[Opinion Needed] When Google rejected the creator of Homebrew (popular package manager on Mac)

I just came across this 2015 tweet by Max Howell, the creator of homebrew:

https://twitter.com/mxcl/status/608682016205344768

So, the question is do you agree with companies that do a whiteboard test during interviews? Some agree and some disagree. People who agree with this philosophy often argue that if you can do it in code, why not on whiteboard? Others who disagree argue that in your day to day work you will probably never have to invert a binary tree.

I believe interviewers should first assess if it's really important to possess such knowledge in algorithms to work for their company. If the answer is no, I don't see the point of whiteboard test.

What do you think?

I think these White Board Tests are not important but similar should be there.

They do not show that you know how to solve that specific problem but

how do you approach any problem
how do you inform others or keep them in the loop about your approach and progress

What we miss in a company interview is that it is not a one-man job; it's a team game and it has to be played as a team.

Consider Football, if Messi could have scored all the goals alone, still he will not be able to win the game as Team is all the gears needed to function as a single unit.

I talked to many developers, and I always come across this question of the importance of this kind of whiteboard interview process.

Whiteboard Interviews or similar process is important to be not thrown away but not that important to reject people straight away.

The above statement is ambiguous in itself but you need to give time to analyze a candidate. And once you have analyzed enough, share your analysis and opinion honestly. Like I value Data Structures more, could you implement with certain Data Structures or what if this problem of 100 entries scales 1000 times

Imho, whiteboard tests are a good way to test a number of things.

Can the applicant

talk to their stakeholders
explain their train of thought to a third party
demonstrate logical thinking
demonstrate the knowledge they claim to possess

They are all important for a professional developer's every day life, so checking them in a sane context is a vital part of an interview.

Imho, the above tweet does not include enough context to conclude that the Google guys are the jerks, or if the whiteboard task was evil. It's hard to put things into relation if you only read a two-sentence rage post from one side.

Marco Alka I think it's more about the classic 'theoretical/academic' VS 'applied/trade' problem. there are different expectations and behaviours based on the experience of the participating persons.

The question of trust is important and a lot of other things. But I do love whiteboard tests ;D I just don't see them as very important and the question is how high or low level should the thinking be? also if you did something 200 times vs doing it the first time :D

interviews always suck ;D .... that's how I see it ;D.

I think binary trees are useful to know as an advanced topic, as they power some useful data structures. But I'd not consider it something every programmer should know, just some fields and some levels.

But useful though they are, I've never had to implement one myself for work (I did as hobby only). And I didn't know what inverting means.

Although if inverting is just "mirroring" the whole tree... I hate to be cruel but that's something a programmer with a few years of experience should be able to do. Is that really too much to ask? It may be useless but it's not hard.

I've found several definitions of "inverting" though - maybe the point of the question was to check if the programmer asks for clarification rather than giving up or just assuming things?

Whiteboards are a stupid medium though.

Mark yeah to problem is usually the missing algorithmic training not to solve it. But yes an inversion based on the representation should be easy ;D esp on white board

no, don't do whiteboard test instead we'll talk to the candidate try to get a gage on what we think their ability levels are from asking technical questions.

Most often its a gut feeling about the person if they seem responsive we'll offer them a trial day with the team to see how they get on.

Mostly we don't expect developers to just hit the ground running with us but require on job training and introductions to our workflows.

I find this works well for both parties.

I've seen Coconut tree, Mango tree, Jack fruit tree... but what is Binary tree? Is "Binary" a fruit?

It's very hard to invert a tree myself (and even on a white broad?). We'll need at least a JCB, right?

How many others here don't know what Binary tree is and how to invert them?

P.S: I don't know :(

A tree in CS is a collection of nodes which are connected to other nodes in a hierarchy. The top-most node is called the "root node", and it and any node below it can have children. All nodes, except for the root node, have a parent. For example:

A binary tree is a tree in which each node has at most three connections: a parent (except for the root node!) and 0..2 children.

// TypeScript
interface Node {
    parent?: Node
    leftChild?: Node
    rightChild?: Node
    value: any
}

Trees are useful in order to represent hierarchic structures in code. For example a JSON object can be represented as a tree. In game-dev, you will find lots of quad-trees (trees with nodes which have at most four children) in 2D games, representing four adjacent field (top, right, down, left) and octrees (trees with nodes which have at most eight children) in 3D games.

Binary trees are especially useful when used as binary search trees, because it is very performant to fill the tree with items (O(n)) and just as performant to read a sorted list of items from the trees (O(n)). Searching in such a tree can be very performant, if the items from which the tree was created, weren't pre-sorted. All other operations are very trivial. Sorting works like this: You start with the first item and make it the root node. Then you compare the second item to the first. Is it smaller or bigger? If it is smaller, it goes to the left child-branch, if it is bigger it goes to the right child-branch. For any further item, you start the comparison at the root node and go down all branches until there is no more item at a branch. That's the spot where the new item is inserted as a new node. Searching is just as easy. Compare all nodes until you find a node which is the same. In case of a search tree, it makes sense to store an object with a key (for comparison) and a value (as payload) inside the node.

As for inverting the tree, I guess it depends on what kind of inversion was asked for. They might want to get a reversed version of the tree (mirrored at the Y axis), which could be achieved by switching the branches of all nodes.

Thread

[Opinion Needed] When Google rejected the creator of Homebrew (popular package manager on Mac)

Responses(5)

Recent in Forum

Search Hashnode

[Opinion Needed] When Google rejected the creator of Homebrew (popular package manager on Mac)

Responses(5)

Recent in Forum