4 ms·

You raise an interesting point. I think most non-FANG engineers would say that that the leetcode culture is getting out of hand, and the FANG's quest for the 10

You raise an interesting point. I think most non-FANG engineers would say that that the leetcode culture is getting out of hand, and the FANG's quest for the 10x engineer leads to both (a) rocket high salaries+competition from the people who can pass leetcode, and (b) passing on lots of smart engineers who could not or did not want to run the interviewing gauntlet. What is the economic sense to not lowering the bar rather than constantly attempting to raise it?

sokoloff 5y ago
There’s coordination costs that scale super-linearly with people and many of the FAANGs have such high leverage from a slightly higher output of a software engineer that the costs aren’t a barrier for them. In other words, if they don’t have to lower their standards for financial reasons, they just don’t.
runawaybottle 5y ago
I made the point in another comment that one of the reasons other industries kind of suck is because of the heavy gate-keeping. You want to be an investment banker? There is no path but to have gone to an Ivy League school. Stuff like this is a tragedy. Industries that are cordoned off in this way are extremely selfish. I understand the need for the elite to be brain surgeons, but for a vast majority of professions, the filtering for elite is not conducive to a healthy village. Leetcode is fine, but the trend I’m hearing is that it is getting more and more absurd and somewhat becoming our version of the same gate-keeping that makes finance, law, politics, etc bad. I’m hopeful overall since you can’t bullshit good software. Over time, the less elite can make better products and make the gatekeeping look stupid when it’s all said and done.
- dasyatidprime 5y ago
  That's very dependent on whether the definition of “good” can be yanked around by what the rest of the industry does, for example by ease of compatibility with other software APIs or by UI and behavior models that are familiar to lower friction for people who have already used something popular.
- dlp211 5y ago
  > You want to be an investment banker? There is no path but to have gone to an Ivy League school This is mostly, but not completely true. There are other paths. I was recruited by a large bank to come interview for an IB position from a recruiter that specifically looks for talent outside of the Ivy's. What made me stand out from my peers was the fact that I served 7 years in a Special Operations military unit, had a perfect GPA, and multiple high recognition internships all while a dual major. I turned down the offer as I had friends that went that route and either burned out or simply hated their jobs and I didn't want that, but the point is that there are other paths into IB then going to an Ivy.
- 908B64B197 5y ago
  > I understand the need for the elite to be brain surgeons, but for a vast majority of professions, the filtering for elite is not conducive to a healthy village. Filtering for surgeons really is: - Who can afford a completely unrelated undergrad before even applying to med school - Who can afford to fly and meet with hospitals for residency spots - Who can be the most sleep deprived during residency (completely useless) > Leetcode is fine, but the trend I’m hearing is that it is getting more and more absurd and somewhat becoming our version of the same gate-keeping that makes finance, law, politics, etc bad. Really, it's possible to bypass leetcoding by just taking a good algorithm and data structure class.
shagie 5y ago
There are a lot of new grad and junior developers who can follow instructions. There are surprisingly few who can solve problems that they haven't seen before and are able to pattern match against. The issue isn't lowering the bar. For a new grad/junior the bar is often lowered to the point of "can you do FizzBuzz in the language we use?" (and there are a lot who can't do that). Rather, the issue is the "we need to solve problem XYZ" and have the developer figure out how to solve it. For example, I just finished debugging a problem of an application that would run fine in the IDE, but fail when run on the command line when packaged into a jar. It turned out that the cause was that you can't open up a File inside a .jar (but you can open up an InputStream). It got passed to me when the developer failed to get it working and gave up on the problem. The happy path failed. That isn't a "lower the bar" thing - the developer who wrote it can write reasonably complex code. It's a lack of problem solving of the unknown problems. This sort of problem isn't something that is easily tested by adjusting the bar. Its a "we're graduating people who don't have problem solving skills and are finding that those skills are rare."
- brutus1213 5y ago
  Great example. Leetcode absolutely does not test for this. Frankly, no FANG interview question I have ever personally seen tests for this. This sort of deep experience comes out when you actually talk to a candidate about their technical work, what was a really subtle bug you tracked down, etc.
  - shagie 5y ago
    Its really hard to test for that as part of an interview question. FANG is able to resolve that by hiring lots of people that pass a sufficiently high bar and then figuring it out as they go... and even when the people don't have the appropriate problem solving skills, there's likely still work for them to do. Smaller shops that don't have the luxury of hiring many developers may find that only 5-10% of their devs have the necessary problem solving skills... and that's a hard pill to swallow. Those skills can be cultivated. Some get there through shear grit, but it takes time. But this is where jumping from one job to another gets problematic - it takes time. If someone is switching jobs each year its harder to know if they've gotten to the point where they can solve problems.

