You can automate a lot of things. Sewing isn't one of them.
September 2, 2019 4:11 AM   Subscribe

This is your occasional reminder that robots can't sew clothes, and that every garment you own - yes, even that $5 t-shirt - was stitched together by a person. [Twitter threadview on Threadreader]

Some highlights:
The ubiquity of clothing makes it easy not to think about how the garments got there, where they came from.
Sewing machines need to be operated by skilled people. The variables inherent in sewing -- the quality of the fabric, the type of thread/stitching required, the applications of trims, etc. -- is why commercial clothing production can't be done by robots.
...I’d like to follow this with the occasional reminder that sewing is high skill work. Sewing is generally assumed to be low skill because women do it, and most of the women doing it professionally aren’t white.
...[A] good way of starting to understand the clothing manufacturing process is through NPR's Planet Money T-Shirt project. It's a good introduction to the mass market garment industry.
Related:
Why is it taking ‘sew’ long for automation in the fashion industry? (Medium, 2018)
...garment creation is an industry that has proven impossible to automate. Human intuition and dexterity when it coming to manipulating fabrics is difficult to program efficiently for a variety of reasons:
1. The variables of fabrics are vast. Fabrics have different levels of stretch, thickness and weaves. Fabrics can crease, fabrics can fold and there is often imperfections and pulls that occur.
2. The garments being manufactured change frequently. Two words: Fast Fashion. H&M and Forever21 both get DAILY shipments of new styles. With each new style, an automated sewing machine would need to be programed with a new set of rules.
3. The movements required to fabricate are complex. While sewing you are often pulling or easing the fabric. Pieces of fabric have to be lined up perfectly or panels won’t match, buttons and holes won’t align and even something as simple as a zipper won’t work
Planet Money podcast – Episode 715: The Sewing Robot (NPR, 2016)
Robots can build cars. They can vacuum your house. Robots can even write news articles. But getting a robot to sew clothes has proven surprisingly difficult. It is a task that is still done almost entirely by people sitting at sewing machines--pretty much how it has worked for decades and decades. Building a sewing robot is something of a frontier for automation.
Is This Sewing Robot The Future Of Fashion? (FastCompany, 2017)
[Experts see] numerous obstacles in overhauling current industry trends—namely, that apparel production, especially for fast fashion, requires expensive machinery and timely delivery at a very low price. With such challenging margins, companies are forced to seek out the lowest wages possible and are unlikely to invest in new technology.
The Case for Letting Robots Make Our Clothes (Motherboard/Vice, 2017)
There are technological limitations to most of the automated "sewbots" on the market right now. [...] machines ... can't do the nimble work of seamstresses when it comes to fancy outfits that have frills and layers of lace. And some of the baubles and additions to clothing can't be done by a standardized machine.
Robots would also fall short in a garment factory since most current machines can only handle specific kinds of materials, not the variety of leather and suede and linen that comprise our market... Cutting different materials would require different devices, whether it's cutting with pressurized water or blades.
...there's already technology that could make t-shirts and other simple garments, but we're not close to replacing human garment labor altogether given the diversity of our designs.
Clothes company backs humans over sewing robots (Financial Times, 2018 - summary of article here)
The march of the robots may be slower than feared: at least on the production lines in emerging markets.
The world’s biggest maker of clothes is betting on human workers rather than automation as it seeks to win more contracts from clients such as Marks and Spencer, Uniqlo and H&M.
Crystal Group, which recently raised $490m in an initial public offering in Hong Kong, said sewing robots could not compete on cost with human labour in developing countries.
[Crystal Group CEO] said innovations such as the Sewbot, designed by Softwear Automation of the US, were “interesting” and would cause disruption. But he did not foresee early-stage sewing robots competing with human labour in low-cost countries in the near future.
posted by bitteschoen (83 comments total) 132 users marked this as a favorite
 
What a great post!

I decided to start sewing clothes last year - I'm a quick learner, and pick up most handcrafts quite quickly, but making clothes was a real challenge. It's incredibly complicated, and very hard to hide mistakes, and even when you do everything mostly right it's possible to end up with a garment that looks fine on the rack but the fit is just weirdly tight/loose in the wrong places, so you never want to wear it.

It's an incredibly underappreciated skill.
posted by twirlypen at 4:52 AM on September 2 [37 favorites]


Chinches are that when they automate clothing manufacture, they'll get rid of sewing altogether (or as much as is possible) and just knit the garments out of very fine yarn in one piece, much in the way that 3D printing is easier to automate than assembly of parts. They already have knitting robots, and algorithms that can translate 3D shapes to knitting instructions.
posted by acb at 5:02 AM on September 2 [28 favorites]


Acb, i really really really do not want to live in that version of distophia
posted by thegirlwiththehat at 5:29 AM on September 2 [8 favorites]


Will the robot scold me for having too big a fabric stash?
posted by Melismata at 5:31 AM on September 2 [16 favorites]


Just wait till you hear how they'll make leather clothing so incredibly form-fitting without any seams.
posted by XMLicious at 5:44 AM on September 2 [26 favorites]


There's been a huge amount of VC money dumped into these "oh, that's got to be easy, poor people can do it" automation projects, especially in the last decade, and there's been very little to show for it. We just had another round of laundry-folding robot companies go bust. Uber is entirely built on the fiction that it's easy to build a self-driving car. Fast food companies are currently learning yet again that building a fully automated kitchen is hard. For some reason we're doing these idiotic pokey little wheeled food delivery robots, again.

The common thread is the notion that these jobs are trivial, and thus it was okay to underpay workers in these industries all along. This is an entire chunk of the tech economy driven, not by economic factors, but by guilt.
posted by phooky at 6:13 AM on September 2 [112 favorites]


There’s a reason the garment industry was at the forefront of the US labor movement, at the beginning of the last century.

Remember the Triangle fire, and the Rana Plaza collapse. The young women who do this work today could have a lot of power, just like the ones who did it then. I hope they use it.
posted by nonasuch at 6:16 AM on September 2 [24 favorites]


Ooh, I also picked up sewing as a hobby a few years ago, and now I feel like clothing should be so much more expensive than it is, because it really is so hard and labor intensive. I also didn't fully realize how manual garment sewing was - I think I did have the vague idea that at least some clothes were, you know, extruded from some machine... now I remind people all the time that someone sat down and sewed everything they are wearing.
posted by catcafe at 6:42 AM on September 2 [10 favorites]


I'd like to follow this with the occasional reminder that sewing is high skill work. Sewing is generally assumed to be low skill because women do it, and most of the women doing it professionally aren't white.

This.

The definition of 'low-skilled work' so often revolves more around whether you can get a comparatively-marginalized group to do it for a fraction of what the labor's actually worth rather than the actual skill required.

Hell, even 'flipping burgers' is far more skilled work than most realize (and actually requires much more skills today than it did back when boomers were teens, making it that much more offensive that the boomer-teens often earned more, inflation-adjusted, than modern teens doing the more intensive modern version of the job).
posted by mystyk at 6:47 AM on September 2 [64 favorites]


I think every piece of clothing should come with the name of the person who sewed it for you.

Controversially, I'd like to see also how much that person was paid for it and use that factor in making my purchasing decisions.
posted by vacapinta at 7:06 AM on September 2 [32 favorites]


When crocheted items were a trend several years ago, I -- a crocheter -- stood in a Target looking at the huge amount of crocheted items in absolute wonder and horror. I explained to my wife, grabbing at something nearby, saying, "Look, machines can't do this look, everything here someone did by hand. A thousand someones are sitting somewhere doing every single one of these stitches."
She was surprised. She didn't know. "Surely some machine can just..." but no. It's absolutely overwhelming. Every part of the process is. I also sew and once sat my wife (who I don't mean to pick on, this is just my area of knowledge, she schools me in chemistry or math every day) at the machine and said, "why don't YOU try" after another off-hand "can't you just sew it/mend it/make it yourself?" request. Those requests are made more respectfully now after she realized how hard it is.
It's at the point where many of us don't even question that we SHOULD or COULD or WOULD know how our clothes or food or other essentials are produced. That not long ago most of us had a handle of the processes of the world. You didn't have to work in a shirtwaist factory to know what there were shirtwaist factories, with women sitting in them making shirtwaists. But knowing would make us less effective consumers, I suppose?
posted by wellifyouinsist at 7:11 AM on September 2 [52 favorites]


Robots are really bad at non-rigid objects, because they don't have predictable shapes. Pile a bunch of non-rigid objects together, and it's a multi-year research project just to figure out where one ends and the next begins.

Here's a film of a robot folding a pile of towels. The video is 50x real time, because the robot needs 20 minutes per towel.

Five years later, they got the robot down to 90 seconds per towel.

