Hacker News Reader

Comment by lisper | Hacker News Reader

> There are some kinds of work that you can't do well without thinking differently from your peers. To be a successful scientist, for example, it's not enough just to be correct. Your ideas have to be both correct and novel.

As someone who once made their living as a scientist, I can tell you from firsthand experience that while this is technically correct, the conclusion PG wishes to draw from this:

> One of the most effective techniques is one practiced unintentionally by most nerds: simply to be less aware what conventional beliefs are.

is very wrong, at least in the domain of scientific research (and probably others as well, but scientific research is the area in which I can speak with some authority). In order to make scientific progress, you have to understand the conventional wisdom. The reason for this is that the conventional wisdom in science is the product of about 350 years (and counting) of hard work by an awful lot of smart people. Those people explored a lot of wrong ideas on the way towards discovering the right ones, and if you are unaware of this history and just strike out on your own thinking that you are so much more brilliant than any other human who has ever walked the planet before you, you are much more likely to go down a well-known dead-end than you are to discover something new. I wasted about ten years of my life learning this lesson the hard way.

Being skeptical of the conventional wisdom can be useful, but keeping yourself intentionally ignorant of it is generally a bad plan.

View on HN

The huge difference between tech startups and science is that in science, the conventional wisdom has recorded justifications and collected evidence that you can look up and understand, while in tech startups the conventional wisdom is advice made up by businessesmen who succeeded once, or sometimes succeeded zero times, and shared over drinks at the local pub. The usefulness gap between scientific knowledge and cultural knowledge is wide, and nowhere wider than business.

Unfortunately the bias against negative results in scientific publications cause a lot of scientific dead-ends to be non-public as well.

A lot starting researchers (such as PhD students) that are left without adequate supervision are likely to reach a dead-end someone previously encountered but did not publish anything about.

I have not been directly involved in academic research, but from what I understood, results gets published, whether they met the goals or not. It is still novel and valuable to publish why this certain approach or idea did not work as expected as long as you followed scientific standards.

I’m sorry you’re being downvoted just because you have a wrong impression of how scientific publishing works.

Nobody is looking for wrong answers, there are simply many uninteresting ways to screw things up. If you are able to publish a wrong result in a worthwhile journal, a lot of right results have to be in it, which I guess makes your work “right”.

I am not sure if there was a missunderstandig:

I know of people who researched topic X with the assumption to find Y. But they did not find Y and rather found out that their setup was insufficient to provide meaningful answers. So they wrote and explained about why this setup did not work for the original question and published it and moved on.

And this makes sense, because they achieved novel information the next student could improve on. So published about a dead end, or not?

Seems like that depends on what kind of conventional startup wisdom you're thinking about. For example, from what I've read about SaaS business plans, the conventional approach is drawn from the experiences and data of hundreds (or thousands?) of SaaS companies. It's not done with scientific rigor, but it's not individual founders making hasty conclusions based on their own outcomes either.

Understanding the conceptual model behind a SaaS business plan, the kind of metrics that people find valuable and concrete techniques that other companies have found effective seems worth doing even if you have your own theory or approach that's explicitly different—for the same reasons lisper mentioned for scientific knowledge, just at a smaller scale.

I single out SaaS not because it's necessarily more systematic than other tech business plans—I suspect that it is, but don't have a real basis for that—but because I've read more about that than other kinds of companies. I'm sure there is a lot of valuable collective knowledge for any sort of product or company you want to build and the advantages you get from learning that should outweigh any priming effect the knowledge has, as long as you are intentional in deviating from the "conventional" path.

But that's collective wisdom for how to do something - not collective wisdom that something can't be done. If the industry had decided that SaaS was impossible, that would be an example of collective wisdom to ignore.

My thoughts went to SaaS startups in particular when reading the essay - most are highly conventional and seem largely a result of VC-like market analysis and application of a well-honed SaaS playbook, not some place where unconventional and novel thinking is deployed and benign ignorance of a market is desirable. The only surprise as an outsider is really the size some markets turn out to be, not the decision to go into the market nor the solution settled upon.

