Slashdot Mirror
MIT's Elegant Schoolbus Algorithm Was No Match For Angry Parents (bostonglobe.com)
"Computers can solve your problem. You may not like the answer," writes the Boston Globe. Slashdot reader sandbagger explains:
"Boston Public Schools asked MIT graduate students Sebastien Martin and Arthur Delarue to build an algorithm that could do the enormously complicated work of changing start times at dozens of schools -- and re-routing the hundreds of buses that serve them. In theory this would also help with student alertness...." MIT also reported that "Approximately 50 superfluous routes could be eliminated using the new method, saving the school district between $3 million and $5 million annually."
The Globe reports: They took to the new project with gusto, working 14- and 15-hour days to meet a tight deadline -- and occasionally waking up in the middle of the night to feed new information to a sprawling MIT data center. The machine they constructed was a marvel. Sorting through 1 novemtrigintillion options -- that's 1 followed by 120 zeroes -- the algorithm landed on a plan that would trim the district's $100 million-plus transportation budget while shifting the overwhelming majority of high school students into later start times.... But no one anticipated the crush of opposition that followed. Angry parents signed an online petition and filled the school committee chamber, turning the plan into one of the biggest crises of Mayor Marty Walsh's tenure. The city summarily dropped it. The failure would eventually play a role in the superintendent's resignation...
Big districts stagger their start times so a single fleet of buses can serve every school: dropping off high school students early in the morning, then circling back to get the elementary and middle school kids. If you're going to push high school start times back, then you've probably got to move a lot of elementary and middle schools into earlier time slots. The district knew that going in, and officials dutifully quizzed thousands of parents and teachers at every grade level about their preferred start times. But they never directly confronted constituents with the sort of dramatic change the algorithm would eventually propose -- shifting school start times at some elementary schools by as much as two hours. Even more... Hundreds of families were facing a 9:30 to 7:15 a.m. shift. And for many, that was intolerable. They'd have to make major changes to work schedules or even quit their jobs...
Nearly 85% of the district had ended up with a new start time, and "In the end, the school start time quandary was more political than technical... This was a fundamentally human conflict, and all the computing power in the world couldn't solve it."
But will the whole drama play out again? "Last year, even after everything went sideways in Boston, some 80 school districts from around the country reached out to the whiz kids from MIT, eager for the algorithm to solve their problems."
The Globe reports: They took to the new project with gusto, working 14- and 15-hour days to meet a tight deadline -- and occasionally waking up in the middle of the night to feed new information to a sprawling MIT data center. The machine they constructed was a marvel. Sorting through 1 novemtrigintillion options -- that's 1 followed by 120 zeroes -- the algorithm landed on a plan that would trim the district's $100 million-plus transportation budget while shifting the overwhelming majority of high school students into later start times.... But no one anticipated the crush of opposition that followed. Angry parents signed an online petition and filled the school committee chamber, turning the plan into one of the biggest crises of Mayor Marty Walsh's tenure. The city summarily dropped it. The failure would eventually play a role in the superintendent's resignation...
Big districts stagger their start times so a single fleet of buses can serve every school: dropping off high school students early in the morning, then circling back to get the elementary and middle school kids. If you're going to push high school start times back, then you've probably got to move a lot of elementary and middle schools into earlier time slots. The district knew that going in, and officials dutifully quizzed thousands of parents and teachers at every grade level about their preferred start times. But they never directly confronted constituents with the sort of dramatic change the algorithm would eventually propose -- shifting school start times at some elementary schools by as much as two hours. Even more... Hundreds of families were facing a 9:30 to 7:15 a.m. shift. And for many, that was intolerable. They'd have to make major changes to work schedules or even quit their jobs...
Nearly 85% of the district had ended up with a new start time, and "In the end, the school start time quandary was more political than technical... This was a fundamentally human conflict, and all the computing power in the world couldn't solve it."
But will the whole drama play out again? "Last year, even after everything went sideways in Boston, some 80 school districts from around the country reached out to the whiz kids from MIT, eager for the algorithm to solve their problems."
