These layoffs don't happen in a really hot market.
As much as people like to think that developers are just cogs on wheels, companies still want experienced devs (senior and above), and they are still difficult to hire.
Arguably python is so easy that junior devs aren't that far off from senior devs.
Start asking questions about Linux, or networking, or security, you won’t find many that can answer.
The difference between a junior and senior developer has nothing to do with the language. Regardless of the language the engineering fundamentals are the same.
Also, there's a billion mile difference between a junior's code and a senior's in any language and this is true of Python as much as anything else.
A languages syntactic or design simplicity is irrelevant. Even if it were not, Python's semantics make it particularly easy to write poor code and trigger footguns.
I've written C, JavaScript, Rust, Go, Python, Java and a bit of Erlang ... Python is easily one of the easiest languages to make a mess in if you've not got a solid engineering mindset.
I could go on, your comment was trash in so many ways, which experience would tell you.
As someone who's done 20 years of professional C++ development along with 15 years of professional python development, as well as 10 years of professional javascript/react development on a systems that go across the stack from the lowest level of sensors and embedded systems to the browser front end I know what I'm talking about.
I've also done rust, go, erlang, java, and I've also done a ton of haskell as well as lisp. I've been all over the place likely more than you.
>The difference between a junior and senior developer has nothing to do with the language. Regardless of the language the engineering fundamentals are the same.
This is beyond wrong. You've done C, you've done rust, then you must know that the fundamentals of systems programming is much more wider ranging then python. What's a pointer? What's a unique pointer or shared pointer? What's move semantics? What's the difference between stack allocation and heap allocation? What's an R value vs. L value? What's a borrowed value? What's virtual memory? What's an interface (arguably with the advent of annotations types are not more relevant in python but there is huge class of senior python programmers who don't use types)
The above are programming fundamentals that a python programmer never needs to know. A python programmer can go from junior to principal and not need to KNOW any of the above simply because python abstracts these concepts away from the user. The fact that a python programmer doesn't need to know any of these things makes it significantly easier. For web apps a company is most efficient if they hire a mid level python developer over some principle python developer with 10 years of experience.
I think your getting off on a different topic. Those guys aren't engineers anymore right? They're more technical managers; or architects.
>So yea, deep language knowledge, while an excellent skill, is not something that defines a senior eng (at least in Meta and I presume in similar FAANG companies).
I agree with this, but this is not what the others are talking about. The topic right now is about python developers. The more wider range role you're talking about here isn't a person with the title "Senior python software developer." Meta likely doesn't have that title.
Additionally, if a company works purely within the python ecosystem of web development, the "architecture" actually doesn't get too complex. Most of this stuff is trivial too, I give it an additional year to learn it.
If you're doing stuff with shared memory and utlra low latency applications then it's a completely different scenario, but you wouldn't be working in python if you had that requirement.
It might have heap allocation, Python has numpy.
Rust might ...
Even so, while there _may_ be more technical scope in some languages, or more immediate surface area, this should be true:
A senior developer primarily in language Y should be technically capable enough to be senior in language Z with some degree of work.
The language isn't relevant. Data structures, engineering mindset, pragmatism, intuition, maths reasoning, abstract thinking, ...
These are things that make a senior developer.
If I find a Python dev without this skillset at a high level they're not being put in a senior position because they'll add technical debt, cost and risk without even knowing it. They're a liability.
Your "senior" Python developers don't sound senior to me.
If you think you can promote anyone to senior or principal without the above, you're a liability too. Good job.
You must be from a hacker school. C has pointers.
>It might have heap allocation, Python has numpy.
Python has heap allocation. Every freaking programming language has heap allocation.
It's just in python you don't have to know when it happens and when it doesn't happen. It's hidden from you. You can guess when it happens but you have to be senior enough to know what it means.
When you need to worry about heap allocation, python is the language to NOT use. If you're a python developer, you'll never hit this problem.
>The language isn't relevant.
It is relevant. It's clear to me you don't even know what a heap allocation is and when it occurs. This concept is complex enough that you need a bit of intuition to know it. Glancing at the wikipedia won't provide you with the needed insight. If you did C++ you'd know it and you would also know how to perform this optimization in python.
In fact in fintech and gaming, they don't hire python programmers to do C++ work for a specific reason. It's 10x harder to program C++. Language MATTERS.
