Thoughts on AWS OpsWorks Application Management

RightScale’s almost-relevant opinion on OpsWorks. Amazon’s move here is to create momentum for engineers, small companies, and those exploring the idea of moving some or all of their infrastructure to the cloud. I don’t think Amazon had “IT departments” at large organizations in mind when they acquired Scalarium. I also don’t think Amazon is particularly interested in the “multi-cloud” movement, seeing as it’s in their best interest to be the only cloud (and they’re doing a damn fine job of it).

GrayBit Retiring

It’s been quite some time since Mike Cherim and I put together GrayBit, a web service that converts any web site into a grayscaled version. The aim behind GrayBit was to provide a helpful service for anyone to evaluate their web site and identify whether the site needed improvements to its contrast or whether folks with color-blindness would be able to effectively browse their site. Although GrayBit has received some attention over the past several years around the web, it’s an ongoing expense to maintain and manage. Mike has graciously provided the hosting and partnered with me on the project for quite some time, and for that I’ll continue to be grateful. Hopefully GrayBit has had some positive impact on the web and accessibility during its life.

That life has come to an end, so to speak, as Mike and I have decided to retire GrayBit officially. The service will cease to be available on March 3, 2013.

If anyone is interested in scooping up the tech behind the service, please contact myself or Mike about it. If you’re asking yourself, “Why don’t you open source it?” the reasons are pretty simple: the technology simply where I’d like it to be in terms of quality; the contribution to the open source community would be more useful if I hadn’t left the core software without improvements or revisions over the past several years. That work needs to be completed before the tech is ready for the open source community.

Thanks for all the good times, GrayBit. Good bye.

Wednesday, Indeed hosted an event called “From 1 to 1 Billion: Evolution of a Document Serving System,” in which Jack Humphrey and Jeff Plaisance walked an audience of entrepreneurs and techies through the evolution of the Indeed business and how their document serving system’s design has grown to meet demands. Jeff dove into the details of the data infrastructure Indeed developed in-house called the LSM Tree, a data architecture that allows O(log(n)) look-up of keys (documents) in a hybrid in-memory / on-disk index. Jeff also compared Indeed’s LSM Tree to both Google’s LevelDb and Tokyo Cabinet, both of which failed to perform as well as the LSM Tree, according to Jeff’s own benchmarks. I was told they’d be posting the slide deck online, but it seems they haven’t gotten around to it yet. According to Indeed, they’ll be continuing to host talks on technology. If I remember right, another event was slated for March 27. I’d be really excited to learn more about Boxcar, as it looks very interesting. Update: Indeed posted video / slides to their engineering blog.
Pop-upView Separately

Wednesday, Indeed hosted an event called “From 1 to 1 Billion: Evolution of a Document Serving System,” in which Jack Humphrey and Jeff Plaisance walked an audience of entrepreneurs and techies through the evolution of the Indeed business and how their document serving system’s design has grown to meet demands. Jeff dove into the details of the data infrastructure Indeed developed in-house called the LSM Tree, a data architecture that allows O(log(n)) look-up of keys (documents) in a hybrid in-memory / on-disk index. Jeff also compared Indeed’s LSM Tree to both Google’s LevelDb and Tokyo Cabinet, both of which failed to perform as well as the LSM Tree, according to Jeff’s own benchmarks. I was told they’d be posting the slide deck online, but it seems they haven’t gotten around to it yet.

According to Indeed, they’ll be continuing to host talks on technology. If I remember right, another event was slated for March 27. I’d be really excited to learn more about Boxcar, as it looks very interesting.

Update: Indeed posted video / slides to their engineering blog.

The Bag

A few weeks ago, Kristyna told me she wanted to look for something new in her career. For quite some time, Kristyna has been a child caretaker working out of various clients’ homes. Although she has a lot of qualities that allow her to excel at taking care of children, she expressed interest in changing things up and pursuing a new direction.

So she told me she was going to quit her job. My initial reaction, of course, was “Please don’t.” But she did anyway, and I’m glad she did. She took the plunge, not knowing what lay in her future, but confident, optimistic, and not taking “no” for an answer. It’s the kind of quality that you find in successful people.

Thus, the job hunt began. Kristyna spent a considerable amount of effort looking for something fresh and interesting. Job hunting is both painful and stressful, and after a while, every opportunity starts to look like a hoax, sham or waste of time. After looking for about a week, Kristyna came upon an account manager (a.k.a. “Charity Rockstar”) position at a company called Goodybag, for which she applied and was subsequently hired.

Goodybag is a local Austin loyalty startup with a charitable twist. The company provides a physical card (keychain size that slips into a plastic case for conversion to a full size card), which is used to “tap-in” at local businesses by pressing the card against the Goodybag-installed Android tablets. These tap-ins equate to donations to local charities. Each time a person taps in, Goodybag donates an amount (more if you signed in with your Facebook account) to the charity chosen by the partnering business. In addition, tap-ins get you karma points, which can be redeemed at any participating business for various coupons or incentives.

Anyway, startup stuff aside, I’m exceedingly proud of Kristyna because she (a) didn’t listen to me and took the risk and (b) found a great position working at a great startup, a great move for her career. She seems to be doing well, having worked there a week now, and enjoying the workplace tremendously. As with anything she sets her mind to, she’ll successfully knock it out of the park!

