Volunteer Recognition Ideas (New: Fill & Download for Free

Basically, the problem that kawalpemilu.org trying to solve is:How do you digitize over 470,000 C1 image scans so that it can be summarized easily?The first idea that come to mind is Optical character recognition. However, I could not find a good enough online OCR tool that can handle handwriting recognition. I did find some websites that do handwriting recognition, but I couldn't find a program that can be readily used (usually the program needs some initial training data beforehand). Since the handwriting of the images come from different people from all over Indonesia, I figured that none of the handwriting recognition would work well (I'd love to be proven wrong).The second idea that come to mind was crowdsourcing. This is not a new idea, since there were a number of other crowdsourcing websites already running (such as realcount.herokuapp.com and http://kawal-suara.appspot.com). The crowdsourcing idea is to build a system that allow users to see the image and manually enter the seen digits into the system. This idea has its own challenges:How do we find the users?How do we verify the digitized results? (how to make it accountable?)How do we make it fast and addictive? (we only have less than 2 weeks to digitize the 470,000+ images)How do we make the system to scale? (handling thousands of concurrent users? crackers?)We gathered insights and learned from the existing crowdsourcing systems to address these challenges. At that time we seem to have the answers to these challenges so we went ahead with the crowdsourcing idea. Essentially, answering how kawalpemilu.org works is the equivalent to answering how it addressed the challenges above. Now, I'm going to elaborate in more detail for each challenge above.How do we find the users?Learning from the other crowdsourcing systems, it seems easy to find volunteers to help in digitizing the images. However, allowing arbitrary volunteer to contribute can backfire. If some volunteers are malicious, they can enter garbage values into the system and it would be very hard to cleanup later. There is a technique that uses at least two users to digitize the same image and only accept the values if both users enter the same value. This technique slows down the data entry by at least twice and may not be effective against collusion attack (i.e., a set of users agrees on certain value for Jokowi and certain value for Prabowo).Our solution is to leverage trusted volunteers. Ainun Najib recruited a set of people he trusted as volunteers, and in turn those volunteers may recruit another volunteers that they trusted and so on. The number of volunteers grew rapidly and after just a few days we stopped the recruitment when we reached 700 volunteers. At that point the digitization process is almost done (around 97%):How do we verify the digitized results?Via another crowdsourcing system :). Yes, we built two systems: the internal system (private) and the external system (public). The internal system is only accessible by the trusted volunteers. The external system is accessible publicly, so anyone around the globe can verify the digitized results that were entered by our trusted volunteers and report any error back to us.The trusted volunteers need to sign in using their Facebook account to use the internal system. The volunteer's Facebook account is associated with the images that they digitized. This is designed on purpose so that malicious volunteers will think twice before they try to mess with the internal system. Moreover, each image can only be digitized at most once, so that malicious user cannot edit the values of the images that were already digitized. Edit privileges are only given to "super trusted" volunteers. We received around 4000 error reports and fixed them (sometimes it involves broken images which we reported to KPU). So far, the error reports we received were due to honest mistakes. I'm really grateful to know the trusted volunteers community who are scattered around the globe:How do we make it fast and addictive?The existing crowdsourcing sites I tried uses randomization for determining the next image to digitize. One advantage for this approach is that it makes it hard for malicious users to collude. On the flip side, I think, it demotivate honest users since the user cannot control which region they want to help digitizing. Moreover, it does not give the users any reward (i.e., the users cannot see the values they have just entered). To motivate the users, one may create a rank list of users who digitized the most images. However, this may backfire (i.e., the users may make more digitizing errors to get higher rank).Since I didn't have enough time to create a game (i.e., to use gamification for the digitization process), I'm thinking a simpler way to give rewards for the users and give the user freedom which region they want to digitize. I figured, what about showing back the users the values they just entered? That seems rewarding enough. For example, after the user did the hard work in digitizing the images from a particular village, they can roll up (drill up) to see who wins for that village. This approach also frees the users to pick which region they want to digitize and