There's a sci-fi story whose name I've forgotten about a man with a home-care robot who eventually catches his robot zoning out. Turns out the robot is a telepresence device operated by a little girl in South America. I say it's sci-fi, but of course we're already doing it.
posted by meaty shoe puppet at 7:23 AM on September 2 [47 favorites]


Middle Son wondered how hard it could be to mend your own clothes, so, with a mixture of the internet and bugging the staff at the local haberdashers, he taught himself sewing, darning and some other basic skills.

He's since bought a sewing machine and gone on from there to make his own shirts, jackets and trousers, and they're good enough that his fashion-conscious sister has asked him to make her a jacket. While I'm impressed with his ability - and he is very good at things that require manual dexterity, unlike me - I've definitely seen how much work it is (and how much practice it takes to do the basic things like sewing in a straight line)

When my favourite pair of jeans reach end-of-life, I'll ask him to use them as a pattern and make me some more.
posted by 43rdAnd9th at 7:37 AM on September 2 [13 favorites]


When I worked in the garment industry I desperately wanted a robot that could simply pick up a T-shirt from a nice flat stack of them and slide it to the appropriate place and alignment on a screen printing pallette, or one that could manage to pull one off the palette and free it from the temporary adhesive holding it there without stretching the shirt, tearing it, or folding it in any way or the wet ink would smudge.

Or a robot that could simply neatly stack and fold t-shirts into dozens or singles, especially out of a tangled pile of them.

Hell, I would have accepted a robot that could just take a dozen count fold of T-shirts out of a box, unfold that dozen into one flat stack and lay it out flat on a table.

I often thought about these robots while doing the exact menial labor I wanted them to do. The work is brutal so of course I wanted to design myself right the hell out of that job.

Even the basic job of pulling t-shirts from the printing palette consumed both your attention and your labor almost completely.

Pulling a T-shirt from a palette goes like this: Reach out to the far edge of the palette, grab the bottom hem of the shirt at the far end, keeping your own shirt, hands, arms etc away from the uncured, wet ink. Peel the shirt carefully from the adhesive holding it to the palette while your back is fully bridged and arched and you're doing about a 5-15 pound reverse curl pull while fully extended, and, no, you definitely can't let your torso or body touch the shirt or palette at all. (This is GREAT for your back, oh yes!) Make sure it stretches not at all or evenly as you pull it as to not disturb or distort the print. Switch your grip to the near side of the palette inside the collar, peel the rest of the shirt off the adhesive, working inside the collar and shoulders. Now pull the shirt carefully from the palette, taking extreme care not to let it flag, fold, wave or flutter so the ink doesn't ruin itself with self contact. Turn and pivot and drape that printed shirt perfectly flat on a moving conveyor belt.

Now do that in about 1-3 seconds, max. Now do it about 1,500 to 2,000+ times an hour. There's only 3600 seconds in an hour, and that machine needs to hit solid production metrics or our slim pennies-per-shirt profit margin and delivery dates just burned up into elemental ash and blew out the window.

Now do that for a 8 or 16 hour shift, day in, day out. I did this for something like 15+ years, starting working in my dad's shop at something like 10, 12?

I spent a whole lot of time thinking about this. We're talking about thousands of hours spent thinking about this, often while working. In those pre-internet days I was even checking out books on machine vision, factory automation, advanced robotics - all the related topics that would go into this sort of thing. I was pretty much robot-mad as a kid and young adult and really wanted a robot to clean my room. Still do.

And my designs and ideas were not shabby. From the beginning I did not set out to build some sort of stupid humanoid robot. The robots I imagined were very specifically focused and more about things like highly polished metal guide rails capable of slipping into the inside of a t-shirt to hold it taut as though on a frame and allow it to slip a T-shirt on a palette, or specific use grippers and manipulators with similar polished shaped or rods to help peel a shirt off a palette. These specific T-shirt handling robots looked (in my head, or on paper) a lot more like brushed metal and polished chrome scarab beetles and had good fundamental industrial design and automation principles.

The fabric folding robots I envisioned in my head sure looked a lot like the marginally useful fabric folding robots of today found in commercial laundries and dedicated entirely to folding bedding and flat hotel sheets, all vacuum-suction manipulators and flat polished panels or rollers.

Just addressing the machine vision problems alone of how to see, identify and pick up a T-shirt from a nice flat stack of them was utterly incredible. How the hell do you even identify if the hem is the upper/front portion of the hem of the t-shirt and not the back portion? Or belonging to the t-shirt below it? How do you deal with a shirt that has the hem folded or wrinkled, or a corner turned up? How do you deal with the fact that they're all the same color and there's often very little usable contrast?

I found myself circling back to solutions that involved even more human labor. "Well, if we make sure to stack all the shirts flat in the first place." or "Maybe sewing in small RFID tags or optical targets is the answer." and no, no it all ended up being just more human labor to make the job easier... for the robots.

Well, in the end I came to the likely accurate conclusion I would never see a robot reliably handle fabric, sewing, folding or other garment industry related tasks in my lifetime. I'm dumb enough to briefly think I might have been able to do this but smart enough that after I did a lot of homework that I realized I was looking at the task of advancing robotics, automation and machine vision forward by half a century or much more and I definitely wasn't smart enough or well funded enough to even begin to crack that nut.

I'm still convinced that I won't see this in the next 50 years, at least not with traditional cut/sew piecework and garments.
posted by loquacious at 7:55 AM on September 2 [63 favorites]


There is some access to power in this situation as a consumer. Buy secondhand if you can, whenever you can; organize/attend clothing swaps in your area; earn a few basic sewing skills for repairs if you don't already; find a local tailor; learn to crochet; never just throw away wearable clothes. I know this won't solve the problem on its own, but at the very least it avoids enriching the industry.

I'm mega fortunate to live in an area thick with thrift stores* and to be able to find clothes that fit me, and I don't work in an industry that requires me to wear expensive clothes, so I've been able to avoid buying new for almost 18 years. It saves oodles of money, it's fun, and because of survivorship bias, the clothes I buy tend to last forever. I encourage everyone who can, go to thrift stores! They need your patronage. Many of them give directly back to the local community, and by keeping used clothes out of landfills, it helps the environment.

*Shout-out to Diversity Thrift in RVA, provider of like 98% of my clothes since 2001
posted by captain afab at 7:55 AM on September 2 [8 favorites]


3D knitting (like 3D printing) as someone mentioned above is the closest to sewing-automation. Sewing is interesting and I like talking about it. I find people are often interested to hear about techniques and whys/wherefors because they never really thought about it before. I remember explaining seam finishes to someone once, and I just kept talking until I was embarrassed, but they assured me the whole thing had been interesting. Later, they mentioned how they'd started looking at the insides of things when shopping.

It's hard to even make sewing into an assembly line because once you get beyond a really simple t-shirt, it's too complex; there's too much room for error and for things shifting and needing to be re-set. Like on a t-shirt, you can have one person cut*, one person piece, the next person do the shoulder seam, the next person do the side seam (inserting sleeves flat--which you can do for a 2 or 3 piece sleeve), the next person finish the neck, the last finish the hem. But with most garments--although there will be a proscribed best order for construction steps--you can't be handing the garment off like that between steps without error or making the process much slower than if one person just did the whole thing.

And since it's probably better to have one person do the whole thing, their mind is constantly shifting from task to task along with their body. Now I'm thinking about holding the tension between these two pieces of fabric to this standard; next I'll be feeding them through the machine at this standard. It's tiring!

Sewing most of the things you see in a fast fashion store is not complicated--even though you are going through those minor adjustments with almost every step. And getting clothing fit the the close-enough standard that ready-to-wear gets is also not complicated. But it's time-consuming. It can be ergonomically challenging and I've never run across a machine with a needle-threader that was any easier than trying to the thread the needle by hand--and that is exhausting on the eyes.

*Also (undoubtedly mentioned in the articles but still my favorite thing about fast fashion), when you try to stack fabric and cut more than a couple layers of the same pattern piece, the variation in size between top layer and bottom layer is so significant, you are no longer sewing a garment of the same dimensions when you get to the bottom of pile as you were at the top.
posted by crush at 7:58 AM on September 2 [15 favorites]


I've had the singular experience of going to a garment factory in a developing country and buying clothes direct from the factory. It's a trip. I wish more people could experience it. Seeing long lines of women operating sewing machines at lightning speed in the heat all working on the same skirt... It's just really something.

One eye opening thing for me was that so many brands use the same factories to make their clothes. If you buy garments made in, say, Vietnam, from any high end brand, it was assembled in one of a small handful of facilities. There are good reasons Banana Republic costs more than Gap, but "made by more talented/better compensated people" isn't one of them for mass market clothes.

I did this probably ten years ago now, and still, when I'm shopping and see "Made in [country]" on a tag I picture that purple skirt. Nothing about the working conditions seemed inhumane to me, let me be clear. It was hot and humid but the whole country was that. The people working were adults. But it did seem like work that should warrant at least as big a paycheck as I'd expect people working under similar circumstances to get in the States, and I doubted that was the case.

