath, but I suppose they figure after wiping has gotten rid of most of the data and shredding has made whatever's left very hard to retrieve, the drive fragments are shipped to a trustworthy recycling operation and any remaining data will quickly be destroyed by the recycler. Of course, the recycler could do the shredding, but since it has to be done anyway Google may as well do it up front.
Plus some customers undoubtedly have security policies that require wiping and shredding.
If you want to argue that the shredding is unnecessary, what you should point to is the fact that all of the customer data on the drives is already encrypted, with keys that are not on the drives. That arguably makes all of the destruction steps unnecessary.
(Full disclosure: I work for Google but have nothing whatsoever to do with data centers or hard drive lifecycle management, and no information about how this stuff is done beyond what I saw in the video.)
I like having the geo-location metadata embedded in my photos. Services like Facebook and Twitter really should sanitize photos by removing most, if not all, of the EXIF data by default, though.
Actually, prior art only refers to prior patents
That is complete rubbish. It just has to be public. Now, fact is, the USPTO will often not check anything but prior patents (if that), and just trust the patent submitter's list of other prior art (no really), but in principle (and what matters in court) prior art that is anything publically released.
Mod parent up. This is correct, GP is flat wrong.
The problem is the climate and the attitude. Too many desperate job seekers, too few jobs. Employers are simply too picky and arbitrary, putting on this big act as if even a rather ordinary job requires such specialized and honed skills that only 0.1% of job seekers can handle.
In the case of Google and other companies like it, I don't think it's a matter of a "big act"... smart technology companies understand that their employees are their single most significant success factor. If they can get great talent and empower it, they'll do well. So it's not a matter of "this job requires these skills", it's a matter of "we can't know what skills matter because we can't know where we should go". So, the solution is to hire the best people available. For companies with the ability to attract top talent, there's really no reason for them to accept less.
In Google's case, I think they have actually lowered their standards a little over the years. I think that because my perception is that a few years ago they wouldn't have hired me, and now they did. I think it also simply because they've increased their hiring rate so much that it seems unlikely that they can really be as selective as they once were. But, still, as long as they can set the bar relatively high, and yet still succeed in hiring enough people to fill their openings, why shouldn't they?
I suspect Google weeds out a lot of good people. There are more good people out there than employers will admit.
Google absolutely does pass over a lot of good people, and the company is fully aware of that, and sees it as a problem to be solved. Unfortunately, no one really knows how to solve it without also increasing the number of less-qualified and unqualified people that get hired. Hiring is a really hard problem.
We all know industry complaints about not enough STEM workers is a self-serving lie, used as a reason for increasing H1Bs among others. I believe Google is above that sort of thing.
Google's approach is to create lots of local offices around the world so they can better hire the available talent, rather than trying to find ways to bring them to the US. But Google's overseas hiring isn't about hiring cheap talent; it's about trying to find the best people they can, knowing that there are lots of capable engineers outside of the US, and about having engineers that understand local culture around the world, because Google's customer base is global.
If there really was a healthy demand for STEM workers, we'd see pay going up, and last I heard, we aren't seeing that.
Actually, we are. There have been some /. articles about rising salaries.
The hiring decision could be made in 15 minutes, or less, with good information. But there isn't good information, or is there? Seems that most of the time needed in Google's interview process is for gathering information, not evaluating it. Another poster complained that most of the information the interviewer gets is biased, and therefore of little value.
Actually, that was me saying that the interviewee can provide biased or inaccurate information (or maybe someone else said it, too). The sucky thing about that situation is that most candidates do provide good information. But if it were relied upon, getting hired would be a simple matter of typing up an outstanding resume.
I'm sorry, but employers haven't deserved better. They play games too. [many real and bothersome problems elided].
Yep. I've seen all that and more. The one good thing is that the employers who play those crappy games the hardest are also the ones you really don't want to work for anyway. I'm more than happy to let them exclude me. But, then, it's easy for me to feel that way, since I've been fully employed by good companies for most of my 20-year career. It's tougher for people who
Thanks for the compliment (calling me a bright kid -- I have enough gray hair that I haven't been called a kid for quite some time, and I'm far less bright than I once was, though I do think that experience makes up the difference).
As for the other part, I'm going to assume you're not trolling, and spell it out.
First, "very, very few bad hires" does not equate to "no bad hires". Assuming making bad hires were the only possible flaw, "very, very few bad hires" would still equate to "not perfect"... because perfection would require that there be NO bad hires. As long as there are any bad hires, there is a need to repair the process.
