Monday, October 20, 2014

GAW and ZenMiner: Creating and Fixing Bugs

In case you've missed it, there's been a ton of ZenMiner/GAW craziness the past week. The short summary is that GAW rolled out an update that allowed the selling of Hashlet Prime (finally), along with a few other modifications. In the process, Hashlet Prime was also made available for purchase again ("Oops! It was a bug!" -- never mind the winking smiley face from jmordica and talk earlier that day saying Primes would be brought back on the market for a short time), but more importantly there was a glitch where buyers would sometimes get a 5MH or 25MH Prime instead of a 1MH Prime.

Some people figured out how to work the system to buy these "discounted" Primes, and they flooded the market with low-cost Primes, purchased more of the glitched codes, etc. The result was that they ended up shutting down ZenMiner along with Hashtalk while they worked through everything in order to roll back purchases -- and there were plenty of locked user accounts and such to go along with the mayhem. It looks like things are mostly back to "normal" now, and for many users (including me) the impact wasn't too significant. Still, this is part of the bigger problem GAW faces.

In short, the schedule of releases and the testing of the software continues to be far too chaotic. When you say you're removing a product from the market and then it comes back, that creates confusion and unhappy customers. Quite a few users for instance bought $70 Primes after the "no more Primes" announcement, only to find that twelve hours later $50 Primes were back -- maybe only temporarily, but still, that's a horrible form of "service".

There's talk of better testing going forward, with community members becoming part of a closed beta, but that's not really a proper solution. The fact that so many bugs have slipped through simply means there are not enough paid developers and testers to keep up the schedule GAW is trying to enforce, and adding more users and features only exacerbates the problem.

It reminds me of a horrible software developer position I had for a couple years back in the late 90s; the CEO would come visit our office about once a month, and every time he stopped by he had some awesome new "vision" of what sort of software we needed to make. He would give us a rough outline that we'd spend two or three weeks trying to formalize, then we'd start working on some early prototypes, and a week or two later he'd look at the prototypes and say, effectively, "That looks great; now let's add some other cool features...." It was developer hell, and nothing ever got finished. Over the course of a year the products we were working on became increasingly buggy and unstable, our working hours skyrocketed, and by the end of my time there people were ready to leave in a mass exodus. Oh, and the company never actually shipped a single piece of working software.

GAW has the benefit of working in the web space and selling something that people are already using, but rather than continually promising new features it would be of great benefit to actually get all of the current features working properly, and then slowly add one or two items at a time, thoroughly testing each one. And make sure when things are supposed to be deactivated, they don't come back -- or things that should be activated stay working. Otherwise, in another year the whole platform powering "HashBase" or whatever will be a complete mess, hackers will make off with tons of Bitcoins, and the company will fade into the annals of Bitcoin oblivion.

No comments:

Post a Comment