Hacker News Reader

Comment by enriquto | Hacker News Reader

> What was surprising was the number of institutions which had signed onto the framework - my alma mater included. I'm curious how many of these still have an Elsevier contract.

If you ever tried to dig deep into the internals of a large organization, you will notice that they are perfectly capable of managing a set of self-contradictory rules. It is actually an amazing power. At least, this was my experience.

For a mathematician used to third-excluded logic, it may seem impossible, since you can prove anything from "p and not p". Yet, these organizations manage to not be able to do arbitrary things from contradictory inputs, but somewhat sane things (some of the time).

View on HN

> they are perfectly capable of managing a set of self-contradictory rules

This is caused by managers' and middle-managers' performance metrics being tied to the wrong outcomes.

The top layer of management's target is the desired outcome. The bottom layer's (the worker) target is also clear, it's to do whatever their manager says to do.

In the middle is where it often breaks down; their objectives are short-sighted and while it works out for their own career, it rarely benefits the top level target. At best, it introduces a large inefficiency, waste of resources and unnecessary busywork ("bullshit jobs" becomes relevant here) and at worst it goes completely against the target set at the top level.

Imagine it as an eventually-consistent system. If you don't change anything and give the system enough time it will eventually achieve its goal. The problem is that the "time" we're talking about is measured in years if not decades and during that period the system is stuck in a contradictory state. Endless restructurings and other external factors often reset this "timer" so the system is even more likely to stay in the contradictory state forever (some other "systems" depend on this system staying in the contradictory state), despite theoretically advancing towards the end-goal.

> At best, it introduces a large inefficiency, waste of resources and unnecessary busywork ("bullshit jobs" becomes relevant here)

If the COVID-19 lockdown has taught us anything, it's that probably a full third of our economy is bullshit jobs. We're at best self-perpetuating economic 'nice-to-haves', at worst parasitic rent seekers, probably somewhere in between, whose real role is simply to distribute GDP more widely, while pretending to add value to society.

I suppose that is a third way to deal with automation.

I usually say, there are two ways to deal with productivity increases (from automation and industrialization): you can either attempt to do more as a civilization (build moon bases, increased consumer goods, etc) or you can have everyone work less (3 or 4 day work weeks, 5 hour work days, etc).

I guess you can also just create lots of bullshit jobs so everyone is still "working" 40+ hours a week without actually doing anything.

I believe your third answer is indeed the reason that despite so many things being automated, we're not seeing benefits of that in terms of reduced workloads on people. The way I see it, as long as creating bullshit jobs is easy, the market will force everyone to work 40+ hour a week to survive. And as long as creating bullshit jobs that just shuffle money around is easier than creating socially valuable jobs, we won't see any Moon bases either.

Look at it from another angle: would people work bullshit jobs if they didn't have to? Fix the incentives, pay people a minimum income where they don't have to work, and if you absolutely need labor you're going to have to pay someone who doesn't have to be there if they don't want to.

If people's basic needs were met, I think you'd see a lot of labor shift to work that is valuable, but currently not compensated for.

Also, you have to adjust constraints to deal with legacy systems: the Federal Reserve works to maximize employment through monetary policy. This is suboptimal, when work will expand to consume the time allowed for it [1]. You need to use a one way policy ratchet to ratchet down the hours per week worked as productivity has increased, otherwise workers will never see productivity gains and society will be stuck on the 40 hour labor work week forever.

[1] https://en.wikipedia.org/wiki/Parkinson%27s_law

> You need to use a one way policy ratchet to ratchet down the hours per week worked as productivity has increased, otherwise workers will never see productivity gains and society will be stuck on the 40 hour labor work week forever.

The feasibility of this (in the US) is strongly dependent on health insurance not being tied to having a full-time job (often defined as just under 40 hours a week).

Plenty of employers have demonstrated that they can be organized to use mostly part-time labor, but their incentive to do this has been avoiding the obligation and expense of contributing to health insurance and other benefits, rather than giving employees a productivity dividend.

I agree. Healthcare must no longer be tied to employment.

It's not just a matter of, is creating bullshit jobs easy?