However, as I explained in my previous post, there are other relevant criteria to consider. One is cost. Google's process is time-consuming and therefore expensive. Another is burden on the candidates. A five-hour interview marathon is exhausting and consumes a whole day that could be used for other things. Another is whether or not the process rejects good candidates. Google's process does, a fact that is very well understood.
So, the process is imperfect in many ways. It has false positives (though relatively few). It has false negatives (likely a substantial number). It's also not clear how even to measure the error rate on either side. Ideally, calibrated interview scoring should provide a strong predictor of job performance (as measured by job performance reviews). Does it? I don't know; though I'm sure that someone at Google is tracking the statistics. Until interview scoring is strongly correlated with job performance, the interview process can be improved. And, of course, it's expensive for Google and time-consuming and intimidating for candidates. I have a couple of friends I think would do a great job at Google, but they're frightened off by the duration and intensity of the interviews.
There is no contradiction between the statement "the process is imperfect" and the statement "there are very few bad hires".
I wonder how many female human beings do you happen to have in your special, white, boy-scout Stanford' grades.....
I think women comprise the majority of the Stanford studentbody, like they do at most universities. I didn't attend Stanford (or any other top-tier university), so perhaps someone who knows will speak up.
If you're asking about Google, I also don't know. My own team is about 10% female, which is a higher ratio than anywhere else I've worked.
I'd really like to have a low-battery-impact location stream for my phone, but my iPhone 4 apparently has this logging feature turned off, because the CELL_LOCATION table on my device is empty.
Anyone know how to turn it on?
I think your line of thinking does a disservice to blacks and other minorities, because assuming the standard for hiring is an accurate reflection of job performance, lowering the bar for hiring will reinforce the perception that minorities are inherently less capable of performing well. If you systematically lower the requirements for any identifiable group, you will cause that group's membership in the company to perform more poorly, on average, than other groups. I think that's exactly the opposite of the effect you're looking for.
Note that this is my personal opinion, and not Google's.
IHBT, I shall flagellate myself appropriately.
Only one? Okay, Google Search :-)
I've seen a lot of different approaches but, yeah, Google's is pretty typical of competent, high-tech companies.
Where you get really different processes is when you start talking to companies who aren't really tech/software companies, but have their own engineering teams tucked over in a corner. Their approaches are all over the map, and may vary significantly depending on who happens to be in the office that day.
So you are saying that during this 5h long test, i have to solve the global peace without even....google-ing?
You're not graded on your ability to perfectly remember details of library APIs, language syntax, etc. Or to remember every algorithm you ever learned about. You are graded on your ability to solve problems and write tight, well-structured code, so googling really wouldn't be helpful.
And just one more comment, you are saying two things that are kind of illogic: 1.Is the system perfect? Clearly not, and Google recognizes that and is constantly looking for ways to improve it. 2.my experience as a Google employee is that there are very, very few bad hires So, please, enlighten me, which one is the correct sentence?
There is no logical inconsistency.
The interview process takes too long, costs too much, is too burdensome on candidates and tends to reject too many good people. But it doesn't often hire weak candidates.
Where is the racial modifier? Does Google not value diversity in hiring? A strict "meritocracy" is just a codeword for "whites only".
First, I have to disagree -- a truly strict meritocracy is race, gender and culture-blind. However, it has to be acknowledged that when you have predominantly white men evaluating candidates for merit, the result is a white male bias. The bias may be purely unintentional, but that doesn't mean it doesn't exist. Google recognizes the problem of bias, whether intentional or not, and does try to address it -- but without lowering the hiring standards. Google has done internal research which shows that diverse teams create better, more innovative solutions, so the company has a strong motivation to increase diversity.
The minority modifier in the process comes in two ways, from what I can see.
First, interviewers are put through diversity training, and taught to err on the positive side when dealing with people of significantly different backgrounds than themselves, as well as to avoid saying or doing things that might make minority candidates uncomfortable.
Second, diversity is an important goal of the second-level hiring committee, and suspicions of exclusion based on cultural/racial/gender differences are a major motivator for overridden no-hire decisions.
There's actually a lot more that Google is doing around this issue, including a lot of detailed analysis to identify biases so it can find ways to work around them, but I'm not sure how much of it I can share. In general, that's a problem for Googlers... there's so much openness and transparency within the company that it's hard to know what's confidential and what's not.
Thanks for clarifying your own closed-minded, data-averse and moronic attitude, as well as your complete inability to state a cogent and well-supported opinion. I can now feel confident that there is nothing of value in your posts.