This suggests to me many/most are started by people looking for a relatively conventional experience, just with a bit more appetite for risk. None of the attributes for these positions are required in much greater degree than elsewhere in tech - curiosity, resistance to being told what to think, or fastidiousness to truth. Perhaps it's a testament to YC's success that the choice to found a startup is more accessible to 'conventional thinkers' these days. No value judgment - we should let a thousand flowers bloom! - but the conflation of startups to the above attributes of 'independent thinkers' is perhaps overblown.

This a million times. You could further prove your point by bringing up some specific examples:

- Webvan vs Instacart. Prior to Instacart, the conventional wisdom was that grocery delivery doesn't work as a standalone business, and that hypothesis was based on the Webvan data point.

- GM vs Tesla. GM's failure with their EV1 was frequently used against Tesla in their early stages. But GM's decision to abandon their project had nothing to do with market demand, and was instead entirely a politically driven decision.

- Concorde vs Boom. Building a supersonic commercial airliner was assumed to be a futile exercise for over 50 years, despite massive changes in regulations, advancements in technology, and overall market demand.

There are so many more, but these are the ones that instantly came to my mind.

An unrelated point that might be worth making: two of those startups are from YC, and the third one was funded by a guy with $100m burning a hole in his pocket. So while going against conventional wisdom has its merits, it does seem that the funding might turn out to be more difficult than for more conventional startups. I suppose this has to do with many VC funds' focus on certain investment themes, which are used to raise their funds and serves as a de facto contract with the GPs. One only has to look at the funding history of Airbnb - even USV couldn't wrap their head around the concept, and Sequoia bought into the concept but then struggled to match them with one of their existing investment themes (I forgot how they finally did it, but I recall it was a bit of a stretch).

> Prior to Instacart, the conventional wisdom was that grocery delivery doesn't work

Is Instacart profitable?

We'll find out soon from their S-1. That in and of itself is pretty remarkable for a business idea that used to be so toxic.

For what its worth, in 07 I had Andy Rachleff as an instructor and we did a case study on Webvan, and Andy's conclusion was that if he had to do it over he'd still invest in Webvan because the business was sound, it was just a bit too early.

There's certainly at least as many opinions about business as there are people working in the private sector, and most are probably wrong, but the fundamental workings of business aren't entirely a random number generator. You can use tools of business to determine the viability of a particular approach, at which point you're left with figuring out the timing and execution of that idea.

Still, not as solid by any means as science.

I think this may attest to the value of (apparent) novelty/innovation in PG's sphere. Seeming unconventional and disruptive has better ROI than spending years mining the history of human thought.

He doesn't do science, he funds and markets products. He's competing to capture pieces of a finite pie, not creating knowledge. More broadly, he promotes a zeitgeist which has made him wealthy. I wonder if this has something to do with his efforts to maintain some semblance of "thought leadership", even if it means recycling tired content.

> He's competing to capture pieces of a finite pie

this is an odd way to describe investing. this would more accurately describe the founders themselves, competing for a finite pool of VC dollars without necessarily having a viable product.

Good point. I meant it in the sense that anyone marketing a product is competing for a finite pool of consumer dollars.

FWIW, I agree with the rest of your post. while I enjoy some of PG's posts, he often makes the classic mistake of presuming that expertise in one specific area makes his opinion on other topics relevant.

Sorry but I strongly disagree with your assertion that building a business is a zero-sum game. At the basic level, a farmer who sows a field and harvests a crop is making the world one harvest richer, and no one poorer. Ditto for the carpenter making a table or mechanic fixing a car. PG addressed this long ago: http://paulgraham.com/gap.html

Scientists turn money into knowledge. Entrepreneurs turn knowledge into money. It's a virtuous cycle. Scientists lament that they don't get their "fair share" but it's wrong to blame innovators, better to reevaluate their own economic strategy. What if we had private research institutes turning out cutting edge research? Or vertically integrated ones that commercialized their own discoveries, like Spacex?

> Sorry but I strongly disagree with your assertion that building a business is a zero-sum game.

It isn't either/or. Nearly all businesses do "create value". Nearly all businesses also compete for a limited supply of customers, because nearly all markets can produce more outputs than are needed.

The ratio between the two matters a lot to the character of the company.

I agree with you completely. There are certainly rent-seeking businesses that do not create value (or net value?). And the majority of businesses are competing in (disrupting) existing markets.