What about providing optimal bus routes without changing start times? Or what about factoring in a cost for changing start times to only do so when the new start time makes a huge difference in the bussing cost? They just need to take into account the political cost of moving start times as another set of parameters.
Hundreds of families were facing a 9:30 to 7:15 a.m. shift. And for many, that was intolerable. They'd have to make major changes to work schedules or even quit their jobs...
This sounds like a perfectly legitimate argument against the plan. The plan wasn't nixed because people were angry to it, the people were angry because it's a bad plan.
Given there have only been around 4.3 x 10^17 seconds since the big bang, it seems unlikely that they actually sorted through 10^120 options.
Another case where the installed base wins over new things because it is too disruptive to change it.
The school scheduling equivalent of COBOL.
Civilized people are dead at 7:15 AM. What kind of an asshole would demand that you get a kid to school so early?
-jcr
The only title of honor that a tyrant can grant is "Enemy of the State."
Programming would be so much easier without the damn user!
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Too-early start times, especially for high schools, are a well known reason for poor academic performance:
http://time.com/4741147/school...
https://www.cbsnews.com/news/s...
Wonder what the public key field is for?
Nearly 85% of the district had ended up with a new start time, and "In the end, the school start time quandary was more political than technical... This was a fundamentally human conflict, and all the computing power in the world couldn't solve it."
No, it wasn't 'political'. The algorithm successfully computed an optimal schedule for the students with regards to bus transport, but did not include any data at all about the optimal schedule for the parents.
If they wanted to find the optimums, they should have included the whole system and not just the least impactful part. The parents schedules are the most important ones since they are responsible for making it all happen; from breakfast to dinner to bedtime.
I see this all the time. Brilliant programmers and mathematicians that think they can just throw the data into an algorithm and get an answer without understanding the data itself or how to interpret it medically/biologically.
You might not be a parent.
If elementary school starts 2 hours earlier, that means it also *ends* 2 hours earlier. Now, where does that elementary school child go for the 2 extra hours (while parents are, presumably, at work)? That's right - now you need pay a nanny.
There is nothing "elegant" about an algorithm that optimizes one variable. You can optimize any one variable to the detriment of others (what if *all* kids go to one giant school, for example :) ) It is not a fault of the algorithm though - it's just relying on poorly specified conditions. To be actually useful to real humans, it should have included as many relevant variables as possible.
No need to blame "angry parents" and no reason to laud poorly specified task.
...if I had a mere dollar for every project that failed because they failed to identify the primary customer and understand their needs.
When Fascism comes to America, it will call itself Anti-Fascism, and tell you to give up your guns.
That's because while they might be hot shot programmers, they are crappy analysts.
I suspect that any grad student from a, "lessor" school, but had been trained in analyzing requirements correctly, this would not have happened. That's because they would have asked the most obvious question, who are the customers and what are their needs.
Failing to account for the work schedules of parents (the real customer ) is a 100,000 watt light sign proclaiming Inadequate Analysis.
When Fascism comes to America, it will call itself Anti-Fascism, and tell you to give up your guns.
What exactly is bad about starting school at 7:15AM?
For most people, that means drop the kids off at school, then head to work.
It's only a problem if you've organized your life around the 9:30 start time. The change would be difficult for some. (I suspect that for many it would be less difficult than they make it out to be. People often complain loudly about change, then when that doesn't work they make some simple adjustments to adapt to the change.)
In Europe, many countries are considering moving the start of the school day from 8-9 am to 10, because it is simply neither healthy nor productive to wake people up so early, child or not.
MIT's algorithm wasn't elegant. It was a complete failure. Still, it is being spun as some sort of success. They're blaming the field of algorithms. The brilliant and elegant men of MIT could not fail.
All this university branding ...
The real root of the parents complaint is the lack of individual situational data into the algorithm. Plug that data into the system, and the results would not be as dramatic, but I guarantee they'd still pay for the project itself multiple times over.
I wonder why they overlooked the individuals as components of the algorithm versus just scheduling the bus. It's like they were programming a motherboard bus where people are 1s and 0s instead of dealing with complex human individuals.