On the one side we have supply constraints on necessities. If there isn't enough good housing for everyone and we're not allowed to build more then people will have to work 40 hours so they can outbid other people on the good housing and not get stuck in the bad housing. We could solve this by building more good housing, but only if we actually do that.

On the other side it's a question of where the surplus goes. You have a company that makes billions of dollars -- far more than it needs to operate. What happens to the rest of the money? If the managers get to use it for empire building, you get bullshit jobs. That happens a lot.

Ideally we'd solve both, but even just one would be progress.

If you don't have to work 40 hours to afford necessities then you're not going to choose a 40 hour bullshit job. You're either going to choose a real job or a job with lower hours so that you can spend more of your time doing something you want to do, which might very well itself be a real job (i.e. starting your own business).

Whereas if we can figure out how to transfer more wealth away from organizations and into the hands of real people, that gets rid of bullshit jobs too, because most of them come from the principal-agent problem and the misalignment between managers and owners/taxpayers. Transferring the money to any person or entity without middle management would be a net improvement.

There's a ton of nice-to-have professions that are not strictly necessary, like musicians and decorative fountain builders. That is a good thing! It's clearly possible to live in a giant grey barrack and eat nothing but soylent but it's not what 99% of the population would prefer. If anything, it would be good for society to make the percentage of not-essential-but-nice-to-have as high as possible.

There's various kinds of "unnecessary work" jobs. Some do really contribute directly to the common welfare: artisans and producers of culture; providers of small conveniences, etc.

There's also a whole lot of administrative, artificial, make-busy-work jobs, where we build complicated systems of bureaucracy to prevent losses that cost orders of magnitude more than the losses they prevent, and that in turn impose layers of bureaucracy on others.

To be fair, nobody knows which parts of a complex bureaucracy are useless and which are important. There is a pretty decent chance that the truly useless parts of an organisation are linked in to complying with regulations.

Most of the parts of the bureaucracy that I think are useless are the parts that are non-negotiable from a legal perspective.

> Most of the parts of the bureaucracy that I think are useless are the parts that are non-negotiable from a legal perspective.

When we look at e.g. higher education, for a place that's particularly guilty of this:

Administrative costs have grown to be approximately equal to instructional costs at universities, from about a quarter of expenditures in 1980.

In essence, our desire to prevent waste and to ensure that money is spent the way we'd like has created an ever-ballooning set of administrative work. It might be better to tolerate a somewhat greater share of instructional and research expenditures being wasted or spent in non-ideal ways.

Any time there's a small bit of waste or money spent in ways we don't approve of, the impulse is to create rules and administrative procedures to prevent it from happening again. But this effectively creates a perpetual overhead for what may have been a one-time, very minor problem.

I dont think the parent comment's bullshit jobs are the same as your definition of nice-to-have jobs, at least assuming parent is talking about David Graeber's use of the term.

Bullshit jobs are ones that could disappear and the impact would unnoticeable or minimal for an organisation's output. Nice-to-have (for society?) professions still have value, otherwise they wouldn't be nice - if all the musicians disappeared people would definitely notice.

I'd also contest the idea that cultural labour is unnecessary. Lots of people in lockdown have depended on all kinds of music, TV, film, etc to maintain their mental health. This seems to go beyond preference, even if not everyone needs the same or as much cultural produce to survive healthily.

It’s roughly the same concept. Those people were hired for a reason, even if it wasn’t a fully rational one. Just as musicians are the quality of life improvers of society at large, so too are code janitors to an IT team.

The idea of bullshit jobs started a few years ago mostly based upon the idea that the people doing themselves felt like they were doing bullshit jobs, and therefore not feeling fullfilled.

Douglas Adams got there in the 80s with his Golgafrincham B Ark storyline.

>Douglas Adams got there in the 80s with his Golgafrincham B Ark storyline.

If COVID-19 has taught us nothing else, it's that (as Adams warned us) "telephone sanitizers" are absolutely essential.

if anything, I'd wager the bullshit jobs make an organization less efficient.

“There is nothing as necessary as the unnecessary” - from the movie Life is Beautiful.

The current losses in public education will be felt on the decades timescale. I think the perspective here on HN is probably shifted due to the tech demographic, where VC and ads have provided a buffer.

Systems have inertia, stuff continues working until it hits a wall. The unemployment systems written in COBOL kept churning until now, for example.