Here’s to you, Beautiful!

How can mint.com be improved?

My answer to the question, “How can Mint.com be improved?”

Answer by Jonathan Fenocchi:

I’ve been using Mint for 1810 days (according to my profile), which equates to roughly 5 years. It’s been an amazing service and product since its inception, all the way up till they got purchased by Intuit, which I believe was in 2010. That’s when I mostly stopped using the service.

I have tried to get in the habit of using it and making it work for me, but I find myself spending far too much time fighting with the service than actually getting anything done. It’s faster and more accurate for me to use a simple spreadsheet to calculate my budget, track bills and income, etc.

Since the question specifically asks how Mint can be improved, I will (as I have done several times now) express my discontent with the service as it is. I’m including this list in order of priority (i.e., most critical issues with the service as-is).

  1. Timeliness. I’ve said it before: getting a text 2 weeks after I pay rent that a few thousand dollars left my bank account scares the living daylights out of me. As a result, I now tend to ignore any notifications I receive from Mint, discounting them as false-positives. The inevitable result is that when Mint does have some useful information, say someone stole my credit card and went on a shopping spree, I’ll just see the text (2 days late!) and assume it’s Mint sending me delayed or irrelevant messages again. Thus, to reiterate, Mint’s business is (personal finance) information, and information of this nature has deteriorating value over time. When information isn’t provided promptly, it ceases to be useful; and irrelevant information is worse than no information.
  2. Mint seems to get confused that I pay rent on the 1st of the month. Half the time, it decides to show that I’ve paid rent twice in one month; the rest of the time, it decides to show I’ve not paid rent at all. I realize this is may be a difficult technical problem to solve, but I’m sure Mint has smart engineers slaving away. None of my data is useful if it’s inaccurate or I can’t trust it. So if I want to look at spending trends, they’re horribly skewed. There goes the Mint’s value proposition, all in one go.
  3. Connectivity issues with my bank happen all the time, always when I’m not around. Mint doesn’t let me know about this until I login to the web site and look at my accounts. I just logged in while writing this answer to find there was a “temporary” problem connecting to my bank account, and that it was last updated 2 days ago. Is it really that hard to design and implement this kind of connectivity?
  4. Speaking of bank accounts, Mint still has several old bank accounts under my main bank account, which I have closed upwards of 3 years ago. I can’t delete them or clean them out without deleting my entire account and losing all my (virtually useless) history.
  5. I have 2 vehicle loans. Mint seems to think that I own them. There’s no intuitive way to add a vehicle and identify the portion that is on loan (debt) versus credit. In addition, the value of the vehicle isn’t updated for me based on NADA or KBB, it’s just static, as if the value of my car will never change. This paints a very unrealistic picture. I could see plenty of opportunities to make some money by providing better integration with loans in this area. Plus, it’d be nice to tie in my monthly car note to a specific vehicle, track my spending on it, the interest, et cetera, and get a pretty good idea on what’s the best way to pay it off based on those factors. If Mint is for personal finance, why can’t I see all of my finances, including assets?
  6. I’ll be damned if I ever try anything Mint recommends again. It’s all so that Mint gets paid for some targeted segment of users. Mint recently informed me that I should update my credit score because it was out of date. It then proceeded to recommend Free Credit Report & History (yes, the one with the goofy commercials). Seriously?! The service is practically bogus and very dangerous — it’s your classic “free” telemarketer borderline scam, designed to rope you in and then charge you out the ass because you didn’t read the 400 pages of legalese. Regardless to say, I would have much rather paid Mint the $150 of fees for a good service without a foul, dishonest recommendation engine.
  7. Why can’t Mint tell the difference between cash, credit, and debt? They are different! Mint only shows “cash” and “debt,” and of that debt, it only shows “credit card” debt. I have debt that isn’t a credit card — say, my auto loans or that Bowflex I bought my wife last Christmas.
  8. The service is slow. I know this sounds like me complaining, but as a devops engineer in a world where cloud computing is all the rage, I know that commodity hardware is cheap and it’s far more valuable to provide a responsive, fast service for customers. And it’s not that hard. I’ll include in this point that the UI is also sluggish (even on a 15” MBP w/ Retina and 16GB/ram) compared to practically anything else out there. And why are we still using Flash, again?
  9. The Transactions section of Mint could use significant improvement. The “tags” system needs to be revamped so that the UI isn’t confusing, difficult, and simply not worth the time. The search function sucks. Even worse, you can’t select a date range, making filtering search results difficult and frustrating. The “manage transaction rules” link is hidden at the bottom of the page. You can’t just export some subset of transactions, like a search result or date range, you have to export all of them.

As an aside, if anyone has contacts or works at Mint, please feel free to get in touch with me. I would really like to continue using Mint, but I frequently find myself searching for the “next” Mint on various app stores (iOS and Chrome Web Store). I think Mint has a tremendous amount of unrealized potential and it never scratched the surface because big guns came in, swooped it up, and total complacency ensued on the part of the leaders at the helm. My intention isn’t to libel Mint, but hopefully someone can put a fire under their ass to get back to working on the product and stop watching the product degenerate slowly. What happened to the innovators at Mint?
View Answer on Quora