In a comment reply by the vid creator he states he was "dead wrong" about his conclusions in this vid about GO vs Python for ML. Leaving this vid online for so long just creates confusion for people getting into ML. I think this vid should at least have a pinned comment stating these facts...
Good idea. But AI is so prominent Python will become just one of all of them since it is affecting everything. Note that I was citing another established professional with a blog making the claims, but PyTorch along is all the evidence most need. Also, Python programmers tend to be unique and unable/unwilling to learn any new languages. I see this constantly at work with dozens of machine learning specialists.
Great news is that you never do. Infrastructure work demands you know several languages and Python is assumed. Just don't let it taint your thinking about other bracket/c-like languages.
INDUSTRY ADAPTION is the issue, not necessarily performance or type-safety alone. Coders just now learning Machine Learning, are learning it in Python with TensorFlow & Keras; while veteran coders are too productive with it to switch. One could also argue that Linux is a superior OS than Windows, and be technically correct. But almost worldwide, Windows dominates the Desktop, while Linux is relegated to servers. If given a choice, I'd take Fedora 34 on a new new PC any day of the week, but pc manufacturers can't justify market demand for it. Python is a lot like PHP: people enjoy crapping on it, but both languages still dominate. Why not learn & use both, specific to the use-case? After all, it's not a binary choice.
I still don't see much support for ML in Go. It's just way easy to just take care of those tasks in python. Maybe you can share some resources that you know in the description of the video or do a follow-up video some time?
I still have to make up my mind how much I agree with him, but it is a pleasure to listen to somebody opinionated who is capable to articulate an argument.
After three months working for an infrastructure support team supporting a 200 person machine learning group I've concluded I was dead wrong about this for machine learning. Cybersecurity, however, continues to replace and create new tools in Go for all the right reasons.
i agree on only 1 thing : typing is mandatory. But you didn't address the title at any moment. Python is the cybersecurity language, same for ML. And i don't see it changing, because all the bad things that python have doesn't matter in theses contexts.
Yeah, these early videos are random, but I do see Python already falling to the wayside for things that were traditionally Python's domain, particularly cybersecurity and machine learning (which I was not expecting).
In my opinion, everyone gonna keep using python since it IS so productive. Geeks or Experts may say this or that but the reality is people are going to choose something that is easy, and getting things done, quickly.. And with current python we can even go with static typing.
It’s a bit too strong to say “everyone”, I reckon. I think it will stay around for a while in the application domains where is best suited. But Google, which had A LOT of Python code is also the company sponsoring world class engineers to develop Go… The value of a language is not just how fast you go to production… there are many factors and different cases will have different trade-offs…
@@a0um i agree. Actually i prefer to code in Go than Python for Backend. But many others on the field disagree with me. Just look at the jobs listing between the 2. I also code in NodeJS since the market asks for it. A language is not easily destroyed since programmers usually polyglot.
@@cepuofficial9025 it’s very popular indeed at the moment. I’ve seen the statistics and I was impressed so much that I did consider to improve my Python skills rather than learning Go . I’d bet that the Python community will remain among the top three for the next decade and I agree: its destruction we will not see, but maybe we’ll see another change of trends. I think so because looking at Java, for example, I’ve seen it started losing popularity because it couldn’t be made simpler. New features were added to try to improve the developers experience but nothing was removed! People much smarter than me said the language is really complex if you dig deeper; not as complex C++ maybe but definitely more than necessary. I’ve seen so many Java developers desperate for an alternative moving to at Kotlin, Clojure, Go, Python, etc I think Python will also struggle to keep the lead for a similar reason: it’s basically impossible to remove features from a language that’s been widely adopted. I think that for larger projects or long lived applications the safety, runtime costs (CPU and memory usage), and maintenance cost concerns will play in favor of other languages. Python language and runtime were not designed for these and it will never be as good as other options that were designed for around those objectives, I don’t think.
Now that I support a 200k node ML k8s infrastructure and see the Python everywhere even if Go does have it, people will stick with Python. But I'm more okay with that now that containers make picking a language largely irrelevant. That has become more clear than ever.
Quite interesting to have found this video. I’m currently using python for data science and ML but I always wondered what about other tools like the Java ML frameworks that work beautifully with AWS services. I just started learning Java and might as well start learning Go, not because Go is getting popular or is going to destroy certain languages but most certainly Go will become a useful tool for future ML (And I just want to be a better programmer). Besides, Go ML community might not be as bigger as python. Hence, ones contributions might be little more appreciated. Thanks for the vid, mate. Any IDLE suggestions for Go?
