Hacker News Reader

Comment by startupdiscuss | Hacker News Reader

I just have a point about the last piece of advice:

"Make small convex investments. Learn a little bit of a lot of things–machine learning, internet of things, virtual and augmented reality, 3D printing."

I find a lot of advice to be diametrically opposed to this. Many other people advise you to "go deep" in a single topic and be the expert on that topic. (It also seems to contradict the point above it, but only if they are both read a certain way.)

So which is better advice? Is there a study out there?

View on HN

I like Scott Adam's take: http://dilbertblog.typepad.com/the_dilbert_blog/2007/07/care...

To be especially valuable and impactful, it's easier to be very good at two things and own the intersection than it is to be the best at one thing.

That's awesome. I hadn't read that before, but it makes perfect sense to me.

Now just to find out where the intersection of my skills are. Right now, I'm not entirely sure what that would be.

There aren't going to be a study on there because the most relevant feature is evolutionary adaptation. Going deep on a single topic works great when that topic is in vogue, but leaves you open to disastrous results if that topic loses its sheen (or becomes over-subscribed with competition). Think, for example, of those poor engineers who invested their lives in BlackBerry systems. Conversely, seeking to build a wide range of skills necessarily reduces your ability to go deep on any one topic, but gives you the ability to more quickly react to shifts in the industry.

Thus, anecdotally at least, my company (and I) advise our staff to become 'T-shaped' - have one area of deep competence, but familiarize yourself with other areas at least to the extent that you can have an intelligent conversation on the topic. This leaves you desirable for your knowledge but not forever tied to that one area.

I know you wanted a study, and the above is anecdotal, but it's based my experience recruiting and many discussions with candidates.

I'm not sure BlackBerry developers are a good example. Some of the best engineers I know working in mobile got there because they worked for years on BlackBerry learning how professionals used their phones, how to optimize software for battery life, how to adapt company workflow to mobile, etc. Making the switch to the modern incarnation of the smartphone with touchscreens wasn't difficult.

It seems that outside a few safe areas (compilers, operating systems, etc) it's really hit or miss on whether your skills are adaptable to the evolving tech landscape.

Lot's of Flash developers got caught with their pants down. If you were good, you moved into iOS (or jumped on the ever morphing javascript train), but lots of them were not T shaped and struggled to transition their skillset.

Fair enough. I'm sure you can think of a technology, though, that made nice coin for a few years until suddenly it didn't.

If not Blackberry, then:

1. Novell Networks.

2. SAP.

3. Peoplesoft.

4. COBOL.

5. Y2K.

6. Perl.

7. Visual Basic.

8. CORBA.

9. XML.

10. Freeway Engineering (if you were in California in the 1970s).

11. Nuclear Engineering (if you were anywhere on Earth after 28 March 1979).

There's an old saying, "Jack of all trades, master of none." Less well-known is the rejoinder: "though ofttimes better than master of one".

The point being that finding yourself with a single competency can be narrowing. Multiple points of expertise give you leverage and pivot capability. The balance is in sorting where and how to develop expertise.

They're not mutually exclusive. You can learn a little bit of a lot of things on the path to "going deep" in a single topic. I believe this is how you end up with T-shaped skills: https://en.wikipedia.org/wiki/T-shaped_skills

That's great, another one of those things that I've never found the right term to articulate but lines up very well with my intuitive understanding.

Going wide also has benefits as you'll stumble across things that have surprising application in the areas where you've gone deep.

The other aspect to look at is if you're only an expert in machine learning, every problem looks like it can be solved with machine learning.[0]

[0] https://en.wikipedia.org/wiki/Law_of_the_instrument

Exactly this. It's about finding a balance: know enough about a specific topic that you're a go-to person ("an expert") and enough about other topics to understand the context and opportunities for your area of expertise.

Too much focus and you can't see the forest for the trees. Too much general knowledge and you're a jack of all trades, master of none.

Finally, note that "balance" is subjective. It's the balance that works for you and where you want to go. Deep, focused experts are valued. Generalists are valued. Find your niche.