sokoloff 5y ago

There’s coordination costs that scale super-linearly with people and many of the FAANGs have such high leverage from a slightly higher output of a software engineer that the costs aren’t a barrier for them. In other words, if they don’t have to lower their standards for financial reasons, they just don’t.

runawaybottle 5y ago

I made the point in another comment that one of the reasons other industries kind of suck is because of the heavy gate-keeping. You want to be an investment banker? There is no path but to have gone to an Ivy League school. Stuff like this is a tragedy. Industries that are cordoned off in this way are extremely selfish. I understand the need for the elite to be brain surgeons, but for a vast majority of professions, the filtering for elite is not conducive to a healthy village. Leetcode is fine, but the trend I’m hearing is that it is getting more and more absurd and somewhat becoming our version of the same gate-keeping that makes finance, law, politics, etc bad. I’m hopeful overall since you can’t bullshit good software. Over time, the less elite can make better products and make the gatekeeping look stupid when it’s all said and done.

dasyatidprime 5y ago
That's very dependent on whether the definition of “good” can be yanked around by what the rest of the industry does, for example by ease of compatibility with other software APIs or by UI and behavior models that are familiar to lower friction for people who have already used something popular.
dlp211 5y ago
> You want to be an investment banker? There is no path but to have gone to an Ivy League school This is mostly, but not completely true. There are other paths. I was recruited by a large bank to come interview for an IB position from a recruiter that specifically looks for talent outside of the Ivy's. What made me stand out from my peers was the fact that I served 7 years in a Special Operations military unit, had a perfect GPA, and multiple high recognition internships all while a dual major. I turned down the offer as I had friends that went that route and either burned out or simply hated their jobs and I didn't want that, but the point is that there are other paths into IB then going to an Ivy.
908B64B197 5y ago
> I understand the need for the elite to be brain surgeons, but for a vast majority of professions, the filtering for elite is not conducive to a healthy village. Filtering for surgeons really is: - Who can afford a completely unrelated undergrad before even applying to med school - Who can afford to fly and meet with hospitals for residency spots - Who can be the most sleep deprived during residency (completely useless) > Leetcode is fine, but the trend I’m hearing is that it is getting more and more absurd and somewhat becoming our version of the same gate-keeping that makes finance, law, politics, etc bad. Really, it's possible to bypass leetcoding by just taking a good algorithm and data structure class.

shagie 5y ago

There are a lot of new grad and junior developers who can follow instructions. There are surprisingly few who can solve problems that they haven't seen before and are able to pattern match against. The issue isn't lowering the bar. For a new grad/junior the bar is often lowered to the point of "can you do FizzBuzz in the language we use?" (and there are a lot who can't do that). Rather, the issue is the "we need to solve problem XYZ" and have the developer figure out how to solve it. For example, I just finished debugging a problem of an application that would run fine in the IDE, but fail when run on the command line when packaged into a jar. It turned out that the cause was that you can't open up a File inside a .jar (but you can open up an InputStream). It got passed to me when the developer failed to get it working and gave up on the problem. The happy path failed. That isn't a "lower the bar" thing - the developer who wrote it can write reasonably complex code. It's a lack of problem solving of the unknown problems. This sort of problem isn't something that is easily tested by adjusting the bar. Its a "we're graduating people who don't have problem solving skills and are finding that those skills are rare."

brutus1213 5y ago
Great example. Leetcode absolutely does not test for this. Frankly, no FANG interview question I have ever personally seen tests for this. This sort of deep experience comes out when you actually talk to a candidate about their technical work, what was a really subtle bug you tracked down, etc.
- shagie 5y ago
  Its really hard to test for that as part of an interview question. FANG is able to resolve that by hiring lots of people that pass a sufficiently high bar and then figuring it out as they go... and even when the people don't have the appropriate problem solving skills, there's likely still work for them to do. Smaller shops that don't have the luxury of hiring many developers may find that only 5-10% of their devs have the necessary problem solving skills... and that's a hard pill to swallow. Those skills can be cultivated. Some get there through shear grit, but it takes time. But this is where jumping from one job to another gets problematic - it takes time. If someone is switching jobs each year its harder to know if they've gotten to the point where they can solve problems.