make it easy for organizing the volunteers to distribute the work.To make the internal system fast, we use fancy dynamic html to auto-advance to the next image when the user hits the submit button for the current image. This allows the user to digitize an image in only 5 seconds. We also crop the images near the numbers and sent to the users a smaller sized image (i.e., 10KB instead of 150KB) to speed up page loading. We built a custom in-memory database using C++ from scratch to ensure low latency for navigation and digitization submission. Below is the latency distribution. Most requests are completed in less than 1 millisecond.I was lucky that at that time I was in Singapore where I have installed SingTel Fibre BroadBand at home. I can download the (70GB) images from the KPU website quickly to make the smaller cropped images. Otherwise, I would have to hot-link the images to the KPU website which was sometimes slow and sometimes unavailable which may slow down the digitization process.How do we make the system to scale?Once the external system is made public (kawalpemilu.org) and covered by media such as MetroTV, detik.com, the number of page views increases rapidly. According to Google Analytics, over the two weeks period (13 - 27 July), we received close to 3 million page views:At peak, we received around 5000 concurrent users (in the realtime overview). We also received DDoS attacks where the external system was hammered with 9000 requests per second.We used Google App Engine (GAE) for the external system. The external system fetches and caches the data from the internal system periodically (e.g., every 10 minutes for recently accessed pages). Thus, no matter how many requests the external system receives, the internal system is unaffected (i.e., the internal system only receives a constant low number of request per second). We rely on GAE for scalability and security. That is, as long as Andrian Kurniady puts in a valid credit card for the GAE account, we can rest assured that the external system will be safe from attackers and stay available :).UPDATE Jan 10, 2015:How does the in-memory database works for the internal website?The in-memory database accepts 18 operations via HTTP requests:https://github.com/kawalpemilu/kawalpemilu2014/blob/master/internal-backend/api.cc#L1092-L1109The main data structure for the administrative area navigation is a Tree data structure (using KPU's internal ids in their database):https://github.com/kawalpemilu/kawalpemilu2014/blob/master/internal-backend/load.h#L110The in-memory database advantages:- It is simple. If you can solve this INC-H problem, then you are qualified to write an in-memory database yourself :). 50 indonesian teams who participated in the Indonesia National Contest 2014 can solve it. So, it's not hard. This in-memory database can be coded in less than 1 hour!- It is single threaded. Since i'm using libuv for http-server, all requests are serialized. Thus, I don't need to think about data-race, synchronizations, etc. This gives the Isolation property in ACID.- High performance. It loads the entire 470K+ nodes into a tree data structure in < 2 secs. Each query is answered in at most 4 ms (mostly it's 32 microseconds). The tree data structure allows each update request to be done in O(1), and maintains the aggregated results in the intermediate nodes, allowing each summarization query to be answered in O(1) as well. Without a tree data structure (normal database don't have it), you would need to run an expensive "GROUP BY" query that scans the entire rows in the database.The in-memory database disadvantages:- Not fully durable. I am using a very crude mmap (memory-mapped) files to persist the data in-memory to disk. So that, if the server crashed or shut down, the data is saved on disk and can be re-loaded. It could be that the last few updates before the server crash, the updated data were not yet saved to disk. Luckily, during the heavy load in the crucial first week, the server didn't crash :D. This is a trade-off I was willing to make. The http-server was well tested over the years for running the uHunt API.- Not atomic. If there is an update request that updates several values in the database, and the server crashed midway, it could be that parts of the updates were successful while the rest are not. To minimize the severity of this, I minimize the things that I need to store to disk. I don't store the aggregated values in the tree, I recompute them upon reload. The things that are persisted to disk are TPS records and error reports only.- Not consistent. Obviously :). It doesn't have a roll-back mechanism in case an update query failed in the middle.As you can see, there's lots of severe disadvantages, yet I still used the in-memory database because I needed the performance. The internal-website is very responsive due to this and it becomes "addictive". There are research that say: if the response time is slow, then the user will "context switch" to other things. With <4 ms response time, the volunteers won't have the time to context switch, and thus stay focused on data entry :), i.e., addicted.That's all about how kawalpemilu.org works. You can still ask for more details in the comments below and I will update the text above.