>Data structures, engineering mindset, pragmatism, intuition, maths reasoning, abstract thinking, ...
pfft. you think this type of thinking exists only for a select few? Let me put it to you this way. Most people in the world of normal intelligence can learn calculus. If you can learn calculus then you have the capacity to learn all of the above. Which is basically most people on earth.
>If I find a Python dev without this skillset at a high level they're not being put in a senior position because they'll add technical debt, cost and risk without even knowing it. They're a liability.
Yeah give them 1-2 years. They will get those abilities if they want it.
>If you think you can promote anyone to senior or principal without the above, you're a liability too. Good job.
You can't promote anyone. At that level it's largely political. How well can you lead, how well can you direct, how much other people respect you. It's also how a company defines the word principal. The people in these roles are not always the best in terms of technical ability.
The promotion is at the purview of the manager and he doesn't choose such a promotion based off of technical ability. If he did then a standardized test would be the most efficient way to measure technical ability.
For C++ it takes probably 4 or 5 years.
This is not nonsense. This is real.
People who can actually engineer in python is becoming more and more common place.
There's still a deficit of talent, even in Python. I don't see anyone denying this reality.
But to your point, there is a flood of bad candidates.
How would you know did you hire these candidates to find out? Or did you dismiss them as bad because they didn't pass some brain teaser algorithm interview?
Those candidates go on to get hired at companies with a lower bar for lower salaries and those lower salaries do influence the market.
Yes.
We'll still interview them, but they're usually pretty bad. I'm struggling to remember one who wasn't. The one's we've hired have been disasters.
You have to give them some take home exercise. That's the best way to determine if they're not a false negative. Unfortunately many good devs just don't have the time to do a takehome.
Did you hire a candidate who say... didn't know what MRO was, but created a stellar result for a take home project?
What I'm saying is that the market is flooded with these false negatives.
Agreed, but that's not how we interview.
>Did you hire a candidate who say... didn't know what MRO was, but created a stellar result for a take home project?
Do you have me confused with the other guy?
>What I'm saying is that the market is flooded with these false negatives.
Understood. The market is also flooded with incompetent devs, capable of talking at length about Python syntax (and sometimes even the computational model, e.g. MRO, etc.), but who aren't actually able to put that knowledge to use in engineering systems. These are systematically catastrophic hires, by which I mean, the engineering team is significantly better off hiring no-one, than hiring them.
May I ask, why does this seem to matter so much to you? If you are a bootcamp graduate yourself, that doesn't preclude you from being a good engineer. I am more than happy to believe you are the one-in-a-thousand counter-example.
I'm not a bootcamp graduate engineer. I'm actually quite good at what I do, interviewing (from both sides) and engineering. This topic matters to me because I like to argue online and I have strong opinions.
It's common for people to manipulate their perception of reality to justify their own situation, I assure you I am not doing that here, however I think you are doing this.
>>What I'm saying is that the market is flooded with these false negatives.
>Understood.
A false negative is someone who is GOOD, but appears bad. He is a negative but that assignment is false. I am saying the market is flooded with capable engineers that are good but appear bad due to interview processes and biases. Hence the term false negative.
>Do you have me confused with the other guy?
Nope you guys touched on similar points. You never mentioned syntax, but you're likely testing on algorithms so it's the same result. You get a ton of false negatives. You pay a premium for the guy who does better at the interview when you can get someone who's just as capable at a much lower cost.
The point is someone will hire those false negatives at a cheaper price and that effects the whole market. Companies hire the best they can get.
>Agreed, but that's not how we interview.
That's why a certain segment of the market looks like there's a dearth of people. It's because the interview clouds your perception. You interview the wrong way, you agree. The contradiction here is that although most people like you agree, you still reject people based off the SAME results as the WRONG interview AND you think the rejection is accurate.
That's how the psychology works. Everyone I talk to agrees interviewing is wrong overall but when they use it to judge on individual candidates they think it's accurate.
The contradiction exists even in your own reply. You agree the interview process you use is screwed yet you're really confident about your judgement of the market even though your measurement methodology is SELF admitted to be wrong. It renders your entire argument into question.
If you yourself are a bad engineer or terrible at hiring, you might mistake the chaff for the wheat, but then you're doomed anyway.
I think you're not talking from experience.
The same thing happened with JS when that was peaky, and it'll happen in other languages too.
What is old is new again.