I've been to other factories (toys, light bulbs, candy) and I can attest that garment factories are VASTLY less automated. It's not like a factory you see on Mr Rogers with an assembly line and tons of special purpose robots. The "assembly line" is lines of people with sewing machines doing the same task over and over then handing off to the next line.
posted by potrzebie at 7:59 AM on September 2 [25 favorites]


Used to teach middle school English to prep school boys. They were required to wear button-down shirts as part of their uniform. I used to read them Pinsky's "Shirt" and then invite them examine the ones they were wearing - find out where they were made (everywhere including Malaysia, China, India, and even very occasionally the United States), count the number of pattern pieces that were sewn together (in a yoked shirt with button bands and cuffs there are anywhere from 12 to 17, including all the facing pieces), and consider the number of buttonholes. And then I gave them information about the Triangle Shirtwaist Factory Fire and we talked. It was a whole-period class.

I was always reminded of how very young middle-schoolers are. They tried everything they could think of in their discussion afterwards to make it so that the seamstresses could get out of the building without dying, for instance. They still believe in the rules of cartoon physics.
posted by Peach at 8:03 AM on September 2 [36 favorites]


One thing that makes me so sad about fast fashion is that because the materials are poor the clothing wears out almost immediately (A lot of it is designed to last about 10 wash cycles). But it still took the same amount of human labor to produce as something that could last for years.
posted by velebita at 8:04 AM on September 2 [61 favorites]


It's also not a coincidence that fabric is one of the most difficult things to simulate well in computer graphics or physics simulation engines.

It is incredibly difficult to model, say, a dress or tunic or other flowing garment on a moving humanoid form. Now try doing it with layers of clothing, say a layered skirt or a tunic over a dress or skirt. Now you have two really complicated physics models involving tensors and meshes to approximate fabric stretch and flow, and now they have to interact with each other, with gravity, with wind resistance and air flow, with a breeze and they have to have full surface collision detection, but the layer beneath can't clip through the one above it, ever, otherwise the simulation immediately fails. If you design a free space buffer between the layers of modeled fabric to prevent clip-through, the simulation fails again because it looks totally unnatural and not like, say, two paper-thin layers of silk in immediate contact and flowing together.

Clothing is ontologically fascinating.
posted by loquacious at 8:05 AM on September 2 [23 favorites]


(I'd just like to point out quickly that garment sweatshops aren't always overseas. About a decade ago when we were looking for some light manufacturing space in Brooklyn, one skeezy broker took us to a couple of sweatshops, one just recently closed and another still active. There was also a time in the mid '00s when I'd take the N over the bridge to work, and as it descended into Manhattan in the summer you'd often see open fire-escape doors ventilating rows of people hunched over sewing machines. It's not as far away as you'd think.)
posted by phooky at 8:08 AM on September 2 [14 favorites]


Remember the Triangle fire, and the Rana Plaza collapse. The young women who do this work today could have a lot of power, just like the ones who did it then. I hope they use it.


Bangladesh strikes: thousands of garment workers clash with police over poor pay

Here's a set of headlines from the Solidarity Center about garment worker strikes, conditions, and retaliation against them. I'd argue that garment workers today are trying to use their power, but the global market has changed substantially since 1909 and they don't have nearly the same power in the face of a globalized economy with cheap and "disposable" labor everywhere.
posted by ChuraChura at 8:36 AM on September 2 [18 favorites]


I grew up with a mother who was using Simplicity patterns to make clothes for us. All kinds of clothes. Shirts and dresses and all of it seemed to be terribly complicated and she worked hard at it. I never realized at the time exactly how much work it was, and at this point in my life I have nothing but respect for her for doing what she could to save our family money by putting in time and effort to create clothes for us.

I should call her and tell her that, to be honest. So much admiration for her and her efforts for our family.
posted by hippybear at 8:40 AM on September 2 [13 favorites]


I just did an alteration on a skirt, and I did a fairly crap job; it's not trivial to sew well. I say it all the time - the US has an absurd oversupply of consumer goods, especially clothing. Totally dependent on exploiting people in distressed economies. Her thread is smart and interesting; thanks for posting it.
posted by theora55 at 8:49 AM on September 2 [3 favorites]


I’ve always thought that any dude using his grandmother as an example of an unsophisticated computer user should try to set up and use a sewing machine.

I have a lot of my clothes made for me because they fit me better than just about anything I can find in stores and I have the opportunity to pay someone a fair wage for their labor.

There’s a running joke in the knitting community about friends asking knitters to make shawls and sweaters. People balk when they’re quoted $600 for a hand knit sweater because they’re so used to fast fashion and don’t realize that the yarn itself can run $200-300 if it’s decent quality. Even a fast knitter has to charge a few hundred bucks if they’re going to beat minimum wage.

If someone gives you a hand sewn or knitted garment as a gift it’s because they love you and not because they’re trying to be cheap.
posted by mikesch at 8:49 AM on September 2 [59 favorites]


Man. I had no idea that robots were not used for sewing fast fashion.

I really had no idea. I’m not what i thought sweat shops did but I honestly didn’t think my stupid t-shirt from H&M was sewn by hand. Christ.
posted by sio42 at 8:52 AM on September 2 [21 favorites]


I also didn't fully realize how manual garment sewing was - I think I did have the vague idea that at least some clothes were, you know, extruded from some machine...

Me too. Especially after I started sewing. The stitching on even the cheapest clothes is just so precise and neat. The tension on sleeves insertions and bindings is so perfect that it never calls attention to itself and you take it for granted. The seam allowances are trimmed more exactly and perfectly than I can see myself ever achieving. Definitely not on a regular basis, and the regularity with which the fiddliest work is executed perfectly is what always made me think cheap clothes just had to have been made with at least some help from automation. To be honest, it made me feel better about how imperfect even my good work is.

When you start to sew, you think all those little things like bindings are going to be easy afterthoughts. Possibly because the bindings on any $5 shirt you've ever worn have been perfect. It's incredibly misleading.
posted by trig at 9:12 AM on September 2 [9 favorites]


It's incredibly misleading.

No, it's incredibly horrible that humans are working to make you a $5 shirt at a skill level that you didn't think humans were involved and that they are being paid a wage where they exert that skill and you still only pay $5 for a shirt.

It makes you question every clothing purchase, and that's what it should do.
posted by hippybear at 9:27 AM on September 2 [16 favorites]


Something something cost of a linen coat something
posted by The Whelk at 9:30 AM on September 2 [4 favorites]


If someone gives you a hand sewn or knitted garment as a gift it’s because they love you and not because they’re trying to be cheap.

So you're saying my gran could have sold that Arran cardigan I was forced to wear for all those long childhood years and bought me a Scalextrix?
posted by biffa at 9:42 AM on September 2 [1 favorite]


Ooo! Pattern-matching and directional prints! I'd love to see a robot handle that. It would be scary and impressive and cool. I mean, there's not a whole lot of attention to pattern-matching in fast fashion--certainly not at the level you see it down in a very expensive bespoke plaid jacket--but it would be wild to see a robot that could handle that!

(Pattern-matching is just making sure that the pattern on your fabric lines up when the seams are sewn together)
posted by crush at 9:44 AM on September 2 [5 favorites]


(i am slowly getting better at it)
posted by crush at 9:44 AM on September 2 [2 favorites]


So you're saying my gran could have sold that Arran cardigan

No, the actual cost of creating clothing has always been undervalued and your gran couldn't have sold that cardigan for anything. Be grateful for the effort others put into your clothes, and hug your gran an extra time.
posted by hippybear at 9:45 AM on September 2 [17 favorites]


Pattern-matching and directional prints! I'd love to see a robot handle that.

Funny you should mention it
posted by OverlappingElvis at 10:18 AM on September 2 [5 favorites]


I'm not understanding the logic- we should automate sewing so that garment workers have no jobs? Is having no job better than having a difficult and underpaid one?
posted by bhnyc at 10:46 AM on September 2 [1 favorite]


One thing that makes me so sad about fast fashion is that because the materials are poor the clothing wears out almost immediately (A lot of it is designed to last about 10 wash cycles). But it still took the same amount of human labor to produce as something that could last for years.

This fact definitely has its downsides, but one of the upsides is that it keeps the workers employed when they might not be if there wasn't such constant demand. I have a friend who is a bespoke suit-maker, and that stuff is designed to last a lifetime. He once told me he effectively has two kinds of customer: Some are the clothes-horses who come back once or twice every year because they love clothes; the others assemble a wardrobe over a few years and then stop coming back because they have all the high-end clothes they need and their bespoke stuff isn't wearing out.
posted by slkinsey at 10:50 AM on September 2 [7 favorites]


I'm not understanding the logic- we should automate sewing so that garment workers have no jobs?