For me, it happened very slowly. I don’t know when it started.It feels like sitting on top of a very very tall bubble, 500 feet high, and it is leaking air. You don’t know where the leak is coming from, and it is very slow. But you know you are sinking and you cannot do anything about it. Your whole body is sinking. But you cannot get off without falling. So you sink.You adjust to your new height a few feet lower, a few feet lower still. You keep sinking, you keep adjusting.YOU haven’t changed, you’re still on the bubble. But your view has changed, you cannot see the things you used to. You miss those things. You feel sad. You see new things, unknown things. You feel scared. You don’t know what to make of these new things.Sometimes the bubble deflates faster and you get sick, lose your orientation and sense of self. You panic, you think you are falling. But you are not, you’re just sinking.Then, finally, the bubble deflates entirely. You are on the ground. And for the first time you can get up, step off of the bubble and walk around. You can walk away. You have no idea what awaits. Perhaps another bubble, perhaps not. It is a brand new world, one you always saw, but never knew.I just stepped off my bubble. I’ll tell you the story.There is one thing I have wanted to be, for as long as I can remember. An elected official, U.S. President to be specific. Scoff all you want, but that has been my dream.I think it was the fourth grade when someone first said, when can I vote for you for President? Those early compliments stick with you. They inform your identity. I promoted myself in roles that were leadership roles and they fit, they made sense. And I received positive recognition that reinforced these feelings. As I matured, I developed compassion and empathy, a deep desire to do good and help people.And I repeatedly saw being an elected official as a way to use my skills and fulfill my desires;I saw the job as someone who takes all of the pain around them, all of the misery, fighting, squabbling, insecurities – absorbs it, and then leads people to be their best selves.I saw the job as someone who makes decisions, who can bring calm to chaos and noise.I saw the job as someone who can be the punching bag she needs to be to make the right choices.I saw an opportunity to test this theory a decade ago when a guy named Mitt Romney ran for Gov. of Massachusetts. What a guy, smart, talented, business-y, commonsense. I quit my job and walked into the campaign.I stayed with him his entire term. I loved it. I loved government, I loved the politics. Seeing how things worked make me more eager, not less, to be involved. That being said, I did become a little less idealistic than I had been. Just a little.So then it became an issue of where would I fit? Geographically I mean. I’m a very moderate Republican. There are only a few places I would fit and be elected. I’m from Michigan, but did not want to move back there. I did not really feel motivation to run in Massachusetts. New Hampshire? Possibly...I got advice; move here, study that, volunteer for this, donate to that. Lots of advice. But I never pulled the trigger to start this career. I convinced myself I would be better as a politician if I knew something of the private sector first. Understood how it worked.So I went to business school. Getting there was a nightmare, I had to take the GMATs 5 times. But I had convinced myself that analytics and management skills would advantage me in my pursuit. I did not respect lifetime politicians; after all, what do they know? I talked about being Republican at school, a compassionate Republican. A few people told me to “let them know when I ran.” I felt a rush again.And then, love came along. I was tethered to someone, in a good way, in a new way that compelled me to make sacrifices. He didn’t ask for them, but was willing to do the same for me so it seemed fair. And still again, I didn’t know the next step to take. So I did consulting, it’s a little bit of everything. Easy right?No, not easy. McKinsey was not easy. It continued to kick my ass. But I stayed until I was promoted, and promoted I was. Not directly working towards my dream, but not veering away from it either.Then, flushed with success, I quit. I was available again, top of my game. But still….I wasn’t ready to pull the trigger. I didn’t want to waste precious time either though…inertia ruled me.So decided to do something to avoid making a life decision and yet, not waste time. I hiked the Appalachian Trail. People tell me how hard this must have been, or how brave it was. My husband and I know the truth, hiking alone in the woods, day -in day- out is easier for me than to pursue a dream that I think I won’t achieve. I cannot explain it, but that is the truth.I finished. I was available again, and even more confident! But I needed a few weeks to process what I had been through. And all of a sudden, my dream wasn’t as clear from my new perspective. Politics seemed dirtier, heavier, I didn’t know if I could be who I needed to be to do it. It was still there, just warped.And then, a life decision was made by someone other than me. It was made by us, this new us that I was part of. My husband got his dream job in London. So we moved here. And he is doing something that is the best of the best in the world. And I couldn’t be more proud. But here I sit, developing a cheddar addiction, Quoraing all day…wondering what is next.I have always asked myself “could I do that and still become President?” If the answer was no, I wouldn’t do it, wouldn’t say it. Simple as that, it was principle, it was what governed me.Recently, however, I have asked myself that, said no, and then…still did it. Quora has helped. I’ve talked about things that “Presidents” don’t talk about. I’ve opened up about being things “Presidents” don’t open up about.And goddamn it if I don’t feel like I’m screaming outloud for the FIRST TIME EVER!!! Holy shit I was constipated for 30 years. And now it just pours out left and right. It feels honest, it feels true, it feels ME. (And a lot of it is crap, sorry about that).I realize, I might be the only person in the world who attended a top business school, applied and worked at a top consulting firm, walked 2000 miles AND moved to a new country – all to avoid failure. I couldn’t kill my dream but couldn’t pursue it. I was trapped. Trapped on the bubble.But along the way, these things that happened had a way of changing me. Letting go of this dream is no longer about failure, it is about what is right for me. Being President (or attempting to be such, since lord knows I would never actually get the job) isn’t right for me. To do that job, and do it well, I would have to make sacrifices I am not willing to make. I cannot lie, I cannot cover things up about myself, I cannot compromise what I believe to win a ticket, and I cannot subject people I love to a life of scrutiny and judgment they did not ask for. Frankly, watching my old boss go through all that when he ran for President was pretty seminal.I found a new principle too. Whereas previously I applied honesty and integrity to my dream, at the expense of giving it to myself and others, I now do the exact opposite. And it feels much better.I don’t know how to incorporate the best parts of myself and my desires and principles into a life, no idea. But I know I will figure it out and I will nurture my new dream because that is how I work. But I think it will be different this time.Yes, I just stepped off the bubble. And as I did, I made a very important discovery, perhaps the most important discovery. I opened my hand. Inside, I found something, something that changed everything.I found a pin.Subconsciously, even consciously, I deflated my own dream. I choose to believe I knew what I was doing all along.

