Hacker News Reader

Comment by Animats | Hacker News Reader

Something that's puzzled me for a while: why haven't we hit "peak office" yet? Why are there still so many people working in offices? By now, we should have many abandoned office blocks, much as we have abandoned factories in the Rust Belt. But that's not happening. We have all this office automation, but too many people in offices.

Are marketing cost and G&A (general and administrative) increasing as a fraction of cost of goods/services sold? Perhaps it's the overhead of capitalism. Advertising just moves demand around; where the savings rate is low, it can't create it. There are a surprising number of products, from movies to telephone service, where the marketing cost exceeds the cost of providing the product or service.

One problem with a winner-take-all economy is that the cost of competition goes up, because being slightly better at the margin becomes more expensive. It's like sports; the additional input required for 1% more performance is a lot more than 1%, once you're close to the record.

View on HN

> We have all this office automation, but too many people in offices.

I can't speak to the general case, but to my experience with big corporations/enterprises.

Automation is rarely implemented, and when it is it's half-assed. And then a year or two later they try again, but the old system remains, and now your workers have to connect two systems manually. And then a year or two later...

Not my domain, I'm busy enough, but in one area of my org there are a ton of people whose job is, at its core, to manually synchronize about six databases. They also interface with customers, so that part can't or shouldn't go away. But they spend 80% of their time (estimate based on friends who worked there) doing that manual sync. Every time someone had the same thoughts as me but the power to do something tried to automate it, they ended up with an extra database that was only partially connected to the rest.

Too many stakeholders, not enough leadership.

Maybe it takes an upstart to displace an incumbent rather than the incumbent reinventing itself. So for example you have Amazon vs Wal*Mart, Tesla vs GM... Now, some incumbets do have enough time to reconfigure (as WalMart is trying).

On the other hand... maybe corporations are better corporate citizens than we give them credit for (i.e. not everyone is a "Chainsaw" Al Dunlap. who want to shed workers as quickly as possible. Maybe some actually want to keep people employed as long as they can.

Maybe people simply have different goals. For instance, it seems pretty obvious to me that no mass-layoffs (and thus a limit on productivity increases in that company) was a condition of the Obama administration's helping out of GM, and the same goes for the various Euro governments who did the same.

https://en.wikipedia.org/wiki/General_Motors_Chapter_11_reor...

> there are a ton of people whose job is, at its core, to manually synchronize about six databases

There was an insightful article that I can't find for the life of me that claimed this sort of thing was the future of much human work - doing drudge work on the margins between automated systems, that's not worth the cost of automating.

That sounds like people are trying to make a new system that does everything. Why is nobody making a system that leaves existing components the same, and just does the syncing? One task at a time, in a way that physically cannot cause you to have more partially-connected databases.

A: It's more productive, and easier to supervise employees.

75% of knowledge worker telecommuters that I've run across who aren't doing tracked tasks are doing very little.

We do have a lot of empty office blocks, especially in midsized cities. Even in NYC, there are plenty of mid rise buildings that are minimally occupied, as automation ate up the armies of clerks who once inhabited them.

I would guess many people see "office work" as a form of UBI, where people self-regulate how much they work. Office blocks may not be physically empty, but the amount of work getting done appears to vary drastically. Lot of (logical) places to hide in large enterprises.

That's a good point. But how do you track non-task work which is mostly creative work, be that in the office or remotely?

In my experience as both and employee and employer, you can't. Even in the office it's not like you can tell people "come up with good solutions". It all turns into make believe work.

As an employer we simply don't hire people anymore who need that type of supervision. As a result our possible talent pool expanded from our city to the entire world.

If nothing else, it's because people like to separate their living space from their working space. For reasons ranging from Easier To Get Into Work Mode, to No Screaming Toddlers.

It turns out renting office space or a big enough apartment/house to have an extra office is expensive. Wouldn't it be nice if somebody else paid for that? Perhaps somebody who also pays you and can then recuperate this cost with scale by putting multiple people together in the same area.

If everyone had their own home office, how much would salary have to increase?

Take San Francisco for example. Renting a 2-bedroom apartment instead of a 1-bedroom would increase my and my girlfriend's rent by about $1,000 per month. This would give one of us a good home office. The startup I work for pays about $10,000 to rent a big-ish office space. There are 15 of us working in it pretty comfortably.

That's a lot cheaper than everyone paying an extra $1,000.