I think the cuts in jobs might give the impression that "we never needed them in the first place," and maybe that's true in some cases, but it's hard to distinguish right now from the effects that take years to surface.

I don't really understand what COVID would tell us about "bullshit jobs." It tells us that restaurants and bars and concerts, etc, aren't "necessary." Duh? Leisure activity has never been necessary, but it's still something we enjoy and value.

The existence of those leisure jobs and the venues they operates certainly adds value to my life.

A lot of businesses have used revenue loss to justify layoffs.

I think it's a little unfair - you will always have jobs that are important in the immediate, short, medium, long terms. Society can function perfectly fine without teachers, engineers, scientists... For a bit. Some teams can function fine with reduced manning, but long term need full or more manning, for training, taking care of admin, ensuring sanity, and doing more than the bare minimum.

Not that I disagree that some jobs are for expensive seat warmers to ensure warm fuzzies

> The top layer of management's target is the desired outcome.

Kinda. Top managers tend to have a much shorter time preference than shareholders. The incentives are aligned only when all shareholders are of the active trading kind (if that ever happens on any company).

Isn't there also an additional misalignment caused by shareholders who hold shares only to speculate on the stock market? Those shareholders don't care about long-term goals of the company, but only about short-term stock performance they can use to jump sell their shares high, and then buy some other shares low.

Short term speculators have no direct influence on the management (other than through share price i.e. the money they promise to give to the long-term shareholders to sell their shares) because they don't hold the shares long enough to vote on the shareholder's meetings - the long- and mid-term shareholders can order the board on what the priority should be or replace the board if it's not aligned with these goals; and there's a class of activist investors who buy shares in order to do just that, to steer the company somewhere else - but in order to do that, you can't be a short-term speculator, you have to buy, vote and wait before selling.

> The top layer of management's target is the desired outcome.

In my working life my experience has been the opposite. I have spent my career in large corporate environments. Often the top management are removed from reality as an example I once had a CIO who thought he could replace a 20 year old database that had several thousand tables and countless stored procedures. First step of such a plan is a POC and guess who gets to implement the POC whilst making sure the application actually bringing in the money keeps working, it is the middle manager. Consultants are brought in but they end up taking up the middle manager's time because middle manager has to explain current application to consultants. Millions later the project is scrapped.

I still have never seen a concrete example of a job which is clearly a bullshit job. Any suggestions?

Has always seemed like a rhetorical device to me. In effect no job is actually bullshit since if it was it would be eliminated. The truest "bullshit" job is one where you are absolutely irreplacable for a single, trivial piece of knowledge. I.E. I'm the only one who can print the records at the end of the year, and otherwise I do (practically) nothing. That's how I've always thought of it, but others may disagree.

Often times it feels like a way for creatives and engineers who make the product to express some (rightful) resentment towards the folks who manage people and business.

Bullshit jobs, worth a listen: https://www.npr.org/2018/08/28/642706138/bs-jobs-how-meaning...

Corporations are not effective and useless positions are not necessary eliminated.

There's also jobs that exist as the result of Moloch traps, e.g. many corporations need to keep a lot of lawyers on staff because their competitors are litigious. If they all agreed to be less litigious, they could all realise a huge net savings. I imagine much of advertising is similar, albeit in a less straightforward way.

I think of as these rent-seeking organizations that don’t provide value and don’t need to exist.

For example, Elsevier has this monopoly on publishing so they need to hire staff to manage everything. Those staff do real jobs- accounting, sales, IT, etc- but there is not value created. So I might call that bullshit.

Layer of layer of real functions and interdependencies of people and groups all costing and resulting in salaries. But if you entirely removed every Elsevier job from existence it could be replaced by scihub with minimal disruption.

Poof all the jobs could disappear. I think of meaningful jobs as those that create some objective value.

Where does scihub get the papers from, in this scenario?

From the researchers.

Totally a tangent, but: the law of the excluded middle ("tertium non datur") doesn't state that "a and not a" is always invalid, it states that "a or not a" is always valid. This is a somewhat important distinction since there are logic systems (intuitionism) where "a or not a" is not a tautology, yet I don't know of any logic that accepts "a and not a".

