Page 2 of 2

Re: Learning animals

PostPosted: Mon Nov 16, 2020 4:23 am
by Mr Reasonable
if u become "powerful" as a result of everyone else getting weaker then youre still the same schmuck that you were before just living in a shittier world.

Re: Learning animals

PostPosted: Mon Nov 16, 2020 4:39 am
by Magnus Anderson
Doesn't that apply to one percent Joe?

Re: Learning animals

PostPosted: Mon Nov 16, 2020 4:51 am
by obsrvr524
Magnus Anderson wrote:Doesn't that apply to one percent Joe?

:lol:

Re: Learning animals

PostPosted: Wed Nov 18, 2020 4:51 am
by Mr Reasonable
Magnus Anderson wrote:Doesn't that apply to one percent Joe?


i dont know who that even is

Re: Learning animals

PostPosted: Wed Nov 18, 2020 5:19 am
by Magnus Anderson
He doesn't know who he is either.

Re: Learning animals

PostPosted: Wed Nov 18, 2020 5:27 am
by Mr Reasonable
you mean the president? he has less money than kabbillionare trump right?

Re: Learning animals

PostPosted: Wed Nov 18, 2020 1:20 pm
by Karpel Tunnel
Mr Reasonable wrote:you mean the president? he has less money than kabbillionare trump right?
So, it's a difference of degree. Biden also making money in real estate and investments.

Re: Learning animals

PostPosted: Thu Nov 19, 2020 3:04 am
by fuse
In the context of software engineering, which is probably the relevant context, industry relies on people who are able to quickly learn and adapt to new technologies, tools, and methods. Google isn't saying that everyone in every industry should be a generalist and not a specialist. They are saying that for their industry, it's what really works. What you have to understand is that there is so much arbitrary convention, scaffolding, and tooling in software that it's more important that you have general logical, analytical, communication, and work ethic abilities than that you are a C++ expert with 15 years of experience, with 10 certifications from 5 years ago. This is why software engineering interviews are typically ~5 hour exams where the point is for the interviewer to observe how you problem solve, analyze, and explain complex concepts in real time.

Re: Learning animals

PostPosted: Thu Nov 19, 2020 11:47 am
by Magnus Anderson
fuse wrote:What you have to understand is that there is so much arbitrary convention, scaffolding, and tooling in software that it's more important that you have general logical, analytical, communication, and work ethic abilities than that you are a C++ expert with 15 years of experience, with 10 certifications from 5 years ago.


I don't have the impression that modern day programmers use "logical, analytical [..] abilities".

One tool is replaced by another in an effort to automate certain manually performed tasks. When someone comes along and tells you that you should stop using whatever tools you've been using in the past and start using new ones, they are in effect stealing your job from you. Perhaps not your entire job but most definitely parts of it. The result is that you have fewer responsibilities requiring little to no thought. Everything is taken care of for you, you just have to sit back and relax.

I am not one of those guys who think they have to do everything on their own. I will happily use other people's tools and help when I see it fitting. But if you're going to force me to use other people's tools for something I can do on my own, then we have a problem.

The worst part is that the advanced tools that you're told you should use don't do the job as well as you yourself (or your own tools) can.

The excuse is that "You are freed from manual labor in order to dedicate yourself to higher tasks." But you didn't ask for it, didn't you? It's not a choice you freely made but something you were forced to because a job has been stolen from you. And it is not a choice because you are not ready for it. You know nothing about, have no experience with, the higher task you are supposed to dedicate yourself to. Your expertise has been, in fact, reduced to zero. There's no longer anything to distinguish you from someone fresh out of school.

And if the average lifespan of a technology is between 3 and 10 years, then no true expertise can ever be attained let alone maintained.

Re: Learning animals

PostPosted: Thu Nov 19, 2020 11:57 am
by fuse
Magnus Anderson wrote:
fuse wrote:What you have to understand is that there is so much arbitrary convention, scaffolding, and tooling in software that it's more important that you have general logical, analytical, communication, and work ethic abilities than that you are a C++ expert with 15 years of experience, with 10 certifications from 5 years ago.


I don't have the impression that modern day programmers use "logical, analytical [..] abilities".

Maybe let's start with what you think the engineers at Google do?

Re: Learning animals

PostPosted: Thu Nov 19, 2020 12:13 pm
by phoneutria
oh oh i know that one
they browse stack overflow
did i get it right?


(this is just a joke)

Re: Learning animals

PostPosted: Thu Nov 19, 2020 12:23 pm
by Magnus Anderson
That's difficult to tell. But I think they all rely heavily on Google search.

