Software That Shapes Workers’ Lives | Hacker News Reader

Software That Shapes Workers’ Lives

newyorker.com

72 pointsjbegley11 comments

Suddenly I want to take this SAP class! I imagine it's something like the worst of Oracle Forms, but nonetheless it's really surprising and cool that the class was one-on-one.

interesting, he says that software like SAP could 'predict/forecast' the spiking demand from customer.

I wonder how this could be done. Because we know how unpredictable of these demand are.

That unpredictability is the reason for a demand planner. They basically act in a similar role to a project manager, except over the demand forecast for a product portfolio. SAP (the ERP system) snakes it's way through the entire company, from sales to transportation to sensors directly on factory production lines. SAP can be configured with various forecasting models that use all that data as input, and the demand planner is a steward over both that model and all the inputs in there. And accounts for fuzzy stuff like unusual spikes anticipated from marketing events, or other macro level market happenings that require manual intervention or adjustments to account for.

All of the hard stats happen within SAP, and the demand planner really just fiddles with the knobs til the output "feels right". While there's likely a team somewhere in the company that is responsible for the statistical models themselves, the demand planners rarely touch them directly or even truly understand them. They just work with whatever is configured in the system for them, and if it doesn't feel right they might fiddle with and override the output until they get to a forecast that makes all the stakeholders happy.

I started my career in this area. Specifically, working on a process improvement team in a company with a top ranked global supply chain, working on validating novel approaches towards demand and supply planning. I'm far from an expert, but got intimately familiar with the underlying systems and processes in this area. If there's anything in particular you'd like to know, feel free to reach out!

I would for sure like to learn more.

1) which gain in efficient use of resource can such a system produce? 1%? More? Less?

2) is there a book or some article about this idea so that I can learn a bit more and think about how to introduce this in my own business?

3) are there other products for this, apart from SAP itself?

1) That's a complex question. The maintenance and curation of the system itself and the data within it is a non-trivial task, which adds overhead. If you don't right-size your process/system to your needs, and maintain it adequately, then it very well could be a net negative on resource efficiency. Plenty of people will blindly follow what the system recommends, even when a basic sanity check would warn you otherwise.

On that note, different companies define resource efficiency differently. You adjust your model according to your priorities. If you're fine with a certain level of out of stock products or long lead times, you can bake that into your model. With the net result (when coupled with supply planning) being that you reduce the amount of working capital tied up in inventory, at risk of reducing sale volume by not having stock when someone requests it. Or you can go the other way and prioritize in-stock status, which will incorporate a buffer into your model to account for enough variability to never be out of stock when an order comes through. But you'll tie up massive amounts of working capital in inventory costs to do so.

2) I don't have any off the top of my head. But some of hte terms to look at are ERP (Enterprise Resource Planning), S&OP (Sales & Operations Planning), demand planning, just in time planning, demand forecasting

3) There are all kinds of systems for different scale businesses and different industries. SAP is the dominant player in large scale supply chains, but there are plenty of smaller players as well both in supply chain and other industries. You also don't necessarily need an external product - at one point in time, I was planning about $300MM in inventory and about $1BN in demand using Excel. We did that manually for almost a year, replicating the entire model and just pulling the raw data from SAP to input into the Excel model, in order to validate a new technique in a variety of our manufacturing plants before dropping ~$100MM+ to have SAP create a custom planning module for us using the methodology. It was a pain to abuse Excel to the extent to use that quantity of data (before the days of PowerQuery and being able to embed an entire SQL database into Excel). But functionally, it worked perfectly.

My email is in my profile - feel free to reach out. I can probably give you a bit more targeted advice if I knew more about your business itself.

Presumably SAP's models encode information about past supply chain behaviors/events, along with any actual data that the customer may have already input.

At least that's what their documentation claims: https://wiki.scn.sap.com/wiki/pages/viewpage.action?pageId=7...

But as the old saying goes, "There's lies, damned lies, and then there's statistics". But I'm sure that this an area where good neural nets could be put to enormous advantage.

The question the author raises with good reason is "Do we want to optimize only for supply chain efficiency?"

BTW, the author is a 'she', Miriam Posner.

There are a lot of low hanging fruits in seasonality and geography, just storing space heaters and toys in the winter and portable AC units in the summer is already a good first step, and ordering them on time to do it. You need quite an involved tool to do that when you have 500,000 SKU in your database.

> Could S.C.M. software include a “workers’-rights” component—a counterpart to PP/DS, incorporating data on working conditions?

That exists, it is called compliance software, and usually serves as an input (of constraints) to the SCM. It might be labor compliance (hours worked, breaks, etc.), it might be food safety compliance (max temperature, max time in storage, etc.), drug compliance, etc.

The article is a bit of a joke, it reads like assigning a humanities professor to report on resource optimization techniques.

Oh, wait, that's exactly what the article is!

But well, it is the New Yorker, no reason to expect something different.

It's no more of a joke than assuming that software developers are going to take ethical considerations into account by default. Categorically dismissing the whole field of humanities seems like the arrogance of ignorance.

Would you please review https://news.ycombinator.com/newsguidelines.html, which ask you not to snark and not to post shallow dismissals here?

Your comment would be fine with just the first paragraph. Smearing the author like that is uncivil, and dissing the publication just adds noise.

To be fair, it was a qualified dismissal (a proof by counter example, essentially) followed by color (or off-color) commentary.

There's probably more polite ways of saying "this is the typical bias of non-stem core commentators on stem core issues". But we can't all to polish our submissions to essay or press release standards, and devaluing negative off-hand remarks will only incentivize positive off-hand remarks, i.e. sycophancy.

Take these as off-hand remarks on the concept of civil discussion. I fully acknowledge your role and experience as the relevant qualified expert.

Reply on news.ycombinator.com