The point being made is not that sewing should be automated. It's that many consumers think it already is, and the truth of how a cheap shirt can cost so little has nothing to do with automation and everything to do with vastly underpaid, highly skilled workers.
posted by potrzebie at 10:54 AM on September 2 [58 favorites]


potrzebie wrote exactly the words that I was going to write, so I'll just point up one comment and say "yes, exactly that".
posted by hippybear at 10:56 AM on September 2 [1 favorite]


As much as I appreciate being able to afford clothes, the people who make them are seriously underpaid. I sew well enough to repair my clothes, quilt, and recently I've started making wall hangings, and even with enough skill to get some decent cash for the wall hangings I'm nervous about whether I'm ready to branch into making my own clothes.

One of my ex-girlfriends makes hers and says she'll help, so that's a plus. Her stuff is great, but like people have said above - it's incredibly time-consuming and it's not cheap. She does it significantly because she can't buy the kinds of things she wants, and I've got similar issues. Hers are more about size and mine are more about gender identity, but either way - it's not a trivial thing to do.

Here's the wall hanging I'm currently working on if anyone is curious, the moon and flying saucers are paper in the photo because I was sending mockups to the buyer.
posted by bile and syntax at 11:04 AM on September 2 [16 favorites]


Putting on my socks while reading this post, I thought surely sock manufacturing is fully automated. How hard can it be for machines to make these?

Answer: How It’s Made Socks (SLYT). Although highly automated, it’s surprising how much human thought, time, and effort is involved, especially with the care and feeding of specialized machines.

There’s more man-made magic under our feet than we realize.
posted by cenoxo at 11:09 AM on September 2 [6 favorites]


What's sad is how much we assume things in our daily life that are cheap are done by machines and we don't even know how humans are involved/exploited for our daily clothing comfort.
posted by hippybear at 11:21 AM on September 2 [5 favorites]


I encourage everyone to try sewing once in their lives. My mother taught me to sew, then I gave it up for a long, long time. Came back to it a few years ago when I got disgusted with how ugly ready-to-wear clothes had become and how badly they fit. Yes, it does take time, but sewing is incredibly satisfying, and if you stick with it, you'll end up with clothes that are one-of-a-kind and fit your body. The great thing, too, is there's an amazing amount of body acceptance in the sewing community. I've found it to be a very fulfilling hobby. It can be hard, but you mostly need to be patient with yourself as you learn and willing to make mistakes. You can find a supportive group of sewists online who will help you. Like others have said here, it gives you a great appreciation for all the people who labor to make not only our clothes, but everything we have.
posted by merrill at 11:32 AM on September 2 [1 favorite]


Did I misunderstand that pattern-matching video? It seemed like it still requires an operator to line up the pattern and the seams and then it can robo-cut the fabric.
posted by crush at 11:33 AM on September 2


There is some access to power in this situation as a consumer. Buy secondhand if you can, whenever you can; organize/attend clothing swaps in your area; earn a few basic sewing skills for repairs if you don't already; find a local tailor; learn to crochet; never just throw away wearable clothes. I know this won't solve the problem on its own, but at the very least it avoids enriching the industry.

I'm not clear how this power allows the consumer here in the west to solve a problem. What exactly is the problem? That people are exploited in clothing manufacture? Buying less clothing solves exploiting part, I guess. By decreasing the earnings of those who were exploited. Less exploitation means less money not only for the industry, but for the people it employs. This is a difficult calculus, as clothing is rarely a market driven solely by the utility of the clothing. Fast fashion, whatever you wish to call it, the garment industry is largely driven by fashion demands, across all price ranges and cultures around the globe that are in a position to utilize garments.

I am curious about the skills involved. The pay of the average garment worker isn't really linked to the skill involved in the work. It's probably more a matter of how many people have (or can obtain) that skill, and are willing to trade that skill for a wage. Supply... demand... crazy how these things seem to have a relationship...

While the skill needed may be considerable, the actual work of garment worker likely rarely called upon to create whole garments, instead engaging in very repetitive piece work. Additionally, one might think of sewing as being a highly portable skill, I suspect garment manufacture is likely a highly portable industry, able to move and set up shop quickly with local labor and minimal demand on specialized local infrastructure. Which means that it's a difficult industry for labor to gain leverage against.
posted by 2N2222 at 12:42 PM on September 2


No, it's incredibly horrible that humans are working to make you a $5 shirt at a skill level that you didn't think humans were involved and that they are being paid a wage where they exert that skill and you still only pay $5 for a shirt.

We're in agreement. I was trying to point out the skill level involved - apologies if that wasn't well expressed.

It makes you question every clothing purchase, and that's what it should do.

This is true, but it's also not so simple, as in every debate on individual action versus systemic change. I buy the cheapest shirts because my labor is also undervalued and I can't afford anything else. (Not that buying more expensive clothes is generally the same as buying clothes whose makers are more fairly compensated, as described above.) I learned to sew specifically because I can't afford to buy the clothes I want, so aside from remaking old clothes I also buy the cheapest fabrics, and I know the people making those fabrics are also underpaid. They, the sewers, and everyone else working for a pittance are also buying the cheapest possible clothing, and the fact that there is decent clothing we can afford is a good thing in that poverty doesn't translate as directly to rags as it used to, and clothing style is less of a definitive class marker than it used to be. The problem isn't exactly that the clothing is cheap; it's that the profits go to the owners and not to the workers, and that too many of us can't afford to vote with our pocketbooks for a better system.

We should all be thinking carefully about our purchases wherever we can afford it. Becoming aware of, and valuing, the work that we rely on is really important. Working towards a society that redefines financial responsibility to include the impact of one's choices on society is extremely important. But choices on an individual scale are only part of the solution: we need to work towards regulation to ensure fair wages and working conditions regardless of the state of the market, towards much fairer distribution of wealth, and in general towards a system where more of us actually have the financial breathing room to not always resort to the cheapest option and to make decisions not born of precarity.
posted by trig at 1:55 PM on September 2 [9 favorites]


Chinches are that when they automate clothing manufacture, they'll get rid of sewing altogether (or as much as is possible) and just knit the garments out of very fine yarn in one piece, much in the way that 3D printing is easier to automate than assembly of parts. They already have knitting robots, and algorithms that can translate 3D shapes to knitting instructions

This is already very much the case for running shoes (for actually running not as much for the athleisure category but it is showing up there as well). They have really moved away from having any sewn parts over to flyknit uppers in a big way in the past two years.
posted by srboisvert at 4:15 PM on September 2 [4 favorites]


When people opine that bras shouldn't cost much because they don't call for much fabric, author (In Intimate Detail) and blogger (The Lingerie Addict) Cora Harrington points out that the cost of a bra is not in the fabric -- it's in the labor needed to make a comfortable, durable, and attractive garment that requires 25 to 30 pieces (a pair of jeans requires 15) and has to hold the breasts in place and cover, lift, shape and support them for hours at a time.
posted by virago at 4:46 PM on September 2 [12 favorites]




nobody has to sew or knit garbage bags
do you see where i am going with this
posted by not_on_display at 5:06 PM on September 2 [5 favorites]


2N2222 are you saying that the conditions aren’t a problem, that there’s nothing that consumers can do that won’t create a different set of problems, or both?
posted by Selena777 at 5:32 PM on September 2 [2 favorites]


I shifted to buying a lot of clothes from Italy over the past few years and at the prices I’m paying I’d be very upset to find out they were being churned out by some fucking robot.
posted by um at 5:48 PM on September 2


The Great British Sewing Bee is a fantastic show, and does a good job at highlighting how hard sewing is and how long it takes. They generally spend two days making three pieces of clothing.

In the final of Season 4, the first challenge was to make a men's collared shirt, and everyone groaned. Buttonholes and collars and thousands of pieces and takes hours. And this is an item that is made en masse by sweat shop labour, and worn everyday by a significant number of people. And it's hard. The woman who won the Season messed up the collar and one of the button holes.

It's hard to learn to sew, and even harder to learn to fit. I know quite a lot of people who buy their child or themselves a sewing machine, and expect to be able to learn how to sew by trial and error. Sure, you can, but lessons are invaluable.
posted by kjs4 at 5:50 PM on September 2 [5 favorites]


In the final of Season 4, the first challenge was to make a men's collared shirt, and everyone groaned. Buttonholes and collars and thousands of pieces and takes hours. And this is an item that is made en masse by sweat shop labour, and worn everyday by a significant number of people. And it's hard. The woman who won the Season messed up the collar and one of the button holes.

It's not really an accident that business formal clothing is incredibly labor intensive and difficult to make and that as a dress code it effectively mandates and messages "I can afford to pay a bunch of indentured garment workers to make my fancy business clothes, have them altered to fit, and take them to the cleaners to be laundered." and that this systemic oppression is and has been an inherent part of the business world wardrobe since the beginning, whether spoken or subtext. (Not judging people who have to participate. There's also a reason why most people hate their work clothes and/or monkey suits!)