Re: Learning animals

PostPosted: Sat Nov 21, 2020 7:39 am
by fuse
Algorithms, complexity analysis, data structures, design patterns, interoperability, and security - these core competencies are much more essential to the job than any particular set of tools or programming languages. The latter can be picked up on the fly, with a firm grasp of the former. In a sense, those core competencies together are the specialty. But each of those topics is also a path of specialization. So do you go all the way in one dimension? There is an argument for that and some level of need for it. Or do you go for multi-dimensional competency, plus the possibility of future specialization, and get the knock-on effect of being able to engineer generally sophisticated systems and applications? This is where being a generalist comes into play.

Magnus Anderson wrote:The thing is, I find it necessary to relate new information to existing information, and if I can't do that, I can't proceed.

We're all doing that, inescapably. Iterating on our baseline model, some more efficiently than others. Googlers aren't amnesiacs. But many were selected because of their general ability to learn and communicate complex ideas (where talking through CS theory and toy programming problems are often thought to be good proxies for this ability). That doesn't mean they don't have core skills and competencies, they do.

Re: Learning animals

PostPosted: Sat Nov 21, 2020 8:01 am
by fuse
I mean, you're free to dissect them for horrific "modernity," if that's what this is about. I'm sure they have that, too. I'm sure you'll find it. Have at it.

Re: Learning animals

PostPosted: Sat Nov 21, 2020 9:12 am
by Magnus Anderson
This thread isn't so much about Google (and IT industry as a whole) as it is about rapid technological development and its consequences.

The basic premise is that the faster the environment we live in changes the less time we have to think -- the less useful intelligence becomes. Rapidly changing environments do not favor intelligence.

I would expect Google to be one of the few companies with a good number of employees who not only have a lot of responsibilities but too many of them. (On the other hand, I believe that Google employees aren't "born equal", so it might be useful to look at how responsibilities are distributed among their employees.)

Re: Learning animals

PostPosted: Tue Nov 24, 2020 4:22 am
by obsrvr524
James S Saint » Thu Mar 19, 2015 1:48 pm wrote:There is absolutely nothing that a human can do that a machine cannot be designed to do better - much, much better. But it has to be sold to the public. So like all social movements, the ones being promoted have to be seen as the poor, helpless, abused underdog until there is no escape.

Your faith in human, conscious superiority is pure superstition and wishful thinking.

Re: Learning animals

PostPosted: Tue Nov 24, 2020 4:37 am
by Mr Reasonable
.

Re: Learning animals

PostPosted: Wed Dec 02, 2020 8:37 am
by fuse
Magnus,

It's an interesting question, but I have to point out that I was responding to the set-up in the OP, which addresses a substantially different question.

The OP essentially asks how to interpret "this process of learning promoted by Google." It has to be done in the context of the software industry, and I've argued that the following interpretation
Magnus Anderson wrote:They are basically people who are able to quickly learn whatever is thrown at them -- they have no other qualifications.

is not what was meant and doesn't really capture how software engineers are vetted. They are vetted for a variety of qualifications, of course, as I mentioned above. By any general standard, a software engineer/dev/etc. is a specialist. That is, they have specialized skills and knowledge for a particular industry, as described earlier. Within software engineering, however, there is a further distinction between being a specialist or being a generalist. If you actually want to know what was meant by the term learning animals, you probably shouldn't divorce the conversation from Google and its industry.

Re: Learning animals

PostPosted: Wed Dec 02, 2020 9:51 am
by fuse
I would say Google's perspective is that the challenges of their business/industry are often very novel. And you need a certain type of person/intelligence to solve novel problems, problems whose scope or fundamental characteristics don't necessarily have antecedents. So as a company let's say they're looking to attract intelligent and skilled people in general (they are), but also that of this class of people they are particularly interested in something like fluid intelligence over crystallized intelligence. I find the fluid/crystallized distinction a useful reference point to get our bearings; I'm not suggesting this is exactly what Google had in mind, but it's an interesting possibility. They know not everyone they hire will be a brilliant thinker who will solve the novel problems of deep learning or quantum computing, and practically they vet applicants by CS and other competencies, but they still want to encourage the idea that they highly value people who are willing and able to integrate complex concepts for which they've had no prior exposure.

