3 ms·

Anecdotal evidence regarding the proprietary technology trap: 1. At another large (but not XXL) "Internet" company, none of the Google SREs have made the SRE h

Anecdotal evidence regarding the proprietary technology trap: 1. At another large (but not XXL) "Internet" company, none of the Google SREs have made the SRE hiring bar, precisely because half their answers to "How would you solve problem XYZ?" are "Oh, I'd just use $ProprietaryTechnologyGHJ", to which the response is, "Well, but we don't have $ProprietaryTechnologyGHJ", to which the response is "Uhhh..." 2. Of the couple Google (non-SRE) software engineer interviewees I am directly aware of, nobody has made it through either, partially for similar reasons (reliance on opaque external services). To be fair, their other shortcomings were more significant. 3. One Googler that I know confided that they are terrified of leaving Google, because they wouldn't (paraphrasing) "know how to code in the outside world." 4. Yet another Googler I talked to said that their stack is unimaginably huge, and nobody knows how all of it works, which was (quoting) "probably by design." Make of that what you will.

jrockway 13y ago
OK... throwaway account, unnamed company, unverified anecdotes... I should really not reply. Oh well. I'm not sure I believe any of this. Very little of the work we do at Google involves feeding data to some tool and sitting back and watching the results. It's software engineering and problem solving as usual, just with some nice libraries and tools to save time for the hard parts. I'm especially confounded as to what SRE tools would cause someone to "not meet the hiring bar" by responding to a question by saying "I'd just use some proprietary technology". I can't think of any tools that are useful enough to be the singular answer to an open-ended question, and I can't imagine anyone wanting a job saying "just use this tool you've never seen before" in an interview. It's unbelieveable. Being used to certain tools is not specific to Google, either. If you ask a Python programmer trivia questions about Ruby, he's not going to know the answer. But given a week, he'll probably know the answer. Another Googler I talked to said that their stack is unimaginably huge, and nobody knows how all of it works, which was (quoting) "probably by design." Make of that what you will. The stack is big but there is nothing preventing one from understanding other than that it mostly works and so one is not forced to understand it. The code is there and you can start it all up locally on your machine, so I don't really see the problem. It's not huge to keep unhappy Googlers from finding other employment opportunities, as you imply. Anyway, I can't believe I replied to this post and am now going to click the reply button, but here goes...
- throwawaykf 13y ago
  I have no reason to make this up :-/ I comment very infrequently, but only use throwaway accounts as a personal policy based in part on HN-the-community's response to my opinions and HN-the-website's aggressive response to downvotes. (I'm actually re-using a previous throwaway because my last throwaway was autodead-on-arrival. Flagged IP maybe?) Anonymity is not really the point; In fact, there are HN members who might figure out my identity simply because they were directly involved in these experiences. The SRE anecdote may seem unbelievable, but why would this have been the case with multiple Google SREs? I did not imply that their work at Google consists of waving a magic wand and things getting done; it's more likely that the basic building blocks of Google's infrastructure are very powerful, and you don't have to worry about a bunch of things that other SREs have to. "Nice libraries and tools" is an understatement. Seems it's not only that they are reliant on $ProprietaryTechnologyG, they could not explain how $ProprietaryTechnologyG works or even describe alternative methods they'd use instead. On the other hand, it is possible that it's only been the lesser SREs that have been interviewing around. Also: The two Googlers I talked about (one of whom has been there longer than you, if I remember your HN comment history correctly), are doing very interesting work, and are very happy at their jobs. As a disclaimer, I know one well and the other I met only once. Nonetheless, they, in no unclear terms, said what I paraphrased and quoted above. I didn't mean to imply that the stack was made so huge to keep unhappy employees in. In fact, I did not mean to imply any intent at all. However, as you said, since employees are not forced to understand it, most of them don't, and this becomes just another factor to consider when thinking of changing jobs. Not knowing the whole stack is perfectly fine, but when critical pieces have no comparable equivalent in the outside world, your skills become less transferable. (Plus, you might not feel like leaving all that good stuff behind!) As I said in another comment upthread, the "proprietary stack" aspect at Google seems to be more pronounced than at other companies like Microsoft. I mean, come on, it's a commonly accepted fact that it takes a year to become productive after you start at Google. What reasonable conclusion about the stack can one draw from that?

jrockway 13y ago

OK... throwaway account, unnamed company, unverified anecdotes... I should really not reply. Oh well. I'm not sure I believe any of this. Very little of the work we do at Google involves feeding data to some tool and sitting back and watching the results. It's software engineering and problem solving as usual, just with some nice libraries and tools to save time for the hard parts. I'm especially confounded as to what SRE tools would cause someone to "not meet the hiring bar" by responding to a question by saying "I'd just use some proprietary technology". I can't think of any tools that are useful enough to be the singular answer to an open-ended question, and I can't imagine anyone wanting a job saying "just use this tool you've never seen before" in an interview. It's unbelieveable. Being used to certain tools is not specific to Google, either. If you ask a Python programmer trivia questions about Ruby, he's not going to know the answer. But given a week, he'll probably know the answer. Another Googler I talked to said that their stack is unimaginably huge, and nobody knows how all of it works, which was (quoting) "probably by design." Make of that what you will. The stack is big but there is nothing preventing one from understanding other than that it mostly works and so one is not forced to understand it. The code is there and you can start it all up locally on your machine, so I don't really see the problem. It's not huge to keep unhappy Googlers from finding other employment opportunities, as you imply. Anyway, I can't believe I replied to this post and am now going to click the reply button, but here goes...

throwawaykf 13y ago
I have no reason to make this up :-/ I comment very infrequently, but only use throwaway accounts as a personal policy based in part on HN-the-community's response to my opinions and HN-the-website's aggressive response to downvotes. (I'm actually re-using a previous throwaway because my last throwaway was autodead-on-arrival. Flagged IP maybe?) Anonymity is not really the point; In fact, there are HN members who might figure out my identity simply because they were directly involved in these experiences. The SRE anecdote may seem unbelievable, but why would this have been the case with multiple Google SREs? I did not imply that their work at Google consists of waving a magic wand and things getting done; it's more likely that the basic building blocks of Google's infrastructure are very powerful, and you don't have to worry about a bunch of things that other SREs have to. "Nice libraries and tools" is an understatement. Seems it's not only that they are reliant on $ProprietaryTechnologyG, they could not explain how $ProprietaryTechnologyG works or even describe alternative methods they'd use instead. On the other hand, it is possible that it's only been the lesser SREs that have been interviewing around. Also: The two Googlers I talked about (one of whom has been there longer than you, if I remember your HN comment history correctly), are doing very interesting work, and are very happy at their jobs. As a disclaimer, I know one well and the other I met only once. Nonetheless, they, in no unclear terms, said what I paraphrased and quoted above. I didn't mean to imply that the stack was made so huge to keep unhappy employees in. In fact, I did not mean to imply any intent at all. However, as you said, since employees are not forced to understand it, most of them don't, and this becomes just another factor to consider when thinking of changing jobs. Not knowing the whole stack is perfectly fine, but when critical pieces have no comparable equivalent in the outside world, your skills become less transferable. (Plus, you might not feel like leaving all that good stuff behind!) As I said in another comment upthread, the "proprietary stack" aspect at Google seems to be more pronounced than at other companies like Microsoft. I mean, come on, it's a commonly accepted fact that it takes a year to become productive after you start at Google. What reasonable conclusion about the stack can one draw from that?