Business suiting and clothing is mainly based on and rooted in military uniforms, specifically the uniforms of colonialist Europe.
posted by loquacious at 8:36 PM on September 2 [13 favorites]


very likely
probably
one might think
I suspect


Treating this as an idle puzzle game for uneducated laypeople to speculate on is exactly the kind of pro-ignorance contempt for garment workers that the original tweets were about. You don't know, so all you can do is suspect. Understandable. But somebody knows. The industry is more than a theoretical playground for cultivating idle suspicions.
posted by queenofbithynia at 9:01 PM on September 2 [6 favorites]


Bringing it back to robots, one takeaway of this is we might be able to defeat, confuse or confound killer robots with little more than fabric and string and a tangled mess that is too chaotic, flexible and dynamic.

So yeah, let's start with giant Jean Claude-Christo fabric art projects everywhere.

I'm also thinking "bolt guns" that take wide bolts or rolls of cloth and use compressed air and high speed pinch rollers to shoot streams of fabric as far as possible and make huge tangled messes. Maybe even shoot a special kind of fabric, like a super light dyneema parachute fabric woven with spectra and/or kevlar threads. Embed tiny rare earth magnets, velcro dots or adhesives in it to increase tangle rates and stickiness - it would make it even harder for an advanced robot to manipulate it and untangle itself. Maybe weave conductive threads in the stuff and zap it with ultracapacitors from the operator end.

It would be very difficult for robots to navigate or untangle. If tangled piles of fabric were attacked with, say, whirling blades or chain saws it'd gum them up. Flamethrowers would make it ignite and turn into glue and napalm. Enough layers of it and it would diffuse or stop projectiles or break them up into fragments. It could be dyed to absorb laser energy.

An effective side arm might simply be a narrower roll of fabric like a packing tape gun. If you could, say, aim a stream of magnetic/sticky/strong fabric that tangles itself up at the sensors and articulation joints of a robot or the rotors of a flying drone and get some range on it, it'd probably be pretty incapacitating to a robot.

Yeah basically describing a cyberpunk tangle gun or riot control foam gun - but for robots.

In all of the sci-fi robot battle games out there I don't know if this idea exists yet and I'm kind of surprised it doesn't exist. I should take this idea all the way to a Battletech or Warhammer40K tournament or something just to see if I can make some nerds mad that I defeated them with - theoretically rainbow colored - high tech kite fabric instead of missiles, chain guns and directed energy weapons.


Bringing it back to sewing I just recently learned how to hand sew a fairly straight follow stitch instead of, oh, randomly spamming whatever inside seam it is I'm trying to repair and hoping it goes ok.

Anyone with any hand sewing experience will rightfully laugh but... I think I was trying to emulate the overlock stitch commonly used on a basic pair of swim trunks or sweat pants or something. With a single hand needle, without really understanding that that overlocker foot is like two needles and bobbins and a bunch of complicated machinery.
posted by loquacious at 9:07 PM on September 2 [11 favorites]


loquacious: I want contact information for your supplier, because that shit seems like it's really strong and awesome!
posted by hippybear at 9:11 PM on September 2 [4 favorites]


When crocheted items were a trend several years ago, I -- a crocheter -- stood in a Target looking at the huge amount of crocheted items in absolute wonder and horror. I explained to my wife, grabbing at something nearby, saying, "Look, machines can't do this look, everything here someone did by hand. A thousand someones are sitting somewhere doing every single one of these stitches."

Pretty much every time I see crocheted items in stores I have to restrain myself from running around like Charlton Heston at the end of Soylent Green to tell them everyone that very thing.

That, or wonder where the hell I could get some of the materials used for those items, because I can't get even the yarn for anything close to how little they were charging for crocheted baskets at a store this weekend. I know it's cheaper in bulk, but... Damn.
posted by asperity at 9:32 PM on September 2 [10 favorites]


I shifted to buying a lot of clothes from Italy over the past few years and at the prices I’m paying I’d be very upset to find out they were being churned out by some fucking robot.

In Gomorrah, Roberto Saviano says that the only way high-end Italian fashion can afford to be made in Italy is because it’s done in Mafia-run sweatshops. I don’t know how universally true that is.
posted by Bloxworth Snout at 2:14 AM on September 3 [5 favorites]


Yeah it wasn’t just Saviano saying it, there have been multiple reports on how "made in Italy" can mean simply "made in a sweatshop in Italy" - google made in italy + sweatshops and browse away. Often it’s immigrant labor, particularly Chinese-owned. Sometimes it’s homeworkers working at sweatshop wages for expensive high fashion brands. There really is no way of knowing where and how the garments are made exactly. Maybe consumers should demand that clothes manufacturing is tracked like with codes on eggs...
posted by bitteschoen at 2:34 AM on September 3 [6 favorites]


It goes without saying there are no guarantees unless I make my own clothes. Most of the stuff I buy comes from northern Italy. I don't care for luxury brands; instead I look for manufacturers which have high standards for the quality of fabric used, the construction, and the finish. All I can say is if there's sweatshop labor involved then they deserve to be proud of their work.
posted by um at 4:42 AM on September 3


I often say I'll start to worry about the robot revolution when they can fold laundry.
posted by lowtide at 6:57 AM on September 3 [1 favorite]


I've sewn a lot of my own clothes, heck, I've even handsewn dresses (very relaxing). Sewing is not the issue, FITTING is. You have to be very skilled to fit around problem areas.

Trying clothes on in a department store, you can immediately tell when something doesn't fit right. A skilled sewer knows if it is something that can be altered to fit. When sewing from a commercial pattern, you best make a muslin first so you are not surprised by the final outcome; you are in effect making two garments but muslin is cheap and good fabric is not.

I dream of those body scanner machines that will output a basic sloper. I know they exist, why aren't they at my local mall??
posted by TWinbrook8 at 7:02 AM on September 3 [1 favorite]


Oh! Consequently, I had the American Fashion podcast recommended to me recently. I've only listened to the "Even the Real Bags are Fake" episode about the leather industry which was quite interesting.
posted by crush at 7:57 AM on September 3 [3 favorites]


And here is their "Just to Clarify, Robots Don't Make Your Clothes" episode from a few years ago.
posted by crush at 8:24 AM on September 3 [3 favorites]


From a business perspective, skill level generally means the amount of time that it takes to learn a job. This article talks about a program that reduced the time needed to train a sewing machine operator in Bangladesh from an average of 32 days to 20 days.

It also relates to aptitude. (No amount of training would make me an NBA player). In Ethiopia, when women come from small villages to apply for garment factory jobs, the factory gives them "a dexterity test and divides them into three categories: gifted 'ones,' fated to work the sewing machines, and less talented 'twos' and 'threes,' who will pack boxes and sweep floors."

These are clearly low-skilled jobs. That does not mean they they are easy.
posted by Mr.Know-it-some at 9:09 AM on September 3


it’s surprising how much human thought, time, and effort is involved, especially with the care and feeding of specialized machines.

Thanks so much for this post, I saw the Twitter thread and was curious but was on my phone, didn't follow links etc.

We are really lucky to have a sock factory in the next town. They have a sale every year where you can get seconds and designs that are out of style. While you're there getting well-made locally-made socks for less than they usually cost (Darn Toughs are not cheap) you can also look in to the factory floor and see all the parts/time that goes into making a pair of socks. I've always felt lucky that the socks I buy help local folks stay in jobs that are decent. And I tell everyone: Hey Darn Toughs are ethical socks!
posted by jessamyn at 10:26 AM on September 3 [7 favorites]


hippybear > loquacious: I want contact information for your supplier, because that shit seems like it's really strong and awesome!

Although it’s out of print, you might get additional inspiration from The Acme Catalog.

While a hydraulic press works ok, a better anti-robot disposall would be a mobile scrap metal shredder. It’s the only way to be sure.
posted by cenoxo at 10:45 AM on September 3 [1 favorite]


(Not wanting to make light of a serious issue), let us remember What the Luddites Really Fought Against, Smithsonian, Richard Conniff March 2011:
...Despite their modern reputation, the original Luddites were neither opposed to technology nor inept at using it. Many were highly skilled machine operators in the textile industry. Nor was the technology they attacked particularly new. Moreover, the idea of smashing machines as a form of industrial protest did not begin or end with them. In truth, the secret of their enduring reputation depends less on what they did than on the name under which they did it. You could say they were good at branding.
...
As the Industrial Revolution began, workers naturally worried about being displaced by increasingly efficient machines. But the Luddites themselves “were totally fine with machines,” says Kevin Binfield, editor of the 2004 collection Writings of the Luddites [article link dead, substituted a review]. They confined their attacks to manufacturers who used machines in what they called “a fraudulent and deceitful manner” to get around standard labor practices. “They just wanted machines that made high-quality goods,” says Binfield, “and they wanted these machines to be run by workers who had gone through an apprenticeship and got paid decent wages. Those were their only concerns.”
posted by cenoxo at 11:19 AM on September 3 [12 favorites]


These are clearly low-skilled jobs. That does not mean they they are easy.