Here's a good exercise for them to remember to think of people:
Create a program to determine where an office full of 15 people can eat lunch on Fridays. Take into consideration, work schedules, personal restaraunt preferences, food preferences, allergies, location of eatery, menu options, service level, eatery reviews, avoid repetition, transportation, weather, traffic, eatery wait times, busyness by hour, price.
Humans negotiate a lot of that in a conversation and usually with a larger group, there's some sacrifices that need to be made. A program should be able to do this, and MIT should have done something like this before attempting the schoolbus problem.
Is that problems like this didn't happen when only one parent had to work to support a family. The 'homemaker' stayed home and handled things like weird school times, plus you know, actually teaching their own kids about life. Now both parents have to work, isn't it nice that now it takes two people working full time jobs to earn enough that used to be done by ONE full time parent? This tells me that your job is only worth half the value it used to be! Progress for corporations, at the expense of your family, indeed.
....it was the choice of cost function. They could have chosen the cost function so that no school had their time moved up more than, for example, 30 minutes. The end result might not have saved as much in bus costs, but by removing the objectionable results, they might have successfully implemented the optimized schedule. If they had asked parents about acceptable start times in the surveys, surely they should have exposed the problem up front.
Traffic is no more dangerous than 30-40 years ago, in fact, it's safer. I don't think a larger % of the population has a proclivity to kidnap and abuse children than before, either.
You sound like part of the problem. In normal places like NYC or parts of Europe, kids walk, bike, or take public transit to school, and parents aren't quaking in their boots in fear.
Home schooling works superbly in the best case.
In the worst case, it's a disaster. Parents don't care, kid watches TV for a year, learns nothing and loses much of what was previously learned. You'd be surprised to find out just what portion of parents are that bad, probably close to one quarter.
Contribute to civilization: ari.aynrand.org/donate
Arriving to work on time sure, thats a fine value
Why is going to bed early and getting up early a value? Its all arbitrary. At least all of the shops I know of and have worked at dont really give a crap what time you show up (at least before noon so you can be there to discuss things with others or to make it to a meeting when scheduled). As long as you aren't inconveniencing someone else by not being there what does it matter, do your 8hrs of work and as long as you get it done adequately who cares
Of course this doesnt apply to shift work, when you have to be there at a certain time, but for anything non shift going to bed early and getting up early is a preference, not a value
"It's just want I asked for, but not what I want!"
The MIT leftie loonies are kept in line by actually having to prove their scientific claims. Harvard, however.....
The problem for the MIT researchers was that they tried to solve a math and a scheduling problem. They'd never attended a town hall meeting in Southie, or in Charlestown, or the political loonie of anything involving the Cambridge Housing Authority. Now *that* is a bunch of overprivileged wealthy white single moms who don't actually have to worry about rent, they just worry about where to find cheaper gas for their SUV that their daddy or ex husband is paying for. Been there, done that, dealt with their spoiled rotten over-medicated kids.
We don't really have those problems here.
In first world countries, it is assumed that people work for a living and that the average work day spans from around 7am to 5pm with a little time for drifting. As such, in first world countries we have government subsidized day care that operates from 7am to 5pm. This means that everyone should be able to make their work window happen during those hours.
For children who are too young to be home alone before and after school, the schools are open early and there are people watching the playgrounds. Then there are programs sponsored by the government to provide after school activities (similar to day care) for kids up to around 6th grade until 5pm.
In these environments, we don't have school buses... we simply have public transportation. The parents drop off and pick up using public buses... even if you live on a farm 500km from civilization... there should be a regularly serviced bus stop nearby.
Parents often make groups to walk kids to and from school each day... and the single parent with a long way to go to get to and from work generally don't have problems because no one would consider making one of their child's friends mother have to quit their job.
Then there's the issue of making sure that mom or dad don't have that problem. Whether you're a 1%er or you're the bottom 1%, the government pays your child welfare to make sure their have what they need. This pops an extra $300-$500 a month into your bank account. So you can afford to have a slightly more flexible job or even to be a student long enough to make things easier later on.