My dad is quite fond of this adage from a journalism school professor in the 80s: "know all about one thing, a little about everything, and where to find the rest."

"T-shaped skills" or whatever you want to call them + the capacity to learn is quite powerful.

Agreed. Learning how to learn is vital, and certainly not easy.

I'd love for this to be true, but I constantly find myself limited by time. I either have time to learn one thing really well or a little bit about a lot of stuff -- if I try to do both, I never end up getting the deep knowledge of the first thing.

A few years ago I gave a talk at a university discussing this very topic. My view is that you should focus more on an "arrow" shape:

- become a senior with one topic (you don't have to be the ultimate master on this) - have mid-level skills on topics surrounding that one main skill - have a junior-level skillset on the outlying topics - have some knowledge of anything beyond that

For example: a back-end developer should be really good at back-end development but they should have a very good understanding of the front-end, DevOps, and DB as well. They should have some rudimentary understanding of how UI/UX works and system administration as well. Beyond that, they should have basic ability to understand what's being discussed regarding product development, marketing, how the OS works on bare-metal, etc. etc.

I feel like that makes a more well-rounded employee that has an easier time moving around their space, understanding their co-workers, and, worst comes to worst, shifting to a new topic when it becomes relevant.

A lot of people have responded by recommending the T shape. I had never heard of it before.

Why is the better than being n or m shaped (great at 2 things; fairly good at 3 things) and so on?

It seems the point about the deep skill going obsolete still stands.

If your one deep skill does go obsolete then being arrow-shaped better enables you to shift one of your mid-level skills to expert level, in which case you do become M-shaped . . .

Personally I like the octagram !!!

Become a specialist expert in something well-paying and bound to be in demand for a long time, but learn enough about everything else that management doesn't consider you outdated when they decide to pivot, and may keep you on during the transition.

Become an obsequious corporate yes-man, but be resourceful enough that management will value you as a "self-starter" and "independent thinker".

Be quirky enough to seem creative to management, but synch up perfectly with whatever process is already in place no matter how bizarre or soul-draining. If you rock the boat too much, you -- not the process -- will be considered the problem, as it's easier and cheaper to fire you than it is to implement real change.

If all these things seem self-contradictory, welcome to the mindset of Corporate fucking America. You should see the correct answers to the Unicru tests which are actually used as employee screening for low-level retail positions.

This is the best advice I've seen on this topic: http://dilbertblog.typepad.com/the_dilbert_blog/2007/07/care...

The advice I see everywhere is to do both. Be a "T shaped" individual who has a shallow knowledge over a broad range and a deep knowledge over a narrow range.

It'll depend upon what everyone else is doing, too. When everyone else is in "spray and pray" mode, it behooves you to go deep and build expertise, because then the depth of your knowledge really makes you stand out in the crowd. When everybody else is going deep and specializing, it behooves you to explore a lot of shallow ideas, because there is probably some low-hanging fruit that someone else is missing.

Which mode are you in now? If it's one of these two, what are you learning?

Everybody is saying both / be T-shaped. There is no free lunch though. You can't have it both ways. If you choose both that means you are going less deep or less wide than you could have gone had you picked one. That may seem obvious but from my experience it is not. Closely related to how we engineers often give unrealistic, overly optimistic work time estimates . Beware!

The T-shape metaphor is imperfect. In reality, going deep doesn't mean digging down into an isolated vertical of knowledge. Knowledge is a graph. To gain more knowledge in one topic, you've got to traverse the graph, following edges most likely to lead you to a better understanding of the topic. Inevitably, during this traversal, you'll encounter many adjacent topics. The deeper you go, the more adjacent topics you'll encounter. Some amount of breadth comes as a natural side-effect of depth. There are few topics where chasing depth will lead you into a totally isolated subgraph for the simple reason that topics that don't relate to other topics tend to be less useful to study.

But then -- to bayonetz point -- are we really giving any negative advice? Just learn? There are only so many hours in the day. Is there anything we are recommending against?

There are two things we are recommending against.

1. Sinking too far into a rut, getting to the point where you have a narrow specialty (fine) but identify with it and are uncomfortable with even thinking about anything outside it.