Ehh, I don't know how clear it is that sewing and garment manufacturing really is low-skilled. I can speak with experience as someone who has worn many hats. I've worked in IT and sat behind a desk, I've been behind a coffee counter, I've been a graphic and commercial designer, I grew up in manufacturing. I've been as low as digging actual ditches and moderately high enough that I was afforded an actual office with a door and a window.

I think we generally refuse to classify sewing as skilled because of the suppressed wages and working conditions. This is true for a lot of difficult, lower paying jobs, but I think it's a really acute problem with commercial sewing and garment work.

Even industrial/commercial cut piece sewing is a lot more like commercial cooking. Or maybe being a machinist. Or maybe even a programmer.

A really skilled and capable tailor/seamstress (notice the -ess and gendered titles! We're going to get back to that!) has to put in considerable training and hands on time to achieve those skills, and in my experience of having seen a little of everything it's easily on par with the amount of effort to become a commodity grade programmer or software engineer.

Just like programming or machining - some people have the gift and knack to be really good at it. That is a skill and high aptitude for something.

In the case of industrial/commercial sewing or garment work this means more than just production rates and volume. Good sewing machine operators have to know a variety of machines, how to operate them, do minor repairs and machine setups and production change overs. They have to know different fabrics and garments, too. Sewing and garment work is a world that is easily as complicated as software engineering.

So let's get back to the -ess in seamstress and the gendered language and history of garment production, especially after the industrial revolution, and how this relates to skilled vs. unskilled and wage/value suppression.

The history of garment work is overwhelmingly dominated by women. Why? Because of sexism, the patriarchy and the lack of economic opportunity for women. It's not because they're inherently better at the work - if that was the case rich men wouldn't go to nearly exclusively male tailors for bespoke suiting.

It's because it was work that they could do that had less competition because sexism culturally informed and shaped us to see it as unskilled women's work that is somehow lower than building steel structures or programming algorithms.

It is this inequality that has kept the price of our clothing suppressed since well before the industrial revolution. It was this inequality that provided the cheap labor, not the level of skill required.

We have had an addiction and sordid love affair with cheap clothes and fashion ever since the industrial revolution and commercial garment production made them so cheap, and there's a really complicated cultural lesson and tale here that involves everything from the history of slavery to the advancements of industrial steam power and electricity to the first hints of programmable computers and the whole thing is steeped in cultural sexism and inequality.

It is, as they say, the fabric of our lives.

In the shop I grew up in we didn't do a lot of sewing in house. Mostly we did neck label changes or small alterations as part of our support offerings for commercial t-shirt printing. We never actually sewed, say, a T-shirt, because that was all handled by huge cotton mills with sewing floors the size of aircraft hangers. (Side note: the neck label changes aren't anything nefarious or counterfeit, it was just a common practice to replace a cotton mill's default neck tag with one for the production company or brand.)

Even with the limited sewing we did we constantly struggled to find good, skilled sewing machine operators just to change neck tags in T-shirts. We only had something like four sewing machines, and at the peak of our business we maybe had something like 50ish total employees, and even out of that large pool we struggled to keep good sewing machine operators in those seats.

And that position paid something like 1.5x to 2x our going base pay rate. As far as I recall it paid roughly as much as a production artist or camera operator in our art department.

It was a skilled, sought after position. Not everyone had the skills to produce and operate the machine even after training.

The people who had the skills to work our sewing machines? Almost invariably all middle aged cultural minority women who had years and decades of considerable skill and experience, both industry experience and general life experience. Middle aged women who likely desperately needed the economic opportunity badly enough that the hard, repetitive but highly skilled work.

And whenever I visited other garment industry businesses, especially anything that did sewing or cut piece work - this demographic was mainly the same. Mostly middle aged immigrant or cultural minority women with a lot of skills and experience working too hard for too little.

You could take a doctor, an architect, a software engineer, an electrical engineer and they would struggle with the complex and arcane skills required to do even commercial production sewing - especially at production quota rates. Doing actual draping, drafting and tailoring, pattern making? Get the hell out of here. You're going to have to go back to school and study, and study hard.

I'd love to see someone from a so-called "skilled" industry or craft - especially a programmer! - try to learn how to sew. It would probably drive most programmers in particular totally batty specifically because of how chaotic and non-discrete it is and how hard it is to automate or repeat.

I really think we excessively undervalue a whole lot of jobs that we culturally consider to be "unskilled" labor, and a lot of that undervaluing is on behalf of protecting the power structure's vested interests in preserving that inequality.
posted by loquacious at 11:33 AM on September 3 [42 favorites]


I'd love to see someone from a so-called "skilled" industry or craft - especially a programmer! - try to learn how to sew.

I think it compares to the "skilled trades", like electrical/plumbing/etc. Except the male-oriented trades pay much better. Huh.
posted by rmd1023 at 12:30 PM on September 3 [3 favorites]


How does it compare to being an electrician?
posted by Selena777 at 1:09 PM on September 3


I've made things on a sewing machine, made things in a machine shop, and soldered things to breadboard (though I don't claim to be an expert in any of the above.)

If you want to be able to sew all kinds of things and customize them, that definitely needs a lot of training. Same if want to be able to machine lots of different stuff on demand, or to fix all kinds of electronic stuff.

If you just need to learn to sew one specific garment over and over again (or one part of one garment) I can believe that would take less than 20 days training. Same if you're installing the same ICs on the same circuit boards every day, or milling the same shape over and over again.

I think sewing is automatable, it's just that right now it would probably require a custom robot for each specific task on each specific garment, and that does not make a lot of financial sense unless you sell just one thing and are confident you'll be selling it for decades.
posted by OnceUponATime at 1:29 PM on September 3 [1 favorite]


Little surprised really that say, 501s, aren't made on an automated line.
posted by Mitheral at 1:37 PM on September 3 [1 favorite]


How does it compare to being an electrician?

My thinking was that there are a bunch of physical tasks/skills you need to master, and implementing them can be different from job to job. Like, I can sew a straight line in fabric, same as I can bend a piece of conduit in a pipe-bender, but boy howdy both tasks get much harder as you start getting more complicated in what you're doing. (It took me forever to get good at bending pipe offsets on the fly, for instance.) Electricianing has a lot of formalized training and certification that sewing doesn't, partially because badly sewn things don't usually cause fires, but I think both roles involve applying a set of basic skills to a complicated and sometimes unpredictable set of tasks. I mean, on a big job, you can have an electrician spend days going around and doing nothing but connecting wire nuts on a fluorescent light fixture in a suspended ceiling, and then you can have more experienced folks who can go and handle anything from snaking in old construction (harder than it looks!) as well as running structured cabling in plenums and wiring up high voltage distribution panels.
posted by rmd1023 at 9:17 AM on September 4 [2 favorites]


I think sewing is automatable, it's just that right now it would probably require a custom robot for each specific task on each specific garment, and that does not make a lot of financial sense unless you sell just one thing and are confident you'll be selling it for decades.

Little surprised really that say, 501s, aren't made on an automated line.

I'm going to dive in deep again because I find this whole topic fascinating. I probably sound like SFIA/Isaac Arthur from YouTube if he were talking about sewing robots and industrial automation instead of Dyson Swarms and Paperclip Maximizers.

They've tried this in depth with some surprisingly high tech solutions thrown at it. There's also been billions of dollars in R/D from the garment industry because the managers and bosses would love to be able to cut any costs or fat they can out of the production process as most of those costs are wages and human labor.

And to some extent, this is where sewing has been automated. Especially when producing popular/classic garments that haven't really changed much, like denim jeans, or button down dress shirts. Boots and shoes - especially classics like work boots or other standards have similar machinery and automation backing up the human labor.

They aren't sewing, say, the button holes or pockets of a dress shirt with a generic sewing machine. They have specialized machines for this sort of thing. Dress shirt collars have a dedicated machine, too, so do cuffs, and there are a wide variety of these machines and attachments that can be reconfigured for multiple tasks like this or styles. A pocket attachment machine will have different sized and styled frames for different sizes of jeans or shirts, etc.

There are thousands of different kinds of customized, task-specific industrial sewing machines, attachments and configurations. Thousands, tens of thousands. I remember pouring over equipment catalogs and just being bewildered by the scope and depth of the industry and all the different specialty machines, sewing heads and attachments. Major garment manufacturers and brands invent entirely new machines specifically to process and sew one part of one garment, because garment manufacturing is all about volume. More pieces per hour means less wages per garment.

While this is true in any industry, this is emphatically true in the high pressure garment industry where per-piece profit margins to a contractor or producer can often be measured in fractions of a single penny.

Buttonholes, plackets and pockets tend to be automated since they benefit from being more uniformly sewn and produced, but also because they have forms and processes that enable automation. A pocket attachment machine has a gripping frame and that holds the pocket fabric to the shirt and zips the sewing needle around the edge of that frame. There's even a lot of attachments that do this in an automated one stem cut/sew process to minimize seam allowances and bulk in, say, a pocket or zipper placket and refine the finish of the process.