"... lean towards rejection", yes indeed! Second, I get just a little tired of claims of exceptionalism. This says, not in words but in this action, that Google thinks themselves very special indeed, whatever anyone may say about their efforts to eschew arrogance.
I don't think it says that Google "thinks themselves very special" it means that Google wants to hire excellent candidates -- significantly above the average -- and doesn't want to have to deal with figuring out how to get rid of employees who turn out not to be able to do the job to the level Google wants. Google may eventually find that this approach results in not being able to hire enough people, at which point a different approach will have to be taken.
Third, it's very expensive to conduct such long interviews.
Indeed it is. Especially when you consider that many such expensive interviews result in a "no-hire" decision. I think the actual hiring rate is confidential, but you can multiply those five engineer-hours by a substantial number to get the engineer-hours-per-hire.
But, the cost of hiring a person who doesn't work out is orders of magnitude higher. This is perhaps more true at Google than other companies. Google has a tremendous amount of homegrown technology (somewhat due to NIH syndrome, but much of it because there simply isn't anything on the market that will do the job -- or at least wasn't when the technology was created), so the learning curve for a new engineer at Google is both steep and long. That means it takes months to really determine that a new engineer isn't going to work out, plus all of the time and effort required to push someone out.
As much as companies try to externalize costs, this shows that when it come to hiring decisions, Google at least has utterly failed at that. I have suggested before that a college degree (of some level from bachelors to PhD, depending on the job), which should be checkable in about 2 seconds, ought to be enough.
That's a much worse way to evaluate candidates. Not only do plenty of people who can't do the job get degrees, but there are plenty of people who can do the job who don't get a degree. Even supposing you don't mind excluding the talented non-degreed individuals, how do you choose among all the people with degrees? Just hire anyone who walks through the door with a degree? Decide which schools do a good job of filtering out the chaff and only hire graduates of those schools? Base your decisions on GPA?
Hiring anyone with a degree is clearly a non-starter. Hiring only from top schools not only misses a lot of capable people who went to lower-tier schools, but also tends to create a monoculture. Google actually struggles with that, to some degree -- too many people from Stanford, MIT, CMU, etc., and so the company tries hard to broaden the pool. Basing hiring decisions on GPA just means that you get the people who play the game well in school; many of the brightest and most talented people don't do well in school.
what's wrong with at least considering something like the results from appropriate subject tests of the GRE?
Because those generally test knowledge, rather than problem-solving ability. Tests of problem-solving ability could be constructed, of course, but a written test gives you very little information about how the candidate solved the problem. Having them work it out, out-loud, in front of a smart interviewer provides much deeper insight into their abilities.
The on-the-spot challenge does, of course, exclude some people who don't do well in that sort of a pressure situation (though Google interviewers try hard to avoid applying pressure, there's no getting around the fact that candidates who want the job feel pressured). That's unfortunate but I, at least, don't know what would work better. The ability to perform under pressure also isn't completely unrelated to the ability to do the j
but on the issue of markets, this person is clearly a grade AAA certified moron
Right, because he disagrees with you.
My opinion is that the situation with respect to markets and regulation isn't so clear-cut. I see huge problems that result from having an inconsistently-regulated market (the primary cause of the recent crash was the assumption of investors that they could rely on regulation to protect them, when in fact the protections had been removed), and there's clearly a problem with markets in which information about is unreliable. In fact, arguably, it was unreliability of information, caused by regulation which wasn't doing the job that everyone expected it was, that caused the sub-prime mortgage bubble and collapse. Had the regulation not existed at all, investors would have been more cautious and would not have purchased mortgage-backed securities, limiting the banks' ability to fund ever-more-foolish loans. Had the regulation been doing what it was supposed to (note that I think there's room for serious debate as to whether that's even possible, but let's suppose it is), then the mortgage-backed securities would have been banned, or at least downgraded, with the same effect.
I think regulation which serves only to ensure the completeness and accuracy of information provided, with no real limitation on action, would provide the most efficient market, and that the more regulation steps in to actually control the behavior of financial markets, the less effective and efficient they will be. At the extreme, the degree of massive regulation you seem to favor will stifle economic progress and centralize economic power to a frightening degree.
But, then, I'm sure I'm a grade-AAA certified moron, too, because I fail to agree with you.
if you never agree to that, you're just another among millions of morons, who speak freely and judgmentally on subject matters they don't even understand
People often evaluate the intelligence of others based on the degree to which those others agree with their own opinions. Rarely, however, do I see someone who is willing to state it so plainly.