@@rwxrob I heard about Mojo just a few weeks ago. I saw a gradually typed, python-like, AI-focused, close-to-native speed language… and I thought, seems the same agenda of Julia Lang. If you cover Mojo and have time to contrast it with Julia I'd find that interesting. Thanks
0:26 Can't you ship Python bytecode without sources? (I know there are downsides to this, just like there are downsides to shipping a Go binary without sources.)
Other than creating operating systems and fuzzing with them like Gamazo, I don't see it. Too slow to write. No need to be safe. Usually have unsafe enabled to do the exploits. C is a much better language to learn for all of that, even Assembly.
Nope. I was dead wrong. And so was that leading AI professional that wrote the blog I was citing. Ironically, Go would still be a better language for many of these things, but the AI community has solidly decided all they need is PyTorch (for the most part).
But for Cybersecurity Go DID destroy the field. It was responsible for a 4000% increase in malware in 2021 and continue to be the favorite of those creating tools and rootkits.
In a comment reply by the vid creator he states he was "dead wrong" about his conclusions in this vid about GO vs Python for ML. Leaving this vid online for so long just creates confusion for people getting into ML. I think this vid should at least have a pinned comment stating these facts...
Good idea. But AI is so prominent Python will become just one of all of them since it is affecting everything. Note that I was citing another established professional with a blog making the claims, but PyTorch along is all the evidence most need. Also, Python programmers tend to be unique and unable/unwilling to learn any new languages. I see this constantly at work with dozens of machine learning specialists.
I'm currently learning Go, and I'm really liking it , but I still think Python is an amazing language. I have no plans to abandon it any time soon.
Great news is that you never do. Infrastructure work demands you know several languages and Python is assumed. Just don't let it taint your thinking about other bracket/c-like languages.
INDUSTRY ADAPTION is the issue, not necessarily performance or type-safety alone. Coders just now learning Machine Learning, are learning it in Python with TensorFlow & Keras; while veteran coders are too productive with it to switch. One could also argue that Linux is a superior OS than Windows, and be technically correct. But almost worldwide, Windows dominates the Desktop, while Linux is relegated to servers. If given a choice, I'd take Fedora 34 on a new new PC any day of the week, but pc manufacturers can't justify market demand for it. Python is a lot like PHP: people enjoy crapping on it, but both languages still dominate. Why not learn & use both, specific to the use-case? After all, it's not a binary choice.
Yep. I know agree with your conclusions. I work with a bunch of machine learning people and yet, Python is all they want to use.
I still don't see much support for ML in Go. It's just way easy to just take care of those tasks in python. Maybe you can share some resources that you know in the description of the video or do a follow-up video some time?
GoLearn library or contribute to open source community.
I still have to make up my mind how much I agree with him, but it is a pleasure to listen to somebody opinionated who is capable to articulate an argument.
After three months working for an infrastructure support team supporting a 200 person machine learning group I've concluded I was dead wrong about this for machine learning. Cybersecurity, however, continues to replace and create new tools in Go for all the right reasons.
@@rwxrob all of the DS, ML, DL libraries for Python are written in C, C++ or Fortran
@@rwxrob is it back to Python for machine learning for you then or onto a new competitor?
@@rwxrob so u saying python is a better option then Go for Machine learning?
this aged well given the amount of Malware wrtten in Go in 2021
i agree on only 1 thing : typing is mandatory. But you didn't address the title at any moment. Python is the cybersecurity language, same for ML. And i don't see it changing, because all the bad things that python have doesn't matter in theses contexts.
Yeah, these early videos are random, but I do see Python already falling to the wayside for things that were traditionally Python's domain, particularly cybersecurity and machine learning (which I was not expecting).
@@rwxrob Oh I see
In my opinion, everyone gonna keep using python since it IS so productive. Geeks or Experts may say this or that but the reality is people are going to choose something that is easy, and getting things done, quickly.. And with current python we can even go with static typing.
Agreed, static typing even if not required is also good practice in Python.
In other words, the people who really get stuff done have one opinion and the unproductive masses have another. Thank you for reminding us.
It’s a bit too strong to say “everyone”, I reckon.
I think it will stay around for a while in the application domains where is best suited. But Google, which had A LOT of Python code is also the company sponsoring world class engineers to develop Go…
The value of a language is not just how fast you go to production… there are many factors and different cases will have different trade-offs…
@@a0um i agree. Actually i prefer to code in Go than Python for Backend. But many others on the field disagree with me. Just look at the jobs listing between the 2. I also code in NodeJS since the market asks for it. A language is not easily destroyed since programmers usually polyglot.