I might be misunderstanding your parent's comment but I think the question wasn't, "why are people still doing their work from offices?" and more, "what are all those people in offices still working on?" There's a lot of work that feels like it should have been automated out of existence and yet there are still lots of people working on it.

This always comes up, so I'll just leave this here:

http://strikemag.org/bullshit-jobs/

Nothing new. The protagonist of the 1961 novel Revolutionary Road[1] by Richard Yates works a job that's mostly just figuring out ways to send stuff from his INBOX to other people with the right kind of notes on it to either delay its coming back for another day or two, or, ideally, result in someone else (many, many note-added forwards later) having to deal with it, and to look busy the rest of (most of) the time. Most of the people in his company apparently work this way. Much of the work itself seems to be generated by people trying to find something to do so they seem busy, rather than for any useful purpose. This doesn't come off as parody, either, but as depicting a situation very close to common reality for office workers at that time, with little or no exaggeration.

(incidentally, the novel is great, still reads fresh, and is not primarily about work life in the 1950s/60s)

[1] https://en.wikipedia.org/wiki/Revolutionary_Road

Sounds like the military bureaucracy in "Catch 22."

I think the idea is that if people didn't have to come to the same office, they could live where it's cheap; thus, they wouldn't have to get that bigger apartment in an expensive area.

People really don't communicate as well when they aren't collocated.

that's an uninteresting opinion. I've been communicating solely with people for years, from home, using chat and sometimes video, but usually just text.

I don't forget what someone just told me.

I have a trail of requests and ideas from clients and coworkers.

if necessary I will jump on a phone call for faster data transfer.

the office is great for talking to coworkers, which is almost always about non work stuff, vs when I'm at home working and all my conversations are about work.

I rarely get interrupted when I'm in the zone unless it's an actual emergency.

I could go on.

Similarly, in my office too many people try to communicate tasks and plans to me verbally. I've had to tell them to send me emails (or put in a JIRA entry or whatever).

I cannot incorporate something into my task list that's undocumented, and text (email or JIRA preferred, chat is sufficient for small things but can't be a priority item) is a fantastic way to ensure that it makes it into the queue.

Indeed. Hell, I apply it to life in general (and I wish people would do less talking and more e-mailing). I often get friends and family make random requests in the middle of me doing something, and at some point I started telling everyone - if you want something important, write me a goddamn e-mail. Otherwise there's a good chance I forget.

... aand since people think it's weird and antisocial to communicate via e-mail, I end up abusing org-capture. I.e. I have a quick hotkey (C-c c t) to input and save a task directly to the self-organizing system I use, so if people absolutely refuse to send e-mail, they get an "uh, ok, thanks; I've noted it down and will deal with this later" from me, and I don't spend even 5 seconds thinking about it until the time is right.

You could go on, but you're never going to explain how your communication has actually worsened because you are unaware of it.

Hell, I could give you a good dozen examples of interesting things that can't be communicated properly over long distance, but I'm never going to prove that here, am I? You're never going to understand how confident or uncertain I am. You're never going to grasp an idea that isn't already weighed down by the bias of my word choices and take it with it's natural abstract flexibility and apply it on your own terms.

So I won't take you at your word.

well that's true, I'm a little fishbowled because I work in devops, so my conversations are constrained to technical matters which don't require (if your team is any good) trust in anything but the code you review and accept.

I restrict conversations about jobs, raises and things that require convincing to face to face encounters, but those are intermittent and not a daily or even weekly occurrence.

since the majority of my conversations are: hey, X needs to be done on Y, or build a mesos cluster on Z, or we're having intermittent connection issues on demo. It makes sense to have everything written down for reference, jira helps too, but getting the business side to create tickets is harder than it looks.

overall maybe it's more of a role based issue.

I agree with everything you just said. Some items should only be in person, but the majority actually work better in some written form. Having an instant log is great for basically any work conversation. JIRA ticket comments and Slack come to mind.

I also find conversations more of a lazy way to communicate. Most just talk and never really think. Forcing someone to validate their idea(s) to the point where they are writing them down helps everyone involved.

Finally, written down helps avoid the circular conversations and keeps others who were not present up to speed. Being able to go off and think and come back to exactly where the written conversation left off is a big win. Contrast this with the often endless meetings where the same things are discussed for a majority of the start of the meeting and at the end 1 extra decision is made.