Hiring is hard, trust me, I've done it for years. However, I claim that silly little tests and so on are just that: silly. I have turned down senior management positions at companies that think they have some sort of Golden Test that candidates need to pass.
Google doesn't have a "Golden Test". I'll describe Google's process below.
There are really only two things to evaluate: (1) Is the candidate smart? and (2) Can the candidate be effective in the position?
Yes, those are the things you need to figure out, but no, neither of them is very easy to evaluate. Especially not when you set your standards as high as Google does. Your approach to evaluating intelligence is especially flawed -- there are lots of people who can talk the talk, but can't perform when given problems to solve.
Here's Google's interview process (the engineering interview process; I don't know about sales, etc.):
The first step is optional, and depends on various things. It's a phone screen. Usually about one hour long, it involves a series of computer science/algorithms questions, and sometimes includes some coding as well, using a shared editor over the web. This screen has nothing to do with the hire/no-hire decision, it's just a filter to verify that it's not a waste of time to bring the candidate on-site.
The on-site interview takes five hours, each hour an interview by another engineer. One of the five "interviews" is lunch, and it has no effect on the hire/no-hire decision; it's mostly an opportunity for the candidates to ask questions and to talk about Google culture.
Each of the actual interviewers gets to ask whatever questions they like (though with some guidance from HR about what kinds of questions need to be avoided). However, there are some recommendations: Questions should be focused on technical topics that evaluate candidates' problem-solving and coding skills, and at least one coding problem must be included. Google interviewers pay no attention to what you have done in the past, except maybe to break the ice and perhaps as a source of technical topics to discuss. Mostly, they ask serious CS questions, requiring you to design (and implement) algorithms to solve problems, and to evaluate the real and asymptotic efficiency of your solutions, and to discuss issues related to scaling your solutions to Google scale (meaning really, really huge).
Afterwards, each interviewer writes up their thoughts, complete with the code you wrote. They do comment a bit on cultural fit, but unless you're really just impossible to work with (e.g. extremely arrogant) that's unlikely to be a problem. Mostly they discuss your problem-solving approach and ability and your coding ability. Each interviewer also rates you on a scale from 0 to 4, and gives their hire/no-hire recommendation. Google's process minimizes and discourages communication between the interviewers, because they don't want one interviewer with an excessively negative or positive opinion to affect the other interviewers' opinions.
After all reviewers have submitted their feedback, the data is compiled and delivered to a hiring committee (again a group of engineers, perhaps with a manager or two, but mostly engineers -- and Google managers are all engineers, too). Based on that information they have to come to a consensus decision to hire, reject or request more interviews (the latter is rare). Candidates who are rejected are not allowed to interview again for six months.
In rare cases, the decision of this hiring committee may be overridden by another, higher-level committee.
At all levels, the direction given to interviewers and committee members is to lean towards rejection. False negatives are perceived as less painful to the company than false positives, so the process is negatively biased.
Is the system perfect? Clearly not, and Google recognizes that and is constantly looking for ways to improve it. I'm not sure how muc
I seem to remember hearing that there were a couple programs that could sniff out a TC volume with a pretty good amount of certainty
Cite?
And who fills their unused hard drive space with random data? If you make that claim, it pretty much sets you up as someone who would likely use TrueCrypt anyway.
I thought we were talking about whether or not Dropbox could identify TrueCrypt volumes in order to refuse to store them, not whether or not someone with files of uniformly-distributed bits is "suspicious".
The application gives me an error about being unable to find the file, but if I identify and locate it manually and open it up with sqlite manager I see that I have the database, and the table, but there are 0 rows in it.
Maybe this feature is disabled on the iPhone 4?
The old policy said our files were encrypted with mil-spec encryption, etc etc. Now they're telling us they'll turn our files over if asked.
There is no contradiction between those two statements.
FYI, TrueCrypt files don't have any distinctive headers, by design. Given a file full of uniformly-distributed random bits, there's no way to tell if it's a TrueCrypt volume or the output of a random number generator.
Can you provide a sample search that shows this problem? The intent of the change was clearly to do the exact opposite of the effect you're describing. It might be useful for someone at Google to look into why their work has backfired in your case.
dog poop dispensers
You mean dogs? OSM must achieve a very high update rate to track all of them.
Two-factor doesn't have to cost much. Your phone can be the second factor. In the case of smartphones, a one-time password generator can be installed as an app, or you can get even more sophisticated and have the web site display a 2D barcode which a phone app photographs, munges into an auth code and sends via the data network. For traditional phones, the site can SMS a one-time password to the phone.
Of course, this assumes that the phone isn't the device accessing FB in the first place.