@@cepuofficial9025 it’s very popular indeed at the moment. I’ve seen the statistics and I was impressed so much that I did consider to improve my Python skills rather than learning Go .
I’d bet that the Python community will remain among the top three for the next decade and I agree: its destruction we will not see, but maybe we’ll see another change of trends.
I think so because looking at Java, for example, I’ve seen it started losing popularity because it couldn’t be made simpler. New features were added to try to improve the developers experience but nothing was removed! People much smarter than me said the language is really complex if you dig deeper; not as complex C++ maybe but definitely more than necessary. I’ve seen so many Java developers desperate for an alternative moving to at Kotlin, Clojure, Go, Python, etc
I think Python will also struggle to keep the lead for a similar reason: it’s basically impossible to remove features from a language that’s been widely adopted. I think that for larger projects or long lived applications the safety, runtime costs (CPU and memory usage), and maintenance cost concerns will play in favor of other languages. Python language and runtime were not designed for these and it will never be as good as other options that were designed for around those objectives, I don’t think.
if go could have libraries for ML like python.
for cybersecurity I see it way better than ML .
Now that I support a 200k node ML k8s infrastructure and see the Python everywhere even if Go does have it, people will stick with Python. But I'm more okay with that now that containers make picking a language largely irrelevant. That has become more clear than ever.
Quite interesting to have found this video. I’m currently using python for data science and ML but I always wondered what about other tools like the Java ML frameworks that work beautifully with AWS services.
I just started learning Java and might as well start learning Go, not because Go is getting popular or is going to destroy certain languages but most certainly Go will become a useful tool for future ML (And I just want to be a better programmer).
Besides, Go ML community might not be as bigger as python. Hence, ones contributions might be little more appreciated.
Thanks for the vid, mate.
Any IDLE suggestions for Go?
VSCode (and that's called an IDE, it's just called IDLE for Python bc of Eric Idle)
Im from learning GO and i completely agree
Two year later I see golang in the same the place 😂
I’m coming from Java backend development and happy to be learning Go.
I wonder, Rob, what do you think of Nim?
I absolutely love Nim, but my money is on Mojo (when it comes out).
@@rwxrob I heard about Mojo just a few weeks ago. I saw a gradually typed, python-like, AI-focused, close-to-native speed language… and I thought, seems the same agenda of Julia Lang. If you cover Mojo and have time to contrast it with Julia I'd find that interesting. Thanks
0:26 Can't you ship Python bytecode without sources? (I know there are downsides to this, just like there are downsides to shipping a Go binary without sources.)
You can.
You can, but you can't run it without python being installed, and not sure if it has to be the same version.
What about the pyjion JIT compiler
Go is Google language. It explains everything. Just next attempt of monopoly and try to dominate market. It not gonna take over anything.
6:09 java is anti OOP? I don't know about that, it literally forces you to write classes for every. single. thing. Do you mean, anti-anti OOP?
Is there a repo for go tensor flow of similar? Would be good to see how it plums together
No idea, I have a feeling I was wrong on this one. Python is really fucking dominating with PyTorch now.
@@rwxrob It's really good. Though I would like to see something similar in Go.
Listen to this whole thing and then it cuts off right when you mention Godot. What we’re you saying?
What about rust in cyber security ?
Other than creating operating systems and fuzzing with them like Gamazo, I don't see it. Too slow to write. No need to be safe. Usually have unsafe enabled to do the exploits. C is a much better language to learn for all of that, even Assembly.
This ghost is super baised, I have seen his vedios on " Rust is Slow" there he promotes Go , now he is saying Go with ML
Yep, but I'm not saying it, ML experts are. But you really don't matter, bubye.
Go👑... is so underrated.
What about Julia for ML?
Well the ending didn't age well :D
Nope. I was dead wrong. And so was that leading AI professional that wrote the blog I was citing. Ironically, Go would still be a better language for many of these things, but the AI community has solidly decided all they need is PyTorch (for the most part).
But for Cybersecurity Go DID destroy the field. It was responsible for a 4000% increase in malware in 2021 and continue to be the favorite of those creating tools and rootkits.
@@rwxrob I take it that the biggest reason for that are go's nice system APIs...
i agree java is the reason why i choose python haha
Heard of numba ?
Golang cannot do nothing against Python at present in ML. Proof me I am wrong. Proofs not Ideas.
Maybe in the future. We have to pick the best tools for a certain task
but one thing there is no better language then c
I actually agree. The one language to rule them all (that I hardly ever write, cuz I don't need it, but probably should).