2. Trying to do everything. With all due respect to Heinlein, specialisation exists for good reason. There will be zillions of worthwhile things you never learn much about because you can't spare the time, and that's okay.

In other words, being I-shaped is a problem when circumstances change, and being rectangular is a fool's errand; that's the context in which we recommend being T-shaped.

It depends on you, your proclivities and what you want out of life.

The value of the "T" depends. If you have real strong expertise in areas like identity or some security practices, those skills are super valuable and you really need a great deal of expertise to even know that the super deep sub-disciplines are. Unless you chose unwisely, a "T" skill set in most things means you'll have a 95th percentile or higher income with a two income family.

IMO, if you "go deep", and make yourself a world-class expert in something you should be deciding to be a contractor and know who the customers are who care. FTEs with deep expertise get screwed in terms of compensation and screwed worse if the winds change and you have to leave involuntarily -- nobody will believe that you are actually an expert if you were an employee. If you want to be and employee, going deep is a waste of time.

There's an approach called "T-learning": be deep in a single topic, but still maintain the broad outlook.

With some luck, you can turn it into Ш-learning later.

I think it is supremely useful to have a deep understanding of one subject, not necessarily so that you're an expert of that subject, but instead so that you experience the process of working hard to understand something. The challenge of learning something deeply requires more focus, and the focus skills you gain can then be reapplied in the future to learn other topics more quickly.

Learn a lot of things, "specialization is for insects" [1]

The type of person who follows this advice will have no problem going deep on the stuff you really like, that will just happen because you find things you love doing. It's much more likely you'll need the push to broaden your interests, especially after you get used to being good at a few things. You'll end up with both a balance of general and specific knowledge and know more about what you personally can do long term.

Of course, it all depends on what your goals are. Self-improvement? Money (specialization is risk, how much are you willing to take on)? Happiness? Fulfillment? Power?

[1] http://www.elise.com/quotes/heinlein_-_specialization_is_for...

Also, what does "convex" mean in this context?

It doesn't seem to be a common term but I found one post saying a convex investment has a lot of upside and little downside, with concave being the opposite: https://www.bogleheads.org/forum/viewtopic.php?t=15932

So I guess the idea is to be ready to position yourself as an semi-expert in a variety of technologies that haven't gotten big yet, but could.

The answer is both. This is the "T-shaped person" archetype who has broad but shallow knowledge of many areas, which gives them a toehold from which to tackle any problem, and deep knowledge of a few areas, allowing them to work on complex problems.

When creating art one must maintain an awareness of how all components fit together to create a beautiful piece that works as a whole. It's important to know what you don't know even if you have no urgent need to learn those things. Anecdotally, figuring out what I enjoy learning (information theory, social systems) and my life goals (early retirement so I can study independently and spend more time with my family) have helped guide me into machine learning and entrepreneurship. Life's too short to learn everything!

I think it's like any other type of investing: iterate on what you invest in every x days/months/years and make adjustments based on how their value changes. Also, diversify.

I think for many folks, at some level, you are enough of an expert in tech ABC that spending a lot more time on it gives diminishing returns. At some point, investing your time in something that's enough of a departure that might only be a 'jr-level' in that tech might be a good thing, while not completely ignoring your previous skills.

Just my own view on this as someone with the 'luxury' of already being 'expert level' enough in certain tech to be able to look at new things.

I don't know about a study, but I figure if you're a real expert in something—especially something less trendy than the suggested topics—your job expectations can become mostly decorrelated from the broader job market. It's like the job equivalent of a hedge fund. What that actually looks like depends on the field your specialize in and the sort of connections you can build but, hey, in the worst case you can always revert to a generic developer job.

My skill/specialization visualization is the paint drip model (generalization of T). You're adding a little of this skill, a little of that, one catches your eye and you dive deep for a while, then a little of this, a little of that.

I assumed that readers had one strong skill already, hence the emphasis on broadening.

From experience - you will naturally "go deep" on something you find really interests you. But it's still good to be aware of various technologies

Reply on news.ycombinator.com