Even if they don't increase total consumer spending by offering better products and winning market share, they do increase consumer _wealth_. I'm glad I can buy a 65" 4k TV for as much as my parents paid for a 15" CRT 20 years ago, adjusted for inflation.

That's the median outcome. Then you get outliers like the sewing machine that greatly increased the productivity of vast amounts of workers, making those users much richer and society wealthier because clothes got cheaper.

I find it disingenuous to claim, as the OP did (perhaps unintentionally), that innovators are bad faith actors playing a zero sum game. Surely that's the exception, not the norm. Just as there are bad actors in academia and any other profession.

> I find it disingenuous to claim, as the OP did (perhaps unintentionally), that innovators are bad faith actors playing a zero sum game. Surely that's the exception, not the norm. Just as there are bad actors in academia and any other profession.

I didn't intend to claim bad faith. As you point out there are surely bad actors in any field. More common, I think, is the (most likely subconscious) Lippmann effect: "We are peculiarly inclined to suppress whatever impugns the security of that to which we have given our allegiance."[0]

[0]: https://www.theatlantic.com/magazine/archive/1919/11/the-bas...

>>Even if they don't increase total consumer spending by offering better products and winning market share, they do increase consumer _wealth_. I'm glad I can buy a 65" 4k TV for as much as my parents paid for a 15" CRT 20 years ago, adjusted for inflation.

I think this depends of how are you calculating the cost of things. In this day in age, misery wages and huge negative externalities have more to do with cheap clothes and appliances than the sewing machine, imho. The later mostly contributed on how many of them you can make by the second, but you still need farms of people to work them.

It's not fixed though. You may think the market for cars only has so much demand, but if you can make a car cheaper, demand goes up. This is one of the most fundamental laws of microeconomics (the non bullshit part of economics.)

There are definitely some zero sum games, I'm not disagreeing with you, just pointing out a nuance. The market for eyeballs/attention is zero sum. Netflix famously thinks of its competition as everything outside of Netflix including "real life."

> Netflix famously thinks of its competition as everything outside of Netflix

Is that true or just marketing? Is Netflix seriously competing against every other type of industry? How could it survive if there was nothing but Netflix? Or are they so megalomaniac that they are trying to take over the pharmaceutical industry too? And agriculture? And military? I can't believe.

There are only two ways civilization gets more stuff, and that’s by either by making the current amount of material go twice as far, making twice as many homes and phones, or we mine it out of the earth.

I don't think building a business is zero-sum in principle. Your point about a farmer creating value is well taken. Our present measures of economic value are notoriously short-sighted though (i.e., willing to ignore hard-to-quantify externalities); did the green revolution create value? It vastly improved yield, but if we don't make serious changes in the next 50 years, it will destroy the viability of agriculture. Value creation is rarely unambiguous.

I would also argue that many VC-backed startups are more about finding unexploited niches or catabolizing existing sectors of the economy than creating new material wealth.

I agree with you that current market incentives do not capture externalities like environmental harm. (eg. soil erosion, in the case of a farmer who doesn't rest his fields)

I hope to convince you to hate the game, not the players. So that we can focus our efforts at the root cause. And to take a measured approach: it's not all bad. I'd rather have our current quality of life (in Canada) than to live in a pre-industrial world in the longhouses of the Aboriginal Peoples who used to live where I do. It seems romantic, but I'm sure I'd be a terrible hunter. I choose to believe that human progress will continue to solve the problems we face, in time. Because historically we always have. Let's root for more Elon Musks instead of hating on the average capitalist.

Where have I implied I hate the players? I certainly don't, nor do I think everything is terrible. I'm sorry it came off as such.

> I choose to believe that human progress will continue to solve the problems we face, in time. Because historically we always have.

This is where we disagree, I think. I don't see good evidence for this. I hope I'm wrong and you're right.

I hope so too :) And fair enough, thanks for the the kind discussion, I didn't mean to put words in your mouth.

Likewise! No problem, friend

Value and compensation aren't necessarily coupled. Think of school teachers. It's clearly an incredibly valuable job preparing the next generation of society, but pays a miserable wage before the intervening of unions, and still not great even with that.

But there's more teachers available in the labor market compared to demand than computer programmers, so the latter command a much higher salary. Even though I would say many deliver much less value to society.

