Cleaning up the Mess?
(Edit: Please do not shoot the messenger đ
As promised, we (along with the Mess Committee of the Student Parliament) interviewed Prof. Girish, incoming chair of the Mess Committee regarding the new changes.
The interview opened with us asking the true reasons behind the change of mess policy considering that the previous one had been working smoothly for all parties involved. Prof. Girish clarified early on that the previous mess system yielded a lot of work on the mess office, and he wanted to try and reduce it. Since the deadline for registering for each meal was 2 days in advance, this required the mess office to place orders for food items only then, reducing the leeway that the mess office had. Prof. Girish drew comparisons to assignments, saying that the daily order of ingredients were like daily assignments . In his opinion, the work of the mess office should also be to inspect messes and ensure their smooth functioning, and the tight deadlines got in the way of doing that. By closing registrations 4 days in advance, the mess office now has the freedom to place orders for multiple days together, improving the overall efficiency of the mess.
Prof. Girish stressed on the reworking of the mess operations. “We took this opportunity [closing of Kadamb] to rethink the way the mess operates, from the mess office to the daily meal preparationsâ. The institute has signed off on an emergency budget allocation towards the renovation of all the messes. The entire Kadamb kitchen is being redesigned by a contractor who has claimed to have kitchens accredited according to the US and EU standards of food safety . The new kitchen is claimed to have separate entries and exits, with dedicated stations for queuing, handwash, serving and dishwashing. To prevent the entry of foreign objects inside food, the kitchen is planned to be entirely enclosed and ventilated sufficiently, including two doors at every entrance (if you are having trouble, imagine an Indian kitchen inside a space station).
Technical Challenges
Far by the most long-lasting change that was introduced were the mess registration caps. Explains Prof. Girish, the idea of capping the number of registrations per mess is intended to respect the inherent capacities of the mess: a combination of the dining capacity, food preparation capacity, and labour capacity. Earlier, certain meals in messes had over 800 people availing, which strained the capacities of the mess. Student sysadmins attempted to add mess-wise caps to the old mess portal, however, they were unable to do it in time, and the decision was made to redo the portal with the help of the IT Team.
While in conversation, it was discovered that most of the changes were introduced to paper over the technical limitations of the current setup.
Unlike earlier, one cannot waltz in and eat in any mess they wish, with the biometric/RFID scanning taking care of the rest of the billing depending on what mess did people eat in. This required storing everyoneâs details locally. Now that only those people are to eat who are registered, and absolutely no one else, the machines need to actually turn down anyone who is not registered. This means that the machine only needs to store the details of people who have registered, and no one else. (Aside: I observed that the biometric machines are much faster now than earlier. Is it possible that we have another case of an accidentally quadratic implementation?)
We now require a slight diversion. The initial plan was to only use RFID for authentication. However, the manner in which the current machines are set up does not allow for decoupling of biometrics with RFID i.e. to add or remove someoneâs RFID details also entails adding (or removing) their biometrics. Naturally, biometric details arenât the smallest form of data, reportedly running into GBs. Additionally, the machines reportedly only provide the functionality to add or remove identities through a GUI, let alone an API.
Still with me? Because, just because, it takes quite long for uploading fingerprint details on each of the machines, and because this requires some poor soul to actually click buttons to upload the identities of every person for all meals, it was decided that this process should happen only once weekly. And that is how we arrived at the decision of eating at the same mess every week. Because the company providing the authentication machines chooses to make them as dumb as a brick with no recourse.
While this does not explain some of the other baffling decisions clubbed into the new mess policy (no re-registrations after cancellations?), it does provide some semblance of perspective.
