Wednesday, November 8, 2017

What to do when you've got a virtual scrum team

Scrum and Agile are suddenly popular in Asia, and because a lot of companies take on outsourced projects, they usually have virtual teams, with members working in Asia, the US, and Europe. In the Agile meetups I've attended so far in this region, one of the most-asked questions is about how to deal with virtual teams. What's a poor Scrum Master to do?

The usual "Scrum answer" is to push for a co-located team and discourage virtual teams. There are many reasons behind this preference, but they are sometimes not obvious. So let's start understanding the problem from the perspective of the Values and Principles in the Agile Manifesto.

The Agile principles recommend:

  1. Customer satisfaction through early and continuous delivery (#1);
  2. Daily interaction between business people and developers (#4); and 
  3. Face to face conversation (#6). 
Note the order of priority, from highest to lowest. Reflect on what gets sacrificed and how customer satisfaction is put to more risk when you have members who are not co-located. Here are some thoughts on this:
  1. Face to face conversation is best because you get shorter feedback loops. Being on the same location allows you to draw collaboratively while talking, use post its or index cards, or see visual cues from people so you could quickly form consensus and decrease misunderstanding. 
  2. There's nothing more efficient and effective than walking to someone and conversing with them to resolve a question. 
  3. Face to face conversation is derived from lean thinking, which recommends it to reduce wait time. If a developer waits for the PO to respond to a request to review the UI of a new web page, that wait time could have been used to work on another task. But working on another task entails switching cost, too. Waiting and switching are waste. 
  4. Lean thinking also views anything that impedes early and continuous delivery as wasteful. In some teams I've worked with, the only useful overlap was a maximum of 2 hours. Their day was ending as ours was just starting. You could take advantage of this situation by handing off tasks to the remote team -- but then again, handoff is waste.
If you're stuck with virtual teams, then the principle of daily interaction must not be violated. I would even recommend that since you cannot have face to face conversation, you must compensate by increasing the frequency of communication. Here are some rules of thumb for communication preferences:
  1. Prefer face to face conversation.
  2. If that's not possible, use Skype or Google Hangouts with video (it's important to receive visual cues, so invest in a good line that allows streaming video).
  3. Compensate for the lack of face to face conversations by over-communicating. 
  4. Discourage instant messengers or email for conversations that need quick feedback. If you use email, use only one thread to avoid multiple messages that are hard to sift through for information.
  5. Compensate for geographic separation by using tools to facilitate virtual teams like Slack, Trello, Skype, Google Hangouts, and Google Docs. Google Docs have neat features that let you collaborate more visually while working concurrently on the same document (I've had a bad experience with Microsoft OneDrive, so I don't recommend it).
  6. Combine the use of these tools so that they make up for each other's limitations. 
  7. Maximize screen-sharing. This helps you to work literally on the same page when reviewing something. 
  8. Trust between the local and remote team members is a key factor. The remoteness and cultural differences provide invisible but critical barriers to smoother collaboration. Bring the remote member to your location so you could work together. This would be great to do for one sprint, or if budget and time are issues, bring the member at sprint review, retro, and planning time. 
If you follow the recommendations above, sooner or later, the team will come into one of these possible conclusions. Either the team will get tired of overcompensating for being fragmented (in which case transition to two separate co-located self-sufficient teams), or they will learn to work remotely with the rest of the team and say the trade-off of over-communicating is worth the benefits of working with the remote members.


Thursday, October 12, 2017

Visualized: Philippine Presidential Elections 2016

Visualizing data helps people to quickly find patterns and trends in the numbers. Here are statistics from the 2016 Presidential Elections.

Monday, October 9, 2017

mBot arduino navigation

The mBot is an Arduino-based rover robot. This video shows how it navigates away from obstacles using the ultrasonic sensors up front. The challenge in programming the obstacle avoidance was to find the a) best distance at which the mBot rover should stop moving and b) by how much it should reverse-drive to change course.

Make (a) too short and it will get trapped in tight spots (there's a point in the video that this happens). Make (a) too long and it will not move at all especially in smaller places. Make (b) too short and it will go nowhere. Make (b) too long and it could back up into other obstacles.

The sensors look like cylindrical eyes. They emit ultrasonic waves that bounce off obstacles. Because of this, the sensors could actually measure distance.

I verified the distance measurements using a tape measure. Not that a tape measure is very accurate, but the distance the ultrasonic sensors reported were actually quite good. I lost my notes on this one, but I will soon reactivate this robot and post the measurement experiments here.

Wednesday, April 5, 2017

Bad UX design could lead to security holes (BPI Express Case)

How bad UX design could lead to security holes (BPI Express Online Case Study)

Like most banks in the Philippines, the Bank of the Philippine Islands (BPI) is a major cause of migraine for customers. Its user experience design (UX) is badly done and the web service is unreliable. Both of these headaches lead to a third, much bigger problem: security holes. This probably explains why some BPI customers recently fell victim to phishing scams where unwitting victims divulged their passwords to a website pretending to be the BPI's online service. 

BPI's UX design problems 

Due to a badly designed user experience, the BPI website creates problems that punish the user. Forgot your password? You would have to call customer service to get a password reset. Maybe they think that this is more secure than doing the password reset online, but really, it's the same thing or even worse -- the human to whom I'm giving my details could jot down my credentials.

Let's discuss the first issue: bad UX. By this I mean not only the clunky design of its user interface that confuses users but also the general lack of empathy with the customer especially in times when the site is down.  