Embroidery is also well automated and is effectively an X/Y axis plotter for a very complicated set of embroidery needles. Volume production embroidery machines often have 10, 20, 30 or more hoops and heads all with their own spindles of thread, all stitching the same logo on a bag or jacket or hat. (And these still require massive amounts of human operation! Embroidery machine operators earn high wages and have tech skills! They're basically tech support for the fussiest printers in the world!)

Another thing that has been relatively automated is fabric spreaders (they lay out rolls of fabric perfectly flat in dozens of layers for pattern cutting tables) and pattern cutting tables themselves, but last time I checked you can still get higher output rates (or lower per pieces costs or higher quality, choose an axis) with experienced fabric cutters and powered cutting tools because they can handle larger/thicker stacks of fabric without messing it up.

Laser cutting piece work is increasingly popular because fabric cutting processes churn through razor sharp blades like they're bubblegum, but there's a limit to how many layers a laser cutter can process, and it's generally like 1-6 layers or so, and it has it's own set of limitations, risks and problems like scorching fabric, melting and even starting fires due to thread dust.

And lasers are mainly used today for specialty processing, like cutting patterns of lace-like holes into fabric or leather for higher end finishes or looks that are nearly impossible to achieve by hand with volume production.

Another highly automated part of the garment industry is bulk weaving, mill and loom work, but working with rolls of bulk fabric is a lot easier than sewing cut pieces. And even the mills/looms need a lot of people to run those machines, because robots aren't so great at handling thread, handling yarn/thread bobbins and even just tying knots yet.

So there is a lot of automation in the garment industry.

Yes, it's really easy to automate some sewing tasks once the fabric piece is in the right place and alignment. And getting that piece properly aligned, tensioned and laying flat is a huge portion of the job of high volume machine sewing.

Now try automatically feeding any automated machine or robot a stack of cut piece fabric or whole sewn garments to load or process. This is the part that people really underestimate or just think "how hard can it be? Make the robot pick up a single piece!"

Cut piece fabric tends to stick to itself. Warp and weft threads tangle at the edges of stacked cut pieces. You might have better luck with something like a really thick heavy canvas or denim, but thin or stretchy fabric like swimsuit spandex, jerseys, knits, linings or lingerie fabric requires dexterity and is difficult to handle.

While I haven't done any bulk cut piece sewing, I have handled a lot of cut piece fabric because some of our jobs were cut piece spot printing. We'd get stacks of cut fabric from a client, usually tied into bundles of 100-ish counts with scraps of fabric and then thrown haphazardly into huge plastic bags or boxes. Between the layers of fabric are usually layers of pattern or indicator paper that is used to help stabilize the pattern cutting process and aid in piece counting. Most clients want you to preserve those paper layers and put them back in the stack in the right counts and orders after a processing step, and they get upset if the stacks aren't well aligned and tidy because it's essential for volume production and handling and keeping the fabric wrinkle-free.

It can be incredibly fiddly to pick up a small piece of cut fabric and place it, especially at volume production speeds. Fabric folds on itself, fabric sticks to itself. Picking up a single piece can mess up the whole feed stack and pull of not just one or two extras, but an entire chain of half a dozen or more pieces messing up your entire feed stack.

So a lot of this pick and place work is done at speed by touch and feel with your fingertips. A human immediately knows when it's picked up two pieces, or the piece is folded, or if they just messed up their feed stack.

For this to be automated we're talking about robots or machinery with enough dexterity to be able to both see and feel fabric layers and with the ability to self-correct feed errors at speed, on the fly.

Maybe think about this another way. We live in a world with some pretty advanced robotics actively in use in industry or research.

The tiny electronic parts on the circuit board of your phone or computer are placed by robots at speeds so fast you can barely see them being placed. We have flying robots that can play ping pong with each other. We have flying robots that can do crazy stuff like acrobatically flip themselves through small slots and barriers in ways no human operator or pilot could manage or even begin to calculate. We have walking robots that can stomp all over rugged outdoor terrain and react to being pushed over. Even large commercial bakeries use high speed delta bots to sort muffins or cookies and carefully place them into their packaging with less breakage than human hands. Just about any commercially and bulk baked good from sliced bread to crackers and muffins you've eaten in the last, oh, 5 years has probably been handled, sorted and packaged by an advanced robot. We have robots that can pipe adhesives or sealing gaskets along complex, convoluted compound shapes at speeds and feed rates that make humans just look bad. We have sorting robots that process bulk food as small as rice or kernels of corn at 50+ miles an hour using high speed machine vision and puffs of air to blow the unwanted rejects out of the stream and conveyor system.

We're even starting to see agricultural robots that can pick strawberries or other produce in the field, recognizing color and ripeness, and these machines do it several windrows at a time with arrays of manipulators and vision systems and practically look like creepy robots from one of the Matrix movies if they were painted all black instead of farmer friendly colors like green.

There are also now many, many automation and robotics companies that can build task specific robots and automated handling systems out of commodity engineering "jelly bean" parts from machine vision to robotic arms and intelligent conveyor systems that can do anything from make pretzels or paperclips or promotional pens with nearly turn key operations, and these systems are affordable and now within reach of small businesses that didn't previously have access to this kind or level of automation.

We even live in a world with self driving robotic cars that have successfully driven an incapacitated passenger to a hospital to deal with a medical emergency! There's self driving freight trucks out there that are doing real world work right now.

Robots and robotics are here in almost every industry and they're not just spot welding cars anymore. They're doing dynamic, complicated things.

And yet they still have difficulty handling fabric, in an industry that's been at the forefront of automation since the industrial revolution, an industry that's basically always been pushing the boundaries of commercial automation and volume production and industrial design. An industry that would desperately love to reduce manual labor costs and keep more of the profits to themselves by using robots to handle fabric. An industry that has extremely tech-forward companies like Nike throwing millions of dollars at production R/D, and who is one of the leaders in production automation.

It is not a trivial problem.

This is actually really fascinating because it really drills deep down into some of the hardest challenges of robotics and one of those challenges is dealing with and successfully handling things that are soft, dynamic and complex. Look at the examples of crochet in this thread. Why can't a robot crochet if it can knit socks?

It also illustrates how complicated fabric and clothing really is. As an object, as a force in our lives, as a labor intensive complicated production process and industry.

For advanced robotics to make inroads here they're going to have to have senses of touch and vision equal to or greater than humans. A manipulator picking up a piece of cut fabric is going to need to be able to feel edges, thicknesses, and weights to determine if it's holding one or two or more pieces. It's going to need to be able to feel or see details in the thousandths of an inch range. It'll need to understand and sense tension, alignment. It'll need to be able to understand the physics of cloth, that if it's pulled by the bias it distorts, or that it folds or drapes one way on the weft and another on the warp. It'll need to be able to feel the weave of fabric as it's being placed for alignment. It'll need to understand wind resistance versus the weight or type of fabric - moving a piece of thick canvas is not like moving a piece of tissue thin silk. It will also need to deal with relatively wide margins and tolerances of size.

To accomplish these things we're basically talking about biomimicry and nearly equal to or greater than human-levels of tactile sensory and spatial perception. I'm not suggesting that hard artificial intelligence or human-like consciousness is required, but at a minimum you're probably going to need some fuzzy logic, neural nets or machine learning, and it's going to challenge the current state of the art of machine vision and learning.

Even with the automation that we do have, this is why I count sewing work as highly skilled work, despite it being seen as and being menial or low-skilled repetitive labor. Specifically because of how difficult it is to automate.

There's a a lot of human skill, attention and emotional labor going into that work and it's exhausting above and beyond the repetitive labor. It's just as attention and focus wearying as coding or people skills in management. You don't just get to check out and be a robot.

I also count things like being a competent barista or short order cook as a form of undervalued skilled labor. There's more to it than just following the directions. We have robots that can make coffee or sandwiches and they're frankly awful.

Why? Why are these unappealing to a human? Why is a cup of hot cocoa from a vending machine so much less appealing than something made by a skilled barista, even if they started with similar ingredients? Why is a home made cookie so much better than a mass produced one?

One of the secret, unspoken ingredients to a lot of these tasks is, for a lack of a better term, love. Empathy for your client, customer, job or product. The discernment of a job well done instead of just done.

When I've worked as a barista or as a cook, as I am working I am asking myself "Would I want to eat this? Does it look appealing? Is it made well? Or does it look like I just threw the ingredients together with a minimum of care and effort with just rote acknowledgement of the instructions?"

I've worked in kitchens or behind counters with coworkers like that who checked out and worked like, well, robots and their product looked and even tasted like crap because they're not putting any care into the work and missing details like, oh, not putting the lettuce on a sandwich before you grill it.