One thing is that talking about "restrategizing our value proposition to better synergize our resource allocation" doesn't look so impressive or meaningful when it's written down. You can get away with a lot more of bullshit when you're talking face-to-face. It's one of the reason why I prefer text.

To drop a data point - as a developer, there's very little in my job that would be done better by personal meeting that isn't handled better via asynchronous communication. Be it IM or e-mail, or whatever. E-mail preferably, because IMs seem to give people the impression they don't have to think before writing. As for things that do require personal meetings - I'd say some initial architectural discussions, when you really want to be together in front of a large whiteboard. That however could be moved to software too, if someone someday writes a non-utterly-shitty collaboration program.

A command-and-control structure is indeed possible even via morse code.

But some office jobs actually involve creativity which seriously profits from being in the same room from time to time,

I've worked at a few companies where inter-office communication and remote-worker communication was excellent. But that's not the norm. People tend to make decisions, especially when they're rushed, with those closest to them.

> Are marketing cost and G&A (general and administrative) increasing as a fraction of cost of goods/services sold? Perhaps it's the overhead of capitalism.

Marketing has a special feature - in a competitive market, it becomes a zero-sum game. You're putting an ever-increasing amount of money into it, but getting only marginal gains out of it. It's a coordination problem. As the industry grows, you end up with chains - ad companies selling tools and services to other ad companies, that bundle those tools and sell services to other companies, which sell their services to companies that use those services to offset each other's use of those services... I think that's why the ad industry is growing so much. There's a lot of space for bullshit jobs there. One would say, almost unlimited.

Hm. This is an argument for not allowing businesses to deduct marketing and advertising expenses.

Except advertising is a stable share of GDP over time. It just shifts channels.

Constant share of GDP for a constant value add is bad. It means there have been no productivity gains. (Which is the main point of the productivity puzzle.) Or more likely, the productivity gains have been eaten up by need to increase the amount of ads in a zero-sum market for attention.

My hypothesis: the optimal amount of advertising spend on a big-picture basis is far less than the actual amount, maybe by an order of magnitude - but a competitive market economy won't get there without some heavy disincentives applied by the state or the public (no tax deductions; "centralized" neutral marketing by journals, blogs, Amazon reviews and clubs like Consumer Reports or Stiftung Warentest; heavy penalties for astroturfing/perverting such neutral reviews etc.).

For the majority of the age of advertising -- basically since newspapers came along -- nobody knew how effective any given advertisement was.

Now we have actual feedback, and the results are tremendously disappointing to everyone except two categories: companies who are completely unknown, and companies who are part of a dominant 2-10 vendor market.

If you are completely unknown, any advertising improves your situation.

If you are a market leader battling for the top few slots in a giant consumer marketplace, huge spending can change consumer perception a little. Getting 2% more sales at the expense of your competitors may be worthwhile.

But the bread-and-butter of advertising was everybody in between those two extremes, and the actual effectiveness is now known to be basically zero.

Is that a "why is there so much office work" question, or "why are there so many people travelling to offices"?

On the former: perhaps the Jevons Paradox at work, or something alligned with it, Baumol's Cost Disease and Amdahl's Law both come to mind.

The Jevons Paradox notes that as a thing becomes cheaper, the total amount of it demanded can increase. A particularly noteable case of this is the cotton gin ("cotton-engine"), which roughly halved or more the cost of removing seeds and chaff from cotton, or alternately, doubled the labour productivity associated with it. A result was a tremendous increase in labour demand, which in Antebellum Dixie meant an increased demand for slaves. The US imported 80,000 slaves from 1790-1809, when imports were banned, then proceeded to increase the number through breeding through the Civil War.

In the case of office jobs -- decreased costs mean an increased demand.

Baumol's Cost Disease has a few elements to it, one is noting that some labour cannot be markedly improved in efficiency. Where physical production and manufacture becomes increasingly efficient, you'll relatively increase the amount of labour elsewhere within the economy, by definition in less-productive roles.

Amdahl's Law similarly notes that in any computer process, parallelisation will serve to increase the significance of the nonparallelisable portion of processing -- if a task is 90% parallelisable and 10% not (by time), then with a tenfold increase in parallelisation will result in roughly 50% of the task being nonparallelisable. The 10% of time is not fungible through parellelisation.

Another factor is being pushed to limits through competition. If other firms are extracting all possible value through increased office-and-administrative tasks -- marketing, HR, and other efficiencies, then companies which don't pursue these ends are at a disadvantage. It's not that there are more gains to be had, but there are the only gains to be had, and you'll lose by not pursuing them.

