Hacker News Reader

Lol no, then the estimators learn to pad their estimates in order to maintain some semblance of an orderly process and management just ends up managing having completely screwed up their scheduling measurements as a predictor of anything. So that practice leads to management in the dark about wether a project is actually at risk for completing on time.

Edit: if you want trust and fast quality execution you can help the team develop best case, median, and worst case timelines.

It's disgusting how people drive away from truth and it causes other people to have to make wild corrections to manage the lies. It snowballs into meaninglessness.

I once gave a detailed estimate of a project I thought would take 2-6 months. I called out the major blockers, explained the highest variability parts of the project, and made suggestions about how we could reduce variability.

Rather than address any of the points I made, the managers focused on nailing me down to a single estimate. Presumably they had to sell it as a quarter-long project.

I sat in the office of the engineering manager, and explained that any single estimate would be a lie. He told me that it wasn't a lie, it's just my best guess. I said my best guess is 2-6 months. He asked for 3 months. I said he can use whatever he wants, but my estimate is the same. I think they ended up using 3 months.

The ordeal complete drained my energy and the project went no where. My future estimates went through the roof and they eventually stopped asking. Now projects don't have any believable plans.

In such case I would recommend to under promise and over deliver. Your estimate conveys that it's risky selling it as a 3 months project and risk free as 6 months. It's in your best interest to sell it as 6 months and have a chance to complete it in 3. You could bargain 5 months by taking few risk. Nonetheless a good manager should have understood your estimation range and act accordingly.

I think that's a good approach as well.

I don't want to belabor the point, but I had been ignored on previous project estimations as well. Management was going to use whatever estimate they wanted.

My focus was on the blockers and risks.

There are many more issues with deliberately misleading employees:

- Some workers will see through the lies and adjust while others won't. The general version of this problem is that there are many lies but only one truth, and lies don't have any self-correcting mechanism built in, resulting in more variance, and in extreme cases, chaos.

- If you mislead people into always-on crisis mode, they will get desensitized and won't trust you, which is especially important when there is a real crisis. The general version of this is crying wolf.

The fact that a cultural phenomenon is irrational and self destructive is perfectly compatible with success, because of noise. Cultural norms do not converge towards optimum quickly, there are too many other factors at play, and social evolution is veeeeery slow. Evolution is probably the most misapplied theory ever proposed.

But at what cost? Don’t get me wrong. I fully believe that what you described here is the rationale used by most management types to make decisions.

But the thing that gets lost is the overhead and time spent on the quality problems created by this model.

I often wonder how much more profit could be realized if teams remained focused on delivering quality and predictable output instead of inevitably evolving into firefighters who can’t innovate because all of the past shortcuts and aggressive timelines have made that impossible.

If I could summarize all of this, I would pose it as a simple question: Has anyone proven that the current model actually delivers better profits than the alternative? We hypothesize that it does, because money is being made, but rarely if ever take a step back to confirm that this is true.

Not long ago, most employers also felt strongly that employees must be in an office to be productive.

It's not actually about profits. Each worker is trying to maximize their status while minimizing risk.

I'd venture to guess most software teams could be replaced with a team of talented engineers 1/4th their size. However, that will never happen. Why? Shrinking your team psychologically signals to your boss that you're less important. Also, you want extra redundancy because that makes each team member more replaceable.

You also don't want to accelerate development too much, otherwise you might end up with a pace that is unsustainable. This risks missed deadlines. It's better to keep engineers occupied with firefighting and rally the troops only if you need a burst of productivity.

I'm not saying I support this but this is what I've figured out in my time as a software developer on a wide variety of teams.

It's wild to me that this is the case, but my observation is the same. So many people are employed just to keep team sizes up and prop up dead projects.

I spent all 2021 working on a project with 10 other developers plus management. Another team was building a replacement at the same time. The existing project was a non-starter for technical and political reasons.

The existing project didn't have many real deliverables. The majority of the work was refactoring that would only pay off if we had years of usage of the original product.

Maybe it makes sense that it was better to throw literally millions of dollars at that project to keep the developers busy until the new project was ready to scale up.

If they don't spend money then their budget shrinks and their empire shrinks.

It's a total waste of money but it doesn't matter as long as they have an excuse that works for their boss. It's all about plausible deniability.

Plausible deniability is also why higher-ups will buy expensive, bloated, enterprise software/hardware. If something goes wrong they can at least claim they bought "industry standard software". Which is something that can be fed to non-technical C-level execs and board members.

Have you read Bullshit Jobs? That team size correlates to organizational power is one of the central tenets of the book.

Prioritizing profits (capital) over well-being is probably a very strong reason in favor of the labor movement. I am glad that Starbucks is unionizing: more unions for the service industry is desperately required.

It's profitable to drastically overwork people, such that you cull the bottom 10% and get fresh blood in the revolving door.

Indeed, supermarkets, hospitals, meat packing plants and other key parts of production have run on this. The covid crisis has meant that door has stopped being as willing to revolve. But still nothing has changed 'cause this society runs on profits and these places, using this strategy, wind-up being the key profit centers.

> Several bosses have explicitly told me they intentionally set an unrealistically aggressive delivery date because of the phenomenon of people taking as much as time as they're given.

That's because even the "non-aggressive, realistic delivery dates" are actually an under-estimation of the amount of work. Practically all deadlines are too soon, and how much you allow to stress out employees parametrizes how much too soon.

Hmm well certainly in education the deadlines are realistic. But it's because they've done the same task before.

That's the key. In software if you want good estimates, get developers to build the same app again and again, throwing away the source code each time.