Here are my top 10 ideas, rather than advises, for a new PhD student. Ideas are not necessary in their order of importance.Art of negotiation: Do not be too agreeable with anyone, especially with your PhD supervisors. Work hard on your experimental hypotheses but do not abandon them just because your supervisor thinks it is a bad idea or because they come up with a better idea. Do not delude yourself into thinking that they know better because they are more experienced. They might but it is not their PhD. Take ownership of your PhD. Remember, the main goal of a PhD is to train yourself towards becoming an independent researcher by developing the necessary critical thinking skills. Challenge your supervisors on their stances, obviously in a reasonable fashion. Most principle investigators (PIs) enjoy and even encourage that sort of a challenge. Those who don’t are perhaps the kind of PIs who want someone (or rather anyone) to carry out the work on their ideas and not your ideas. Beware of such scientists.Always clarify your authorship position on any research article before starting the manuscript: You might hear many horrific tales about PhD students not being adequately or reasonably recognised for their contribution towards a journal article. Most of them are true (though be wary of some overtly exaggerated tales) This is more common in big labs where multiple PhD students and post-docs work on a research project. Problems arise when multiple people work under the fairy-tale assumption that they would be the first author on any research paper that arises out of this research. They carry on working without bothering to clarify this with their PI. You would be surprised to know that some post-docs have even claimed attribution/recognition to the Nobel Prizes awarded to their PIs! These controversies are difficult to regulate as no one can really isolate the importance of any particular individual’s contribution in a project. Nevertheless, do not be afraid to raise this issue.Never pay for your conference trips: Make sure that the conference expenses are covered through the funding that supports your PhD. If not, insist that the School/Department pays for the conference. Some PhD students do not realise that a conference is a place where the University gets a bloody good publicity of the research being conducted by them. The least they can do is to cover the conference expenses in exchange for the publicity. Do not delude yourself into thinking that a conference is a mere personal development or networking exercise and that it is only reasonable to pay for my own career development. That is plain wrong.Handling loneliness: You will sooner or later experience loneliness. This is not uncommon. For whatever it is, loneliness can be an asshole. It crawls around you slowly in an unsuspecting fashion and before you know, you are now under the spell of ‘Mr and Mrs Loneliness’. Personally, I have an opinion that the ability to fight this loneliness completely depends upon the biology of a person. For e.g. narcissistic individuals rarely would be affected by the bouts of loneliness as they will inevitably find a way to get out of it. Many other personalities though can get pretty damn stuck. But speaking objectively, you can beat loneliness by having a good set of friends, both within and outside the PhD community (which I will cover in the next point). Don’t feel embarrassed to seek help from professionals. Don’t feel anxious about taking anti-depressants (though I never did). Don’t feel ashamed to do simple things like calling your parents and expressing your angst. Don’t feel it is a waste of time to journal your thoughts down. Every little helps. Don’t suffer in silence.Having a good social circle: This includes not just the social circle involving your PhD friends but also the people outside of it i.e. with people who have nothing to do with research. Lets face it: during my PhD you are going to get tired of seeing the same faces everyday (especially if you work in a lab). Hanging out with the same people outside work hours might sound cool initially but guess what; even your social life would become meaningless after a certain period of time. Also bear in mind, too much social time with work colleagues can have some ugly unexpected consequences. Gossips might spread like a wild fire, drunk talks might become hard facts, small groups might polarise within larger groups, things might get personal etc. All these add up to your already stressed out PhD journey. Be good friends with everyone. Diplomacy is underrated.Self-time: Just be aware of this fact: burnout from work is real and mental health is important. Don’t always try to off-set heavy work with a under-cool social life. Do try and incorporate a ‘Me-time’. Your brain relaxes in different ways during socials and during