I don't think it's the speed of changing conditions which would disfavor intelligence, but the non-regularity (arbitrariness) of changing conditions that would disfavor intelligence as we know it. Without some kind of regularity, adherence to patterns or "laws," changing conditions become unpredictable by reason or experience. Rapid change however is simply a kind of intelligence test. (There may be some forms of intelligence that don't fare well in this kind of test, but it doesn't make sense to me to say that intelligence in general, intelligence qua intelligence, becomes useless or non-viable amidst rapid change.) Of course, if the change is too rapid for the group to handle, we as a species might fail that test. A rapidly changing environment certainly does not favor poor intelligence. We should also define a bit more explicitly what we mean by rapidly changing environments so that it's clear we're referring to something like the pace of technological -- and by extension social/cultural -- change, and not, say, change at the level of a sudden lightening strike or rockslide (where reflex, for example, would reasonably be thought to be favored over intelligence).

Re: Learning animals

PostPosted: Wed Dec 02, 2020 8:47 pm
by Magnus Anderson
The way I understood it, and I think it's in any one of the articles you can find on the Internet regarding the concept of "learning animal", is that they need these "learning animals" as a consequence of rapid technological development. That suggests to me they need people who can cope with a lot of change ("whatever is thrown at them") which means people who rely more on intuition (which is lesser form of intelligence) than reason.

If you need to act quickly in order to survive, you need to make decisions fast. And if you haven't already figured out how to make quick but high-quality decisions (and they obviously haven't otherwise they wouldn't be playing down the value of experience), you have no choice but to make quick but low-quality decisions.

It's quantity over quality or r-selection over K-selection.

Re: Learning animals

PostPosted: Sun Dec 13, 2020 1:01 pm
by fuse
Magnus Anderson wrote:The way I understood it, and I think it's in any one of the articles you can find on the Internet regarding the concept of "learning animal", is that they need these "learning animals" as a consequence of rapid technological development.

Well yes, it's in Rosenberg's quote:

Rosenburg wrote:I think that people don’t realize that, fundamentally, we’re focused on learning animals or generalists as opposed to specialists. And the main reason is that when you’re in a dynamic industry where the conditions are changing so fast, then things like experience and the way you’ve done a role before isn’t nearly as important as your ability to think.

"When you're in a dynamic industry," say, Google's industry, which is synonymous with rapid technological development, it's plausible that the ability to think and learn in general could at some point take precedence over your particular career experience so far (your resume) or simply knowing the trivia of x or y programming language or past conventions. In fact, I hardly see how his statement could be that controversial. For one, it would be odd to presume that the "ability to think" has nothing to do with reasoning ability. The ability to think and learn has a lot to do with reason and is of general benefit to pretty much every discipline. If you're pursuing a static field like the study of Ancient Greek, where neither the object of study nor the conventions of the discipline will ever change much - then, yeah, it's your experience and years spent with the language that carry the most importance. Still other fields might require a different balance between experience and raw ability to think (including reason).

More controversial are the implications of such rapidly advancing technology for civilization. It doesn't seem likely that the tech explosion we're living through will be rolled back of our own volition. It's kind of like pandora's box has been opened. It's scary, also fascinating and potentially hopeful at the same time. But it really isn't a question, at this point, of do we want a rapidly changing environment where we need to learn quickly? It's already here, we must. And the best way to do that, in pretty much any context, is to seek out quicker feedback cycles. Rapid iteration. Move fast and break things**.. but also don't mistake motion for progress. These are major themes of software engineering.

**everyone seems to have an opinion about what this phrase really means; imo it refers to the hacker ethos of testing out new ideas and approaches, of breaking the right things and learning from it. Instead of spending time speculating, perfecting, or optimizing, people learn best by doing and testing and learning from the results/errors; the more frequent this feedback cycle, the more we learn.

Magnus Anderson wrote:If you need to act quickly in order to survive, you need to make decisions fast. And if you haven't already figured out how to make quick but high-quality decisions (and they obviously haven't otherwise they wouldn't be playing down the value of experience), you have no choice but to make quick but low-quality decisions.

I just want to note that in context Rosenburg isn't talking about reacting with split-second decisions. He's talking about an eight-hour-ish work day where people are designing, building, and testing at a fast pace, in an industry where paradigms shift frequently and new ground is being broken all the time. The point is you have to be able to make high-quality decisions without the benefit of experience. You have to reason. You have to test. If you had all the experience, then you would simply know what to do and you wouldn't really need to think much about it or test your ideas; they'd have already been tested. But on the bleeding edge, you can't expect the answers to be known. There's won't always be a fully-formed body of knowledge yet for you to consult. I mentioned before, there are still core skills and competencies - experience is still valued - but at a certain point raw ability to think and learn takes precedence.