Or negative value like Facebook engineers. Sorry, I couldn't resist that cheap shot, and I do believe Facebook is a net negative for society.

I'm not even sure it's true that there are more teachers than programmers. But the pool of money for paying programmers is large and expanding, because we can easily calculate the value they produce, while the monetary value of teaching is harder to measure.

It doesn't help that teaching is presented as a "caring" profession, that you do for the love of your charges. Teachers don't threaten to jump to a different school district for more money, and when they campaign for more money, they're presented as not caring about the students. Nobody expects programmers to have any loyalty, so they get to demand money.

I would love to have schools fight to get the most talented teachers, instead of settling for "any woman who likes kids". (Those "caring" professions are usually ones associated with women.) I'd love to see school systems compete with software companies and research institutions, and to have school districts run by the people who might otherwise run Fortune 500 companies. But that won't happen without money, and until we start deciding to pay for it, we're going to get a lot of mediocre teaching.

> because we can easily calculate the value they produce,

I am not trolling but I would really, really like a citation on this. I hear this sentiment so often but nobody ever produced that calculation. Would be great (or maybe not ;)) for negotiations going forward.

I mean it in the simplistic sense that we can sum up the revenue of the software industry. That's not really a complete calculation but it's clearer than trying to measure the value add of teachers.... even though it's arguably more.

well, the software industry is filled with non-software engineers. and if you go with the revenue angle, what do you do with companies that don't produce "value" (i.e. product doesn't sell)?

It wasn't my intention to claim that it could be calculated exactly. It was my intention to point out the distinction between software engineers, where there's a bottom line somewhere to be counted, and teachers, where the bottom line is effectively impossible to calculate.

This isn't about how much programmers are worth, but about why it is we get away with paying teachers so little while programmers are paid so much: their contribution to a corporate profit. The details are beyond the scope of the post.

"and to have school districts run by the people who might otherwise run Fortune 500 companies"

What bothers me with this "race for the best", is that there is a finite amount of people running fortune 500 companies.

My main problem with the school system is that even good teachers can't teach their full potential with all the outside constraints regulating their classes.

> At the basic level, a farmer who sows a field and harvests a crop is making the world one harvest richer, and no one poorer.

Not quite true; the farmer using that land to grow crops is externalizing an opportunity cost onto those who might use that land for something else, e.g. for the carpenter's workshop or the mechanic's garage. Same with the water consumed by those crops. With these natural and finite resources, there very much is a zero-sum game, since your use of that resource is at everyone else's expense.

This is something Graham entirely misses in his article: the wealth, contrary to his core assertion, does flow from a common source, specifically land. The wealthy, in turn, are such specifically because they happened to be the ones able to control that land at the exclusion of everyone else. This was the basis of Henry George's advocacy for a land value tax.

"But YellowApple," I can already hear, "${ARBITRARY_FAANG}'s wealth ain't tied up in real estate!" Au contraire. Their software is developed in offices that consume land. Their hardware is built in factories that consume land. Their hardware is further constructed from materials extracted from that land. It's land all the way down, and the company externalizes those opportunity costs on everyone else. That is where the injustice in their wealth lies; it is unjust for them to reap the benefits of that land without adequately compensating the rest of society for the opportunity cost the members thereof now bear.

> Same with the water consumed by those crops.

Every drop of water on Earth has been "consumed" many, many times over. It doesn't just get used once and then disappear.

There may be localized shortages of water (e.g., in California), but that doesn't mean that the Earth's water overall gets "used up", in any meaningful sense.

That ain't what I'm saying. Every inch of land on Earth has been "reused" many times over, too. That doesn't change the fact that there's a finite amount of it, and the possession of any quantity of it is at the expense of everyone else needing it. Same goes for water; yes, the water itself doesn't magically disappear, but of the 1,386,000,000 cubic kilometers of water on this planet, if you're storing a trillion of those km³ for yourself in water tanks, that's a lot less that other people can use.

That is: as defined above, consumption ≠ destruction, but rather consumption = possession.

(This ain't even mentioning the distinction between actual potable water v. water contaminated with salt or outright toxins - and the resulting labor and energy costs of converting the latter back into the former. It also ain't even mentioning the labor and energy costs of moving water from where it's abundant to where it's scarce.)