As I write this, the BPI website is, yet again, under maintenance. The screenshot below is the usual message that customers get when that happens (and all too frequently at that, if I may add).


There is no information on how long the maintenance will take. Should I just keep clicking the refresh button?

In the homepage, it turns out that there's an inconspicuous and cryptic message saying "Electronic Channels Upgrade Advisory":



Does the extra click I spend give me more helpful information? Nope. Check out the resulting page:

It's just telling me what I already know. At least tell me how long I should wait or when to try again. The message is as helpful as a flood warning a day after the flood has submerged the town. 

My retries produced a more problematic error message that gives away database details:


Maybe I stumbled upon a critical procedure during the maintenance. But shouldn't the BPI team be cautious about this and prevent this kind of sensitive error message to be published? 

UX design problems create security holes

The examples above are just one aspect of BPI's UX design issue. What's more problematic is that the UX design could lead to major security holes. For example, BPI requires users to change passwords once in every three months. You cannot repeat passwords so you have to create a new one every quarter. 

Can you imagine the burden of remembering a new password every three months, especially since your password cannot be a string of letters? This unreasonable policy forces people to write down their passwords -- which defeats the purpose of a strong password in the first place. 

Another consequence of frequently changing passwords is that people will forget their passwords and would have to call customer service, which leads us to another security hole. BPI does not have an online password reset service. If you forget your password, you have to call a customer representative to reset your password for you. Adding a human introduces a weakness in the security chain. Before the customer representative resets your password, you must answer security questions that force you to divulge private details to a stranger. 

In forcing users to call a human instead of offering an online password reset service, BPI probably thought it was creating tighter security. Yet this did not prevent users from giving away their passwords from a phishing scam, did it? 

Many know that UX design is important to any piece of software to improve usability. But as I explained above, UX can also lead to security problems that could be exploited by online criminals.   

The case I outlined above also shows how corporations build their security measures based on outdated assumptions. But that is for another blog which I will be writing soon. 

Saturday, March 25, 2017

Yummy Filipino Adobo

Filipino Adobo stores well and gets better as you keep it for several days. Store it in the fridge and fry/heat it up as you consume. The recipe/s below combines different recipes I've learned from friends and relatives. If you're new to cooking adobo, try the Basic Prep instructions first.  

The Kapampangan adobo recipe varies from the one I record here. When I get the time, I'll also write that one down. There are many variations of Filipino adobo, as much as there are Filipino families, I bet. The word adobo is Spanish for sauce or marinate, so don't be confuse Filipino adobo with the Mexican version.  

Got some tips? Share them in the comments below. 

Enjoy!

Yummy Filipino Adobo

Ingredients

  • 2 lbs - pork (belly and/or ribs is great, but any cut is okay)
  • 1 cup - soy sauce
  • 1 cup - white vinegar (rice vinegar is okay)
  • 4-6 pcs - bay or laurel leaves
  • 1-2 tbsp -  ground/cracked black pepper
  • 1-2 bulbs - of garlic - with cloves crushed, peeled, and sliced (the more garlic, the better)
  • 1 bulb - onion, sliced (optional, see Instructions for Tastier prep, below)

I. Basic prep

  1. Rinse pork and drain away water.
  2. Put pork in a pot and mix in the crushed sliced garlic cloves with the pork.
  3. Pour in soy sauce and vinegar.
    1. Note: If you want the sauce on the saltier side, add more soy sauce.
  4. Add black pepper.
  5. Crumple/crack the laurel/bay leaves and add them to the pot.
  6. Put in stove, bring to a boil, then lower the temp to a slow boil for 1.5 to 2 hours (the longer the cook time, the more tender the pork will be).
  7. Continue the slow boil until the meat is tender, has soaked in the sauce, and the sauce is reduced to a thicker consistency.
  8. Taste and keep cooking until you’re happy with the taste.
  9. Serve with rice.

II. Tastier prep options

Try each of the following options separately, from top to bottom or do them all at the same time. 
  1. Marinate the meat in the soy sauce-vinegar sauce for 1-3 hours before putting on the stove. This will help the meat absorb the sauce even more.
  2. When the meat is tender and has started to absorb the sauce, take it out and fry it. Drizzle some sauce with cooked garlic on the meat while frying. Meanwhile, let the sauce simmer and reduce further in the pot. When the sauce is ready, put the meat back and serve.
  3. Instead of frying the meat, try baking, broiling, or grilling it.
  4. Substitute 1 lb of chicken instead of pork. Note: chicken softens faster than pork, so you can add the chicken later. Marinate the chicken in the sauce so it absorbs the sauce (see II-1). 
  5. Fry some potato wedges (season lightly with salt and pepper) and add them to the pot when the adobo is ready.
  6. Onion - make a bed of sliced onions in the pan before adding the meat. This makes the adobo even tastier, but you’ll need to reduce the sauce even more as the onion waters down the sauce if not cooked well. To fix this, just reduce the sauce some more. The onion will dissolve and thicken the sauce. Try also putting some of the cooked onions with the meat when frying/baking. 
  7. After a few days, fry the meat with some sauce. As you fry it, pull the meat apart. This makes for some great tasting pulled pork/chicken. 



What to do when you've got a virtual scrum team

Scrum and Agile are suddenly popular in Asia, and because a lot of companies take on outsourced projects, they usually have virtual teams, w...