Re: Learning animals

PostPosted: Sun Dec 13, 2020 6:04 pm
by Magnus Anderson
Hi fuse, I appreciate the detailed response :)

fuse wrote:"When you're in a dynamic industry," say, Google's industry, which is synonymous with rapid technological development, it's plausible that the ability to think and learn in general could at some point take precedence over your particular career experience so far (your resume) or simply knowing the trivia of x or y programming language or past conventions.


I think that experience makes you efficient. When you do something over and over again, you become good at that particular thing. And that applies to thinking too. When you think on certain subject for an extended period of time, you become good at thinking on that particular subject.

Of course, in many if not most cases, when you become good at one thing, you also become better at some other things (due to the fact that most things have plenty in common.) So when you become good at X, you also become better at Y. But whenever you try to use what you learned by doing X in order to do Y, you will perform significantly worse than someone who has direct experience with Y. And this applies to thinking too.

You're probably aware of the fact that one can be a good mathematician but a poor sociologist. Even though the two subjects share a lot in common (you can become a considerably better sociologist by becoming better at mathematics), there's still a lot they don't.

I don't think you can become good at chess merely by becoming good at mathematics. You can become better at it but to a very limited extent. In order to become good at chess, you need to specifically think about chess.

That said, I do seriously think that you can't become good at anything without spending enough time learning how to do, not something else, but that particular thing.

In fact, I hardly see how his statement could be that controversial. For one, it would be odd to presume that the "ability to think" has nothing to do with reasoning ability. The ability to think and learn has a lot to do with reason and is of general benefit to pretty much every discipline. If you're pursuing a static field like the study of Ancient Greek, where neither the object of study nor the conventions of the discipline will ever change much - then, yeah, it's your experience and years spent with the language that carry the most importance. Still other fields might require a different balance between experience and raw ability to think (including reason).


The problem is with "the general ability to think". For any kind of job one can think of, that's simply not enough. You need to specialize. Otherwise, you will be easily outcompeted by specialists. (Unless, of course, someone is regulating the so-called free market such that specialists stand no chance.)

The point is you have to be able to make high-quality decisions without the benefit of experience.


And how is that possible?

If you had all the experience, then you would simply know what to do and you wouldn't really need to think much about it or test your ideas; they'd have already been tested.


Well, I don't think anyone has all the experience. But I don't think the only remaining option is "No experience" :)

Re: Learning animals

PostPosted: Mon Dec 14, 2020 6:58 am
by fuse
I appreciate the conversation, too.
Sorry for the delay in responses - will reply soon when I have some more time.

Re: Learning animals

PostPosted: Mon Dec 14, 2020 8:19 am
by obsrvr524
I think in every business there are only 3 kinds of people being sought for hire -
  • People who can invent or develop new product
  • People already trained on a product
  • People easily trainable on a product

That is true for every kind of business.

Only a few are ever needed for the first kind. They invent something new then something else then something else and quickly the backlog of new product leaves them with nothing to do except wait for the roll out to catch up so they can invent new product based on the new product. Designers can run out of work quickly. This kind is at the top of the pyramid of workers - sometimes only the business owner.

The second kind of employee is the most favored because he costs least to produce most. He hits the ground running. But in a quickly changing field those are hard to find. And again if the product is quickly changing that person is only good for a short time. So those get hired for more but get laid off or reassigned quickly.

The third is obviously the most prevalent in the population and least payed - untrained. IF they can learn quickly they become useful quickly but then might have to drop that and be able to learn another product quickly - again and again. Before long, it is cheaper to just drop that employee and get another beginner at even a lower wage.

So the first and the third become the mainstay of fast developing businesses - the inventors (very few) and the learners (the much greater portion). The second kind who already know the current product are harder to find and often not worth their temporary usefulness.

Re: Learning animals

PostPosted: Thu Dec 17, 2020 4:47 am
by Magnus Anderson
And what exactly do these "trainable animals" end up learning? :)

When they get a job, they learn whatever is necessary to do the job. Could be anything really but in most cases it's a trivial task that can be learned quickly. Then they use their newly acquired knowledge for a while until they lose their job and are forced to look for another one.

What kind of knowledge do they end up with?

Sure, they end up learning a bit about everything, but since they spend very little time applying what they learn, how long does it stay in their memory?

Perhaps they end up forgetting it, perhaps they end up pushing it in the part of their brain from which it is difficult to retrieve information. In any case, it doesn't seem like it ever becomes their "second nature".

What they become good at, it appears to me, is how to do whatever they are told to do.

That seems to be their specialty.

Here's the thing:

If you can keep a job, you will focus on becoming good at that job.

If you cannot keep a job, you will focus on becoming good at finding another one.

Your job becomes finding a job.