It's somewhat related (though I'm explaining it poorly) to the tendencies of those under extreme duress to take actions which seem otherwise irrational. Investors will buy high and sell low because they have an increased liquidity need when market prices are low. Dust bowl farmers would increase their farming intensity in the presence of both drought and low market prices because they had fixed debt obligations they had to meet to remain solvent.

(Gordon mentions the investor case in Rise and Fall.)

Re "Jevons Paradox". That doesn't mean employment goes up. There's a little book, "Chapters on Machinery and Labor", from a century ago which outlines three ways this worked out.

The "good case" was mechanized typesetting, the Linotype. This produced a huge increase in print shop productivity and a huge increase in the amount of printed material. The result was an increase in printing employment (That lasted until the 1980s; printing today is almost dead as a trade, and print shops have few people in them.)

The "medium case" was glassblowing. The introduction of automatic bottle-making machines did automatically what used to take a skilled team of five men working closely together. Bottle consumption went up sharply, and but employment in bottlemaking did not. The skill level went down, since feeding a bottle-making machine isn't that hard, so wages dropped.

The "bad case" was the stone planer. This was a machine to make stone lintels for the tops of doors and windows in brick construction. The stone planer made it much cheaper to do this. But lintel consumption didn't go up, because lintels are a small fraction of building cost. Lots of men with hammers and chisels become unemployed.

Good points, and it's complicated.

What the Jevons Paradox meansm more specifically, is that the cost of a good or service is reduced. Taking costs as opportunity costs (Krugman, Economics, 2008, "All costs are opportunity costs."), you're making a thing cheaper.

The response depends on what that thing is and what the relative demand is.

Printing (and communications/information in general) is, well, an opportunity for minting money. The capacity to spew more crap is pretty much endless (hence: the modern WWW). Reducing cost constraints will increase output.

The glassblowing case, and skill requirements, are one where you're probably seeing interactions with a larger, oversupplied, labour market. Offer a lower-skilled position, avail a larger pool, and cut labour costs whilst increasing output.

Stone planers meet an intrinsically limited demand, unless you can suddenly increase the demand for the product. Creating an interest in stone statuary, or stone walls, or something of the sort might work. Pet rocks, Beanie Babies, Bratz Dolls, iPods.

That's a marketing problem....

I've been kicking around other areas and interactions of reduced costs.

Reduced travel costs (railroads) made holidays and holiday resorts possible, as well as commuter suburbs. Automobiles increased these trends, and created back-seat sex (mobility, privacy, opportunity, virility...).

Low-to-zero-cost email created spam.

Low-to-zero-cost webhosting, plus copious data and ad networks: weaponised viral clickbait.

Low-cost long-distance service: robocalls, Carol from Account Services, and Microsoft Support calls.

Increased travel almost always introduces more disease vectors -- from the Silk Road to Roman highways to Columbus and the Age of Discovery cross-polinating old and new world diseases. James Burke of Connections suggested that the biggest legacy of the jet airliner age might be in epidemics. He made that observation in the mid-1990s, with hindsight on the AIDS epidemic (one "patient zero" was a flight attendent).

I'm trying to come up with a good general term for these effects -- along with pollution and systemic problems (crime, fraud, corruption, public health crises), I'm leaning toward "hygiene factors" -- things that can reduce your capabilities. Open to suggestions though.

Very sadly, that little book wasn't published a century ago, but in 1926. Two years after the 1924 copyright blackhole cutoff.

https://www.worldcat.org/title/chapters-on-machinery-and-lab...

TIA finds no matches: https://archive.org/search.php?query=Chapters+on+Machinery+a...

SciHubbed review http://www.journals.uchicago.edu.sci-hub.bz/doi/10.1086/2539...

I'll have to get and scan that.

Meetings. People love meetings. It is much easier to have face to face meetings if you are in the same area. Of course the meetings are mostly pointless, but people still love to have them. Even in an environment where you can do 90 - 100 percent is your work remotely they still insist on you coming into an office. Even though slack or Skype can handle most meetings.

s/people/management. As someone whose job is to produce things, I absolutely hate meetings. And probably so does most of people here on HN.

You might argue that we did: when folks actually had offices. Now the cram-em-in open floor-plan approach is sort of the "now we use less office per person" era.

Reply on news.ycombinator.com