I'm confused. Or maybe you're confusing AND with OR? How can "A and not A" ever be true, let alone tautologically so? Also see https://en.wikipedia.org/wiki/Law_of_noncontradiction

Quantum states can be A and not A.

Unknown questions are A and not A.

There’s interesting alternative logic systems that try to tackle this.

We get into interesting territory when using English and then “is” as a statement. Most English is assumed to have a certain value because of to be.

English is almost inherently flawed with logic bugs and assumptions.

E-Prime may help with this.

1. https://en.wikipedia.org/wiki/Catu%E1%B9%A3ko%E1%B9%ADi

2. https://en.wikipedia.org/wiki/Jaina_seven-valued_logic

3. https://en.wikipedia.org/wiki/Infinite-valued_logic

Okay but I still don't see how the law of the excluded middle says that "a and not a" is always valid.

Grandparent said that the law doesn't say that "a and not a" is always INvalid. (The law also doesn't say that "a and not a" is sometimes valid. It makes no comment at all, but only claims that "a or not a" is always valid. According to the grandparent. Now I need to lie down for a bit....)

Yes, this.

The point was that you can reject the law of the excluded middle without getting into the absurdity that would result from accepting "a and not a".

Ahh! Now I see I misread your comment, I see what you were trying to say now.

[this was misleading]

Are you sure you're not confusing AND and OR? https://en.wikipedia.org/wiki/Intuitionistic_logic#Relation_...

Literally copying from Wikipedia: The system of classical logic is obtained by adding any one of the following axioms: ϕ ∨ ¬ϕ (Law of the excluded middle), ...

where ∨ means OR, not AND... right?

Yes. In classical logic, p OR !p is always true. You have to add these axioms to get classical logic. Intuitionistic logic doesn't have that. p OR !p is not necessarily true in intuitionistic logic.

> A and not A are not necessarily contradictions

Nobody was saying p OR !p is necessarily true. The argument was about p AND !p. It really seems to me like you're confusing AND and OR...

The intuitionist will accept that "A and not A" cannot be true https://en.wikipedia.org/wiki/Intuitionism#Truth_and_proof

I didn't claim that it could be true, as I did in fact state that I know of no logic that accepts "A or not A". The argument was solely about what the law of the excluded middle is. IOW, law of the excluded middle != law of noncontradiction. Classical logic accepts both of these laws, intuitionism only the second.

edit: looking at Wikipedia does seem to indicate that there might be systems of logic where "A and not A" might not be valid (paraconsistent logic), but I haven't really looked into that, and I don't know how much such systems are used.

> you will notice that they are perfectly capable of managing a set of self-contradictory rules. It is actually an amazing power.

That's how life is, though—a jumble of rules that don't all agree and aren't all consistent. So you get used to juggling them all, making them work together as well as you can, so you can get your job done and move on.

Yes, but to me this (the contradictory nature of a large set of rules in life) is a nuisance, not a feature to be welcomed. Sometimes it is unavoidable or very expensive to clean, but we should discourage adding complex rules contradicting our existing set if possible. My 2c.

Consider the state and intent of a group of humanity as a spin glass, and you can apply a little bit of vaguely applicable analysis.

I've tried for most of my life to internalize that most of the difficulty in life is caused by different people having different viewpoints and goals and conditions. Any endeavour only works as well as the sum of its implementors.

Light behaves as both a wave and a particle. This goes well beyond human eccentricities. The universe does not care that our simple brains want to model it simply.

The wave and particle models are our own. They don't necessarily respect the true nature of reality. A bit like the old anecdote about blindfolded people feeling an elephant.

The dual, contradictory model predicts reality better than a unified one can. A set of contradictory rules might create better outcomes than a set of straightforward rules.

I don't think this is downvote-worthy, even if it is something of a personal opinion. Heck, half of what transpires for discourse on HN technical topics boils down to personal opinion. On days when I am being cynical, I think that the phrase "Best practices say that" is a veiled synonym for "I prefer that", for example.