>If you're storing a trillion of those km³ for yourself in water tanks

All the water tanks in the world don't add up to a trillion km³. Not by many orders of magnitude.

> the resulting labor and energy costs of converting the latter back into the former

The sun does that for free, and has done for billions of years.

> All the water tanks in the world don't add up to a trillion km³. Not by many orders of magnitude.

You're missing my point: that hoarding water is at the expense of anyone else who needs water. One milliliter, one gigaliter, doesn't matter; that's still less for everyone else unless and until it is released, and during that time the mere storage of that water externalizes an opportunity cost on everyone else needing that water.

> The sun does that for free, and has done for billions of years.

Right, because the sun magically drops all precipitation into lakes and rivers, and not a drop of it into the oceans. I'm sure the sun has some sort of tracking system that realizes which molecules of water come from where and puts them right back whence they came, right?

> You're missing my point: that hoarding water is at the expense of anyone else who needs water.

Someone with a water tank in (say) Seattle, Washington is not "hoarding water" "at the expense" of someone in Addis Ababa, Ethiopia, because there is no practical way of getting the "hoarded" water from Washington State to Ethiopia. If the Seattleite doesn't "hoard" that water, it's going to run straight into Puget Sound. It's not going to somehow appear at a tap in Addis Ababa.

> Right, because the sun magically drops all precipitation into lakes and rivers, and not a drop of it into the oceans.

Now you're just strawmanning. Of course a lot of it falls on the oceans. But enough of it falls on land that the hydrologic cycle continues, as it has for billions of years.

I think fundamentally you're just wedded to the the idea that water gets "used up" in the same sense that, say, oil, gets used up.

It doesn't.

> Someone with a water tank in (say) Seattle, Washington is not "hoarding water" "at the expense" of someone in Addis Ababa, Ethiopia,

Well it sure ain't contributing to the "hydrologic cycle" if it's in a water tank. That's a tankful less water in that cycle. And multiply that by everyone else storing water, and before you know it that's a noticeable impact globally, including both for Addis Ababa and the rest of Seattle.

This is, mind you, precisely why a lot of municipalities don't take kindly to people collecting rainfall, or to people damming up streams or rivers without doing the necessary due diligence on ecological impact assessment. There are downstream impacts to these seemingly-innocuous things.

> I think fundamentally you're just wedded to the the idea that water gets "used up" in the same sense that, say, oil, gets used up.

Nowhere have I even suggested that to be the case, and yet somehow you think I'm the one strawmanning here. If you're going to deliberately ignore my point and substitute it for one that's obviously false, then why even bother to respond?

Yes. The "pie" (while certainly finite) is nonetheless immensely bigger than it was in even 1980, much less 1880 or 1380.

> Being skeptical of the conventional wisdom can be useful, but keeping yourself intentionally ignorant of it is generally a bad plan.

I can vouch for that.

I write software, but my original training was as an EE. My software development methodology was almost entirely self-taught.

This has been both good and bad. I will sometimes come up with some really interesting orthogonal approaches, but I will also sometimes "reinvent the wheel," with a naive approach.

Experience has taught me to seek out prior art, and established patterns. My creative way may still be the best way, but I've learned that is not always the case.

I think ordering of exposure makes a difference. Being submersed in prior art and then trying to come up with something novel is much harder or less likely than having conviction around a novel idea/experience and then digging into prior art to strengthen your offering.

But doesn't that seem like a strategy for indulging in confirmation bias?

Uh that's entirely wrong. If you don't even know the tools available to the state-of-the-art, then how can you begin to build new things?

For example, in cryptography, if you don't know know about basic things like elliptic curves and hash functions, how are you going to solve any big problem?

Bitcoin is probably an example of this. I really doubt its creator knew about the state of the art in the field, yet he still created the state of the art in the field.

But yes, you are unlikely to do state of the art math without extensive education. However most other fields aren't like that, software engineering included.

As an EEE graduate myself I hear you. But now that I work in pure IT where I write programs for customer's requirements where nobody cares about a rock solid product vs a shiny new one that can ship next week (and I believe that was the context the article was written), and I hate to admit this - can't fault the author here.

And, like most kinds of work these days, science progress is mainly collaborative, rather than the result of one independent thinker.

pg seems predisposed to dichotomous thinking. This could be about his rhetorical style. It's easier to write an essay, or a screenplay, with good guys and bad guys. The world is more like a novel now, with multiple independent processes in shades of gray.