self-times. During socials, it mostly forgets. During me-times, it mostly reflects. Always maintain a delicate balance between socials and self-time. This might sound a bit mystical but personally, the reflections that have occurred during my self-times have actually enhanced my ability to deal with heavy work; this happens through enhanced concentration, less noise within the mind and more room for tolerance. Explore it. Highly suggested.Overwork glorification: Don’t ever buy into the overtly pretentious made-up maxim of traditional academia that “If you are not suffering, you are not working hard enough”. Anyone who comes up with ideas like these must be asked to f**k off. Anyone who comes up slightly diplomatic versions of similar ideas must be politely requested to go back to the caves. You chose to do a PhD because you love research. You might be bad at it, never mind. But no one has the right to make you suffer in their pursuit of ‘transforming’ you into a good scientist. These tyrannical methods are archaic and highly overrated. The good news is that these views are no longer appreciated and even looked down upon within academia. Most modern scientists love to treat their PhD students as their colleagues or at least give them the permission to ask stupid questions and even fail in expensive ways with their own ideas without having to be afraid of being judged. Things have changed, we (modern scientists) are moving ahead.Acerbic academics: Some academics are plain acidic. This is obviously true with every other profession but the difference here is that academia is a sanctuary of the ‘intellectuals’. You are training yourself to be an intellectual. You probably would ignore a highly piercing remark about your intelligence from your friends, family or even a fellow PhD students. But when a similar remark is made by your mentor or by any other academic whom you highly venerate, you are more likely to engulf their remarks without challenging its veracity. I am not referring to the academics who like to talk to their students in an informal manner, filled with humour and sarcasm. I am referring to the academics who have a clear goal of damaging the self-esteem of anyone who challenges their ego. Stay away from such ‘intellectuals’, no matter how smart they are.Avoid comparison: A deep seated problem of the human psyche is comparison with fellow beings. It is quite rampant in academia too. It is very common to hear someone lamenting, “Oh, I am in my third year of my PhD and I don’t have a single publication but Sarah who is just in her second year has two publications already! Either I am not working hard enough or my project is crap.” Everyone’s project is unique and although all PhDs are roughly 3–4 years long (in the UK), the nature of the work plays a huge role in the output time. Some projects are so high-throughput that you can obtain a gamut of data in just a few weeks and start writing a paper. Some projects are so novel and multi-disciplinary that it might take an entire year just to lay down the ground work. It might not pay off immediately but sooner or later it will. The more riskier the project, the less sooner the output time but the more novel will your project be. In deed, many funding bodies these days are encouraging ‘high-risk, high-gain’ projects whose ambitions are so big that there are no clear, laid out milestones initially, only an huge overall vision of the project. So, do not place a huge amount of emphasis on the publication record of your colleagues. It is okay if they have a Nature paper already, appreciate them for their work. Your time is yet to come and perhaps, in a different place?Explore: Don’t stick to the vicious cycle of work-social-home-work-social-home. Explore, explore, explore! Read books that challenge your assumptions, talk to people whom you would avoid on a good day, Travel along to places, ask random people of you could click their pictures, do something that freaks you out (for me it was asking someone out:massive fail!), draw your bizarre ideas on a paper, attend strange workshops, do lots of volunteering. I learnt so much about life during my volunteering times with a homeless outreach group. It made me a better scientist as well. Life is all interlinked. Do not deceive yourself into thinking that tunneling all your energies into your subject matter will land you a Nobel Prize. In fact, a study has shown that the more eclectic you are outside work, the more chances of you bumping into a Nobel Prize. For e.g. scientists who could dance, play piano, write stories etc. were more likely to become Nobel laureates than their ‘work-eat-sleep-work’ counterparts. Go on, who said you can’t have a fulfilling life as a scientist!This is not an exhaustive list obviously. I would like to add more points as and when possible.Enjoy!Pranav