And I'm all in favor of teaching robots how to love, have empathy and give a shit, but if/when we do we're probably going to find that they don't like having their work undervalued or underpaid, too.


I think sewing can be automated, too, I just think it's going to be a lot more challenging than people think it is, and what will likely happen first is we'll change how fabric or clothing is made long before we automate more traditional garment work with woven or knit cut piece fabrics.

You can see some of this right now, like the dimensionally knit uppers of running shoes as mentioned in this thread. Or that my nice high tech Goretex rain shell is basically all laser cut composite/layered synthetic fabric and non-woven membranes and employs things like sonic welding and seam taping in many places where it would have been sewn. Even the interior pocket is glued and bonded on instead of stitched.
posted by loquacious at 10:09 AM on September 4 [45 favorites]


Electricianing has a lot of formalized training and certification that sewing doesn't, partially because badly sewn things don't usually cause fires, but I think both roles involve applying a set of basic skills to a complicated and sometimes unpredictable set of tasks. I mean, on a big job, you can have an electrician spend days going around and doing nothing but connecting wire nuts on a fluorescent light fixture in a suspended ceiling, and then you can have more experienced folks who can go and handle anything from snaking in old construction (harder than it looks!) as well as running structured cabling in plenums and wiring up high voltage distribution panels.

Right, and that's another thing I wanted to address. In the garment industry employees have to be able to move around and cross train a lot to adapt to the product. You have to pivot and pivot fast.

In my dad's shop it was rare that someone didn't do a little of everything. There were crunch times when we even had the art department out on the production floor helping with manual labor.

Another thing I wanted to mention and point out is that the garment industry is wildly and energetically inventive specifically because of the lack of regulation and lack of critical risk like, say, an electrical installation catching fire.

You go into any garment industry production business with more than a couple of employees and they probably have some weird, one of a kind machine or process they invented in house. They probably have more than one.

The garment industry is definitely not afraid of innovation. We got up to some really wild stuff at my dad's shop just trying to improve anything we could in the process, and these innovations ranged from production machinery to experimenting with chemical dye processes and more.

There's a running inside joke that's probably 40 years old at this point in the screen printing industry that the most commonly used tool isn't a squeegee but a plain claw hammer, because you ended up using one for everything from building old style wood framed screens to gently coaxing or tapping screens into registration to banging on balky machinery to make it work or you're building a new work table or jig for a specialty order.

More than once I wished we had a machine or fabrication shop complete with a lathe, a mill and a welding table, and not just because I wanted to build robots. It wasn't uncommon for us to have to go to a fabricator to make custom printing pallets or workholding devices, or to call in welders for expensive on site repairs.

One small thing I managed to invent was a laser alignment device so we could run sewn t-shirts through a textile screen printing machine twice to combine two different incompatible printing technologies, which was previously considered to be impossible to do at production speeds because of how fabric distorts while being processed and handled. Traditionally, once you pulled a shirt of fabric from a platen you don't print anything else in or near that print because of this distortion and difficulty.

This was, oh, 30 years ago. Daylight-visible lasers were still relatively expensive. My invention was prototyped using a barely visible and very small helium neon tube laser, and it just projected a set of alignment marks with mirrors on the loading station of the printing machine. We upgraded this to what was then a very expensive set of much brighter visible diode laser modules. The total cost of the device ended up being around $500 in about 1990 dollars, so it wasn't exactly cheap. For $500 back then you could pick up a used, manual 4 color press and be half way to starting a one person garage sized screen printing business.

But due to the volume production increase and being the sole provider of this process, it probably paid for itself in a week or less. We ended up using it a lot even for other kinds of printing because it was so useful for more accurate substrate/fabric loading.

Today for less than $500 I could replace that system with a multicolored laser or even just a video projector controlled by a microcomputer like an Arduino and have it scan and project an actual image or outline of the second print needing to be aligned to an existing print. Throw in some more money and it could be a dynamic system with machine vision and live dynamic visual feedback being project.

And we'd still probably need a human operator to pick up that shirt and load it smoothly and accurately on the loading pallet.

Fascinating.
posted by loquacious at 10:39 AM on September 4 [13 favorites]


I "know how to sew" in the sense that I know how to read a sewing pattern and operate a sewing machine, but I don't really know how to sew. I can bang together a Halloween costume, but I would need a lot of training to be able to produce a garment that didn't look crappy.

One thing that hasn't come up yet in this conversation is how most cheap clothing these days is made of stretchy fabric, and is usually assembled with a serger rather than a conventional sewing machine. Probably because stretchy clothing is much more forgiving of an imprecise fit, which is the only kind of fit you'll get from off-the-rack clothing. If I wanted to make clothes that resembled most of the clothes I can buy in stores, I wouldn't be able to do it with the machine I currently own, even if I improved my skills a lot.
posted by showbiz_liz at 12:14 PM on September 4 [3 favorites]


This is actually really fascinating because it really drills deep down into some of the hardest challenges of robotics and one of those challenges is dealing with and successfully handling things that are soft, dynamic and complex. Look at the examples of crochet in this thread. Why can't a robot crochet if it can knit socks?

I pondered the machine crochet question here a while back, and I'm not sure I could do much better now. Though I'm doing some ridiculous tapestry crochet thing right now and omg why am I doing this thing where I have to switch colors every two stitches it is madness there is no flow state.

In conclusion, please enjoy AI-generated crochet patterns.
posted by asperity at 12:27 PM on September 4 [3 favorites]


I thought some of the advantage of stretchy cotton/lycra fabric is that the lycra in it lets them make fabric with shorter cotton fibers (with the lycra providing more structure) so you can make it cheaper. Also related to why some of the so-soft cheap t-shirts are so soft. I don't remember a source for this, though.
posted by rmd1023 at 10:16 AM on September 5


2N2222 are you saying that the conditions aren’t a problem, that there’s nothing that consumers can do that won’t create a different set of problems, or both?

Conditions are a problem when they're a problem. Consumer actions are difficult to assess. The industry is opaque to a large degree, one, because relatively few people are involved in it, and two, because so much of the industry is shielded by geography, language and culture. If deciding that you're going to stop buying new clothes makes you feel good, then do it. If you think it will help the plight of a garment worker, whose conditions you don't know, whose need for the work you don't know, well, the best you're doing is guessing. And it strains logic to see how a garment worker is helped by lowering demand for the worker's skill.

Treating this as an idle puzzle game for uneducated laypeople to speculate on is exactly the kind of pro-ignorance contempt for garment workers that the original tweets were about. You don't know, so all you can do is suspect. Understandable. But somebody knows. The industry is more than a theoretical playground for cultivating idle suspicions.

Few people here have an extensive insider knowledge of the industry. Cultivating idle suspicions, as you disparagingly put it, is is the nature of Metafilter. Would you say none of them is qualified to post an opinion? Somebody does know. But nobody knows everything. And some of us do know pieces. As do the people who created the links. I work in a trade. I can tell you all about the particulars about what I know with authority. But I can't tell you so much about the economics that make the business viable. The logistics involved in coordinating, scheduling production is mostly mysterious to me. The funny thing is , it's sometimes mysterious to the people actually doing it. Some practices are vestigial, some are done out of pure habit. Sometimes they're documented, sometimes not.

Back to the topic, garment manufacture involves certain skills. People seem to take umbrage, perhaps feeling that the skills are being belittled somehow. Whatever skills are needed, they are not particularly unique. That the industry can move across the globe so easily and utilize local workers is an indication that this is the case. It's not disrespect to acknowledge there are enough people who have, or can obtain, such skills to keep garment manufacture an endeavor that rarely rises past the "sweatshop" level. If that skill were rare, the people who possessed it would have the leverage to demand better compensation for it.
posted by 2N2222 at 9:16 PM on September 5 [1 favorite]


I’ve always thought that any dude using his grandmother as an example of an unsophisticated computer user should try to set up and use a sewing machine. I am a grandmother and I can do tech support as well as thread and use a sewing machine. Thank you, I feel like a superhero.
posted by theora55 at 1:48 PM on September 7 [14 favorites]


There was a study that tried to account for inflation all the way back to the middle ages to see how much a medieval peasant's t-shirt would cost in recent times. It is impossible to get an incontrovertible figure, but the number put forth was $3,500.

Food for thought in this context.
posted by fragmede at 12:52 AM on September 14 [3 favorites]


If that skill were rare, the people who possessed it would have the leverage to demand better compensation for it.

Are you sure? Sweatshops aren't just called that because they're warm. They're famous for ethical and safety violations, for owners carving out "special economic zones" so the laws of the country don't apply to them, and for generally appalling treatment of workers. Owners threaten your family or withhold pay if you try to negotiate, and are often the only employer in town.

Being rare doesn't give as much leverage as an economics textbook would have you believe.
posted by harriet vane at 3:04 AM on September 17 [4 favorites]


« Older incorrect horse battery staple   |   All Power To All Workers Newer »


You are not currently logged in. Log in or create a new account to post comments.