We pay for this as tax payers in the first world and don't think anything of it. It doesn't matter whether we choose to have children of our own or not. What matters is that the people we work with need to be healthy. The people who work for us need to be healthy. The people who pick up the trash on the road need to be healthy. The people who we pass on the street need to be healthy. Otherwise, you get second world problems like school shootings because people aren't healthy. Or equally disgusting... people live in neighborhoods with security gates and guards because they're terrified of their own lives.
The first world is willing to live with a little less to get a lot more. We have governments with parties who we don't trust, but are smart enough to make sure there are enough parties that they can't make any choices without actually debating those laws openly. So while we don't trust the people in the government... we trust their enmity towards one another to keep them from hurting us. We also trust the government to make sure our tax payer money is spent in a way that will get them reelected because we can see, touch and feel how much better our lives are than the second world Americans on TV.
Well, and this just goes to show that the Netherlands has never had to deal with "forced bussing".
Once upon a time, it was determined that black kids got worse educations at their local schools than white kids at their local schools. The obvious solution? Bus some of the white kids to the black schools, and some of the black kids to the white schools. Net result? Most kids don't go to the close-to-home schools, they go across town, no matter where they live (the exception, of course, is the children of the upper class, who could (and did) afford to go to private schools).
Anyways, for the last 50 years, major cities didn't allow for the option of "go to the school just down the street"....
"I do not agree with what you say, but I will defend to the death your right to say it"
Actually, Massachusetts almost totally got rid of "forced busing" rules around 2013. Many towns and cities the size of Cambridge (~100k population) in the US have their own districts and route kids to local elementary schools within a mile or two of their homes.
Even bigger US cities do this -- in NYC, kids typically go to local elementary schools, semi-local middle schools, and then take public transportation to high schools.
A long long time ago, known as the '80s, we still had people called "general business consultants". These people were hired by businesses, for lots of money, purely to look around and make decisions. They didn't justify those decisions (with any sort of data) at all. They simply said thing like "I believe this is the best course of action.". You believed them because they had a good track record and experience in the field, or you didn't.
Nowadays, general business consultants have been replaced by data engineers -- people who like to collect huge amounts of data points, and have dumb-ass machines make decisions based on those data points. Alas, like every study that's ever been done with data points, it all comes down to whether or not you have enough of the right data points, and not too many of the wrong data points. And that's a skill that absolutely none of these data engineers has ever had.
I can make the cost of transportation absolutely $0. It's really easy. I'll just cancel all of the buses. Oh, wait, you actually want buses? I hadn't thought about that. Okay, I'll take your children at midnight. Oh? You don't like that either? Here's a thought, I'll get more buses, not stagger anything, and you'll be happy. Oh wait, we don't have that much money?
Look at that. Balancing costs and services can't make everyone happy. Maybe happiness costs money, or customers. Shock of a lifetime.
Maybe one day, data engineers will be able to put in the very important data point that says we're never trying to solve a problem. Solving any problem is ridiculously easy. We're always trying to solve a problem within another problem -- within a context. Like, in this case, within a parent's business day. That's hard, if not impossible, every time.
And apparently I need to learn how to insert paragraph breaks. Mucho apologies.
Comment removed based on user account deletion
There is a fallacy of averages in play here.
We try to provide an education for all students, and there are federal laws protecting that right to an education. But some students cost more to educate than others. Special needs students are very expensive to educate. Most charter and voucher schools find ways to get out of taking their fair share of special needs students, and few parents will have the resources to home school them.
But voucher systems typically pay the district-wide average of student cost, rather than the average cost of educating a non-special-needs student. As a result, they overpay for what the schools are delivering. Students who are less costly to educate leave the public school system, leaving that system with a higher percentage of those expensive students while simultaneously damaging its economies of scale. The result is a downward spiral of public education.
I can certainly understand the desire for more mixed schools; some schools in Amsterdam could certainly use being a bit more mixed.
But still, that doesn't mean everybody has to be bussed, just half of the kids living in a neighbourhood that is itself completely homogenous as well as too far away from a different neighbourhood.
Of course the real solution to that problem would be to encourage more mixed neighbourhoods. Add bigger, more expensive houses in poor neighbourhoods and more affordable housing in wealthy neighbourhoods.