This thing -- that different rules get layered in inconsistent ways that cannot easily be algorithmized -- is indeed an inconvenient part of adulthood. But it is a subset of an even more difficult problem. Shermer put it this way, that "smart people are very good at defending beliefs which they arrived at by non-smart reasons." Perhaps the most jarring studies proving this happen on split-brain patients -- people who, because of for example seizures, had a medical procedure where their corpus callosum was severed and now their two halves of the brain cannot talk to each other anymore. Some studies of such patients document trying to prompt their right half of the brain to do something, then verbally ask the left half why they are doing that thing. The shocking thing is that they usually do not get some "I don't know" response from the left-brain; they apparently usually get a comprehensive justification of the action complete with all sorts of details except for the crucial one ("I did it because you asked me to"). So like if the behavior was standing up they will explain how they felt uncomfortable about sitting too long and needed to stretch their legs, but the actual reason was that the experimenters asked the other half of their brain to stand up and it felt agreeable to that suggestion.

Justification is in other words something that we generally backtrack to find. And this in many ways kindly resolves the problems of inconsistencies. Because a better way to look at it is that we have various values which our actions can either serve or betray. Those values are not perfectly orthogonal, and thus alien to any sort of conflict: rather, certain actions will invite you to trade off your value of (say) living an ambitious daring life of adventures with the value of (say) being compassionate to others. So if you were to have an affair you will need to evaluate a moderate betrayal of your value of simplicity/honesty and a large betrayal of your compassion (and if you have this value a massive betrayal of your humility and self-vanishing) for a minor benefit to your curiosity and a moderate benefit to your risky ambitiousness. All of these values come in to speak a separate statement over that choice, and the judgment must be made by you-as-judge to weigh whether the benefits are worth the costs. [This particular schema of five values was in fact the result of a large amount of introspection in my more vulnerable years but now I am not sure that I find it correct -- but it is still useful for this discussion I think.] In my case my judgment is simply that I'd lose way more than I'd gain, so cheating is immoral -- and not just immoral "for me" but that in my ideology pretty much everyone will lose way more than they'd gain, so that "all other things being equal" it is immoral to cheat on your spouse. And indeed if these values are objective, as we have some reason to believe they would be -- like, human biology just functions better if we can generate societies, and those societies just function better if they can have values such as these -- then objectively cheating is immoral. So the fact that we can backtrack from our actions to our values and weigh those actions with those values really frees us from having to somehow "orthogonalize" our values into nonoverlapping Kantian precepts that might hopefully never conflict with each other and generate a logical derivation of what we should do, "People should never lie unless they are saving folks from Nazis or they are preserving the surprise of a surprise party or... or... or..."

There's a whole class of formal logic systems that accept "a and not a".

They are called Paraconsistent Logics (https://en.wikipedia.org/wiki/Paraconsistent_logic), which allow contradiction without explosion.

Ideal for mathematically representing the kind of reasoning that bubbles up from contradictory positions and goals as you described, where organisations manage to do sane things anyway, sometimes!

It's easy to make this non-contradictory by making the rules apply only to future contracts, or only when the decision maker wants them to apply.

this guy manages!

Maybe it doesn’t make sense to think about large organizations as monolithic entities. They’re composed of numerous sub entities that may follow different rules. A platoon may be ordered to shoot, but it’s conceivable that not all soldiers will comply. This is not at all contradictory.

I can't help but think of the legal system.

You would think it was rigorous and logical and would have parallels in science in some way, but when you can "exclude evidence" you realize they are different animals.

as to large organizations... Look, it's just Game of Thrones all the way down.

Ruthlessly applying logical rules onto reality tends to create singularities :)

Organizations are composed from many people doing independent decisions and forgetting about old decisions. There is nothing difficult about organization being inconsistent, the difficult leadership task is to make it consistent.

Inconsistency is natural state even when we ignore the fact that all humans are internally inconsistent - including mathematicians.

I mean, the reality of any large institution (or even a single human being) is that there will inevitably exist many contradictory "rules" that must be handled. I'm not sure how it is amazing. It's a fundamental necessity for their existence.

Policy compliance is a constant battle of wills in most large organisation. Enough of a battle in some places that they have compliance staff.

Just because they sign up to do something in a particular way doesn't mean everybody in that org is going to follow those rules.

Just like the humans that make them, then.

> they are perfectly capable of managing a set of self-contradictory rules. It is actually an amazing power.

Not amazing. Rules are for the ruled. The executive is not bound by them.

Reply on news.ycombinator.com