Bare minimums
When pressed about what is truly required for replacing the temporary system in-place, Prof. Girish had only three requirements:
- Capping the number of registrations per mess
- Ensuring that the mess office gets an accurate number atleast 4 days in advance
- Integrating other systems like the biometric/RFID scanners and the ingredient ordering to minimise manual effort
Prof. Girish is relatively unruffled about the introduction of the new mess policy, claiming that it went âsmoother than expectedâ. Despite the outcry from the students, he remains focussed on the work ahead. âWhen you are managing [the meals of] 2000 people, people will complain. [It] doesnât mean that we donât get work doneâŚWe are still running a highly flexible system.â Due to the rollout of a radically new policy, the mess office was rather careful in their implementation and are beginning to relax several of the restrictions. Amongst these include the removal of the roll number scheme in Kadamb (already announced as of writing), the availability of a highly limited pay-to-eat meals at all messes to account for people outside the system and the possibility of moving cancellations closer to the date of the meal. Prof. Girish also agreed to introduce longer periods of registrations, instead of the current 14 days, so that it is easier for students to register in more relaxed intervals.
Prof. Girish is also optimistic about students being involved in building a new, completely open sourced, mess portal from scratch, and that includes the hardware. His vision is to make something that can be used across various institutes, not just IIIT, and students who work on it should be able to get credit for it. Only time will tell how far he can take his vision.
Edit: We contacted Prof. Girish asking for details regarding the claim of running into GBs (thank you to the anonymous commenter who ran the numbers!). His reply follows:
Institute uses a device provided by external vendor. As far as I know, it is used as a black box. We don’t know how they store the data. What I can confirm is for uploading data for 600 students to the device, it takes more than 3 hrs currently using the process/ software provided by the vendor. The IT team is not happy with this and is requesting students help is building an alternative system which can be much more efficient as you pointed out. We will be happy to buy and give the interested students the required hardware and support.
The Mess-y Situation 
Qu’ils mangent de la grenouille! (Let Them Eat Frogs!) 
Tale of Two Cheenties 
Peace of mind. 
Boats and Valorant 
Blessings 
A perspective on sports in IIIT 
Paintings of IIIT 
The Tale of Jagruti 
Cleaning up the Mess?
> Earlier, certain meals in messes had over 800 people availing, which strained the capacities of the mess
I would actually love to see the data, we can prolly do a bit more statistics to help because number of people per meal is just too opaque a metric not taking into account rush hour etc.
> Student sysadmins attempted to add mess-wise caps to the old mess portal, however, they were unable to do it in time
Whytf are they so insecure about the code. They had a month. If the code was open source, i’m sure the student community couldve done a satisfactory implementation (unless the issue is with integrating with the hardware)
> the decision was made to redo the portal with the help of the IT Team.
i would also like the see the timeline of events that happened here, im guessing they started thinking about it too late and gave very little time for the sysadmins and gave into their ims fetish.
> While in conversation, it was discovered that most of the changes were introduced to paper over the technical limitations of the current setup.
I do not understand this statement
> This required storing everyoneâs details locally.
I’m assuming this is a limitation of the device?
> The manner in which the current machines are set up does not allow for decoupling of biometrics with RFID
Is it not possible to change the set up? how long will it take? can we buy new machines?
> biometric details arenât the smallest form of data, reportedly running into GBs
you might wanna fact check that one
> the machines reportedly only provide the functionality to add or remove identities through a GUI, let alone an API
Seriously, how did they not anticipate this to be a problem when setting this up initially (hindisight ig)
> Still with me?
Nope
> poor soul to actually click buttons to upload the identities of every person for all meals, it was decided that this process should happen only once weekly
Autohotkey time? yeah i think its autohotkey time (idk, i dont have enough info to comment ig)
> Because the company providing the authentication machines chooses to make them as dumb as a brick with no recourse
FR!!
> Ensuring that the mess office gets an accurate number atleast 4 days in advance
Is the deviation from week to week so different? Again having the data would be nice
Since the regs are capped won’t they already have that data for the most part? or do we have fewer students than total slots?
Same. We did request, but we received a non-comittal reply.
Mostly correct.
Your assumption is spot on.
We did ask this. It turns out that while during the semester, there isn’t as much variation, it really shoots up during exam weeks for example. Again, we did ask for concrete numbers, but we could not get them.
This is great and really helps to improve transparency. I do, however, have some questions though:
1. Until better machines are brought about, will the weekly registration system continue? Has a timeline been given for when they’ll be able to procure better machines? This is specifically in relation to his claim that this system is flexible, which I personally don’t agree with.