The difference between true genius and your regular old smart person is that it takes about all the brainpower the smart person has to understand something that's pretty complicated, and their thinking then becomes locked in. The effort not only cements their conceptual understanding; it cements their emotional attachment to their understanding. By contrast, the genius has the brainpower to understand a thing, understand alternative explanations just as well, and then innovate beyond all that.

I think this is true in less extreme cases too.

People have a tendency to do more of what works for them. People with more raw intelligence have lower cost (effort) in considering multiple possibilities and keeping them all alive in their heads at once. (I'll assume just for this paragraph that "raw intelligence" is a thing, and means what I want it to mean.) If other, less capable people attempted to do the same thing, they would get hopelessly confused and be unable to decide or act. So for them, it is smarter to stick to conventional wisdom; they'll be more likely to get a positive outcome.

Combine that with the fact that different people have different interests, inclinations, and abilities in different areas, and the result is that the whole dichotomy of "conventional vs unconventional" quickly falls apart.

Do what's right for you, not what's right for pg.

Another option to:

"One of the most effective techniques is one practiced unintentionally by most nerds: simply to be less aware what conventional beliefs are. "

Is to be aware of them, and to be aware that they may not be correct, or they may be correct and there are better and more effective beliefs now (new theories, but they may not be as well proven, yet), and better ways in the future (new theories/frameworks that aren't proven, or can't yet exist due to other lacking ecosystem components like technology or math or thought processes (Like open source back in the day, and then it won).

My response to all this is to consciously be aware of what I know and how old it is, and that there are probably better ways to solve these problems, or simply engineer the problem away (e.g. better password policy vs simply replacing it with SSO with good MFA).

>Being skeptical of the conventional wisdom can be useful, but keeping yourself intentionally ignorant of it is generally a bad plan.

Yes, I agree with this. It follows along the lines of the saying: You must learn the rules before you can break them. With that said, I have seen many 1st year graduate students spend weeks/months going through the literature only to come up with project ideas that are close derivatives of published works. This usually happens because a student put too much focus on absorbing the "conventional wisdom" of the field rather than starting with a question and seeing what research has already been done to address it.

However, there are plenty of scientists who disagree with this sentiment. I had a colleague who would quote Feynman, "I never pay any attention to anything by experts. I calculate everything myself." Maybe that works if you are Feynman, but I seem to recall John Preskill saying that he didn't think this mentality served Feynman well, at least in his later years.

In many domains of science I will agree with you: success in physics is unlikely if you don't have a strong physics background.

But many modern technologies are much less mature. For example, a good general background in software engineering helps, but following state of the art is not critical (and may be counterproductive) for building new products. My 2c.

And yet many of the greatest mathematicians produced their best work before their 30th birthday.

This used to be true for scientists too, 100 or more years ago.

Body of human knowledge is much bigger now-a-days. It takes 4 yrs honors and first 1/2 yrs in PhD to even begin thinking about fundamental research. That number is likely higher for some other fields (like doctors I guess).

>Body of human knowledge is much bigger now-a-days

that's what Einstein' contemporaries could have said as well.

The growing body of human knowledge is increasingly noise, bad or redundant theories, with entire gigantic fields based on wrong assumptions and fueled by groupthink and premature optimization, just like 100 or 500 years ago. Deep specialization is not always helpful and often detrimental for fundamental research as it assumes deep indoctrination into consensus theory of the day.

Thinking about fundamental research can start once one learns the fundamentals which doesn't take more than a few years in a breadth-first search manner. You don't have to cover all the branches of that giant tree, some of the branches don't have a right to exist in the first place and some are incredibly redundant or marginal improvements of the core parent idea.

I think redundancy and noise have definitely increased because more people are involved in research. So the only think that might be harder now vs. 100 years ago is filtering signal from noise. But the internet, and the rise of search engines is actually pretty helpful with that.

There are certain classes of creative work that especially turn this essay on its head: writing a textbook, typesetting a page, creating a film as an homage to a particular time period or genre. To do these things exceptionally well means noticing and meticulously following conventions that nobody realized existed before you did.

Reply on news.ycombinator.com