Read Henry Ford's autobiography. You'd think he'd agree with you. He'd want to overwork his employees to produce more cars. But, actually, he was much more of a perfectionist around all the things that go into producing things efficiently. And over-working employees certainly wasn't efficient, partly because turn over is high and also work place injuries go up. (For example, he introduce a 40 hour working week https://www.thevintagenews.com/2019/01/24/henry-ford/.)

There is certainly a culture of over-working employees in some businesses. But is it really a _universal_ rule of capitalism that overworking people is profitable? Look at Germany. Highly productive economy, very strong labour laws and no (/limited) culture of overworking.

Have you not missed the 'meta' the article is talking about: that it's an Americanism to think that more hours == more product.

"Image 'slow productivity' in a startup" - what? How long did it take the inventor of bluetooth to invent bluetooth? How long did it take scientists to develop the internet? How long did it take the people to develop computers and electronics? People talk about "fast" productivity in start-ups because there's money involved and investors are greedy and no-one wants anything slowly. But, you know what, plenty of good products take 10 years to make or more. https://www.joelonsoftware.com/2001/07/21/good-software-take...

Successful businesses aren't merely defined by being profit maximising. It's fine to create a good business slowly that has a positive net impact on society. That's in some ways better than a business that abuses everyone to make a small circle rich.

If you judge the success of a business merely by it's profitability, you're a fool. There's more to life than profit.

"I can't help but think he's missing the meta we're all playing within: profit is the motive (not human flourishing)." No, we are not all playing within that 'meta'. We reject that. Playing purely for profit is often sick and deranged. Get a life. Experience something more meaningful than just profit. Reject those procrustean views.

Yes, we all need to make enough to get by, and that means making some profit. But no, we don't have to define or constrain ourselves to being trapped in a pure-profit game. That's for the foolish.

Perhaps my tone was lost, but I was describing the typical American tech company, not what I think is right. I'm agree with your values, but I don't think the economy is setup to support those values.

Looks like Henry Ford discovered that profit was affected by overworking. He was still maximizing that function, no?

Sometimes that function includes making the worker happier, but sometimes it is inversely related.

re: startups, my point was that they have near zero chance of succeeding against large incumbents by moving at a leisurely pace. Sure, great technologies can be invented slowly. But once great tech is given VC money, it's a race. That race has created such a lead between the US and every other country in the tech world.

Sure. Yeah, I apologise if I was aggressive. I agree with you that yeah profit maximising has no necessary relationship to the happiness of workers. But, I was surprised when I read Ford's biography that a lot of (aggressive) practices I thought 'hey this is just bosses being bosses', actually might be poor management.

Re, VC money creating a rat-race, I think you're probably right.

Though I think there must remain a slice of the market that's not determined by the rat race. But it could well be the slice of the market that VCs aren't interested in funding because it's not big or fast enough. I guess I would cite something like Linux as an example. It's not really a start-up but by not being fuelled by money, and being open-source, it was able to under-cut and out-compete Microsoft and a bunch of other OSs in the hobbyists -> server -> mobile. But I suppose it became a platform that a bunch of other businesses like Redhat and Android were able to launch off.

But yeah I guess the general trend is profit invites competition, and competition is amoral.

> But once great tech is given VC money, it's a race. That race has created such a lead between the US and every other country in the tech world.

Let's be clear, it's access to capital rather than technology that has made US startups more valuable (which is weird, given that the easiest way to make a startup more valuable is to invest more money at a higher valuation).

If you look at (almost) all of the recent tech giants, it was the capital they got that allowed them to outcompete the rest of the world, not the technology (which is mostly available to the whole world).

It is expensive to constantly hire new employees and waiting for them to get up to speed. Having a stable group of expert employees who know the systems inside out, and know how to work well together, is a huge competitive advantage. Pushing employees beyond steady state is a clear sign of incompetent management.

> Several bosses have explicitly told me they intentionally set an unrealistically aggressive delivery date because of the phenomenon of people taking as much as time as they're given.

Some people call this “phenomenon” planning against the deadline.

I think a lot of the real world occurrences of this phenomenon are actually due to bad incentive structures though. For example if you finish early (or go under budget) on a project you can generally expect the next project to have a tighter deadline/allowance.

A friend of mine was a postal worker for a brief period a couple decades ago, and at that time (in his location at least) he was told to make sure he took the full day to complete his route, because if there was a consistent pattern of finishing early they would lay off some of the workers.

It's not always possible to make sure incentives are well balanced, but managers should at least take a minute to put themselves in the shoes of the employees. If finishing early has only negative consequences, then maybe that could be addressed directly instead of just setting impossibly early deadlines

Shouldn’t you plan to get the job done as quickly as you can while meeting quality expectations?

If I have 10 hours to make a car trip, but it only takes 5, I don’t drive at half the speed limit. Sure, there may be some extra bathroom and snack breaks, but probably not 5 hours worth.

I see people underperforming due to low expectations quite often, possibly following the classic 80/20 rule. I think this “planning against the deadline” behavior is one of the difference-makers between a hungry startup that outperforms, and the legacy player that barely keeps up despite having 10x the staff.

No, you should treat people like adults, set a deadline, and let them execute to that deadline. Many people balance multiple tasks, objectives, and goals and so taking 10 hours for a task that was assigned 10 hours is perfectly reasonable. If you keep planning 10 hour car rides and they continuously come in at 5 hours, you're bad at estimating.

So you're ultimately saying people should be given 5 hours for the car trip, even though giving them 10 would only result in a ~6-6.5 hour trip with healthy bladder relief?

No, I’m saying if you have 10 hours to do something that can be done in 5, but is optimally done in 6.5, take 6.5 hours and use the remaining 3.5 to do something useful. Maybe you’ll go on a sightseeing tour at your destination; maybe you’ll resolve a customer-reported bug you didn’t think you had time for this release.

I see, thanks.