2. Bringing on caps is a bandaid on a broken system right? A higher demand for a particular mess (let us take the example of Sunday Biryani or a Wednesday Dosa) highlights the need for better alternatives rather than a FCFS system that is just unable to provide the food. Were these points raised with Prof. Girish? Especially with the weekly system (which is here to stay for awhile), this forces students to deal with subpar options, and encourages them to turn towards unhealthy and expensive options.
No. As per our knowledge, they are working towards redoing the mess portal, and then the hardware. So it can be quite some time until things are back to normal.
Nor do we. Which is why we wanted to highlight it.
We are curious to what ideas you have (no snark). All the alternatives we came up with either did not satisfy the requirements, or were thought about by the Mess Office already, or were far more complicated to implement and reason about.
I have problem with only one of their radical changes, which is not allowing people to choose their meals from different messes. And what I gather from the article and the comment section it seems that the college, cheap-ed out while buying the authentication machines (as the article says there is no API, only a GUI for uploading fingerprint data) and/or some of their data is incorrect/they are lying. According to ISO/IEC 19794-4, storing a compressed fingerprint + some numeric identifiers should be KBs not GBs……and with the (ineffective) registration caps introduced for the messes, it would be (2x650x20 KB) [2 fingers per person, largest mess caps are at 650, assuming fingerprint data takes 20KB with Identifiers according to ISO 19794-4] which comes out to be 26000 KB = 26 MB
(unrelated tangent, one of the authentication machines used in North Mess used to say “Main Gate”, does that mean, one can spoof it? lol)
Back to data storage in authentication machines…..now if they were to store 3 sets of fingerprint data in each machine, 1 each for breakfast, lunch and dinner for a day….it would be 78 MB…..assuming each day the data set changed…then it would require 546 MB of data for a week…..which is not close to the claim of GBs.
To conclude, what I gather from my reading of ISO 19794-4 and the article, college cheaped out while buying authentication machines or the college is in such a dire condition that they can’t afford new authentication machines, hence they have decided to take away the freedom of students to choose their own meals.
This new system in no way benefits the students, what was this man high on while coming up with this system….I need some of that to cope up with my grades đ
PS : iirc PJN said in an FSIS that making changes to IMS takes a lot of time and it wouldn’t be possible to make changes to the course registration forms on IMS and yet somehow the magical IT team managed to implement a dogwater mess registration system to suck money out of our/our parents’ pockets.
Thank you for the effort! We were also extremely suspicious of this claim, but we did not have any numbers to back it up. We are reaching out to Prof. Girish as we speak to check on this.
Quoting a response from Prof. Girish: (credit to him for his quick reply)
If you wish to contribute to this effort, reaching out to your MP would be helpful.
Did you ask any questions about how the mess caps were arrived at? They do not add up to the student count, leaving a lot of people without any registered mess. The randomised allocation has also not been put into place, meaning that if you forget to register, you just go without food for the next week.
Another thing is that limiting people to one mess for the entire week makes no sense even with logistics in mind. The previous system, with simply a 4-day limit to booking meals, should still work, what is the logic for completely doing away with it?
It has long been a joke that the best software system which works in this supposedly IT college is the mess portal. I’m not happy about its demise
Yes. The reply we got was that the caps are set to 3 times the seating capacity (rounded off). We are planning to cross-verify the claim independently.
What apparantly transpired was that they wanted a final list of people from Jan 9th (when the course registration window closes). Until then, people were free to not register. However, the mess office is suspiciously quiet on these aspects.
While I thank Monish for putting out most of what we discussed, one important point was missed.
Most institutes that I know of run an monthly registration system (student is tied to a mess for an entire month). In IIT Hyderabad and Mahendra University nearby, they have a single mess for 4000+ students. Compared to this, even with the weekly registration system, IIIT Hyderabad runs a highly student friendly system with 4 mess for 2000 students.
Its very easy to take this for granted. Maybe (especially ug students) don’t know how it’s in other places. Current flexible mess system is maintained by putting extra efforts from staff members and faculty. We want to refocus more on Hiegene which means work of staff members on the registration system needs to be reduced. I am glad that few ug1 students have come forward with a clear plan in automating tasks and maintaining flexibility. Hopefully this will be ready in a month or two along with the renovated mess.