I've been hiring roughly the same amount for 10 years. I'm actually really extremely good at modern python. My skills with python syntax are so good that I can do in 1 line exclusively with generators what a junior programmer probably needs 10 lines and twice as much mutating state to pull off. My code is also likely more modular and neater then a junior but you have to realize none of this matters that much. My skill level with python is beyond the curve of what is necessary and so is any other developer that goes beyond 2 years with python.
In the end every programmer I ever hired exclusively for python will hit this tier within 1 or 2 years, very few exceptions. Roughly half of the junior programmers I hire are actually already within this tier upon getting hired. Literally before hitting there first job. After two years the only thing left to pick up are some more unique patterns and tricks but that's it.
I am 100% talking from experience.
It isn't unique to Python.
"Of what is necessary" is subjective. Regardless of the language or framework I'm hiring within they must have the same level of engineering skill. I've hired people for Python projects who were outstanding Go or Java programmers who never touched Python commercially.
My point is the language isn't relevant to your arguments.
That's why you're wrong.
There's also plenty of Python developers two years in (that I've interviewed and rejected) who couldn't write me a decorator or explain MRO.
There's plenty of 2-year Go programmers who can't deal with goroutines well.
First off both MRO and decorators can be learned in 1 week max by anyone. And one week is REALLY forgiving. Literally. You have to be mentally deficient not to be able to learn it. They simply didn't encounter the python feature in their experience. You would just reject them for not knowing about those features?
Second off decorators are syntactical tricks. You can get around python without ever using them. If you encounter them in libraries you just need to know what they do at a high level, you don't need to know how to implement them. If you ever do need to implement one just look it up. I don't judge people for not knowing syntactic tricks.
Third, MRO is something that should be not heavily used at all. You should avoid inheritance as much as possible. Additionally you should avoid double inheritance all together. The MRO/inheritance concept is a left over thing from an outdated programming style. At most you should know inheritance, and at most have one level of it if you must and try to use interfaces instead. Nowadays I expect only to see inheritance in existing libraries/tech debt and not newly written code from senior programmers. If someone encounters MRO issues, again, I just expect them to look it up.
I'll actually judge someone who expects others to know MRO given how outdated it is. I won't think they're stupid, or I won't not hire them or anything like that. I just know they aren't up to date with the current methodology but they can easily learn it if needed.
Overall, you're just setting a high bar artificially. It's like hiring janitors who only have phds in chemistry to know what's in the cleaning chemicals.
> Yes but the same timeline can be done for C, C#, Java or anything else.
No Java and C# languages are complex enough that it can be done in a 2 year timeline but these languages are hard enough that they are in a different class then python. C with memory management is harder then both java and C#. Memory management simply isn't intuitive.
Isn't all programming just syntactical tricks?
We have a fundamentally different view of these things. Mine is serving me well. Yours isn't challenging mine.
Let's agree to disagree.
No. The decorator is really just a syntactic shortcut. There are other facilities in python to do the same thing. The decorator is just a convenient way of doing it. The concept at hand is a function that takes a function and returns a new function. Nothing crazy here.
decorator: Callable[[Callable[[Any]. Any]], Callable[[Any], Any]] = lambda x: (lambda y: x(y)) #does nothing
g: Callable[[Any], Any] = lambda x: x #identity func
decorated_g: Callable[[Any], Any] = decorator(g) #decorator that does nothing to identity returns identity func
>We have a fundamentally different view of these things. Mine is serving me well. Yours isn't challenging mine.
Exactly see. Just because my view is "better" or more "modern" you still function fine within your role which is my point. Someone could have knowledge far superior to a person with 2 years of experience, functionally speaking, it doesn't matter too much. You still do "fine" at your role. Your view serves you well in the capacity of a "senior" engineer" even though your view points are ultimately wrong and inferior they still work well enough.
Also it isn't a viewpoint. Two of the newest languages to hit the mainstream have eschewed inheritance completely from their syntax. Neither Golang or Rust have the concept. Both use interfaces exclusively to promote the concept of similar types. No inheritance anywhere. I have tons of evidence on this even papers, quotations from prominent people. It's a well known fact inheritance is outdated.
We can agree to disagree, but it's like agreeing to disagree with a child who doesn't know any better. Sure.
This is indeed what is happening right now.
I really don't worry about this stuff until it starts looking like the dot-com bomb again. I survived that, I'll survive this.