Let me tell you something about TPS reports …
A few years ago the contact center I then worked at slashed their workforce by about 25%, it was the start of the Great Recession. This proved almost instantly to be a disastrous move. Call waiting times were already pushing 15 mins or more and even the fastest ticket turn-around times were measured in hours. But after that lay-off, literally overnight, the hold time went to hours and the ticket-time went to days. The hold time soon grew so great the system couldn’t hold that many open lines in queue, it began crashing periodically and hanging up on people, often after they’d been on hold for an hour or more. If you think they were in a good mood when they finally got through, think again. The crash in the phone system was nothing compared to the crash in customer satisfactions scores that soon followed.
It was a full-blown debacle, something had to be done, so senior management came up with a brilliant solution: hang up on 90% of the calls after they got through. Just dump them out of the system! Customers didn’t even get a recorded message inviting them to visit the website at first, not at first, for some bizarre reason that took another few weeks to get set up. So, just BAM, it would hang up on you. Not only was no one fired, as I recall a round of promotions and title upgrades swept through senior management, confirming the whole episode from start to finish had in all likelihood been orchestrated at the highest levels.
This is the reality in the private sector, they penny-pinch and cost-cut to the absolute bone, often with significant, dire consequences for service and support. After 2008, it got much, much worse. Those departments found themselves among the most popular to gut, outsource, or simply eliminate. They were easy targets. In part because customer service and tech support are generally staffed by the worst paid, least influential, short-term and part-time employees in an entire company.
But the crew on Morning Joe isn’t having any of this. What’s really funny is some of the same people who voted 40 times to repeal Obamacare, then did their best to defund and hobble the law at every level, up to and including shutting down the government, are now whining the website wasn’t properly funded and the WH should fly out to Silicon Valley and hire billionaire developers to speed it up.
What’s funny about seeing the website under attack is how much better those efforts would have worked had Ted Cruz and his buddies not provided so much cover. But even if it had been front and center starting at the first of the month, I’m not sure it would matter to many people outside of the usual suspects, the clowns who get a woody every time they think they have a new scheme to with which to attack Obama.
The premise of these attacks, that your average non-partisan American has never encountered a slow website, and/or will therefore feel outrage over same even though they will probably never use this one, sounds almost comical. Hell, the credit card only vending machine in my apartment complex just glitched out on me longer than it took Healthcare.gov to load right before I hammered this post out. And for what it’s worth, I remember when Amazon started, I remember when iPads and iPhones came out, it wasn’t that long ago after all folks, and they crashed and sputtered and had all kinds of issues like everything else.
But on Morning Joe, all those sites and systems and devices worked flawlessly. On Morning Joe, a slow government website, and only a slow government website, is on par with the slow government response to Hurricane Katrina.
There was a time when I would follow this up with the obligatory warning that the wingnut Wurlitzer can make a mountain out of any molehill. But watching Scar try to inject his talking points into this discussion, I think even that once admirable spin machinery has been weakened over the past few years.
I wouldn’t be at all surprised if the website is suffering some level of DDoS attack.
Obviously these folks do not play online video games like Diablo 3, Simcity, or Grand Theft Auto Online. If they did, they’d know that rolling out an online service only to have the anticipated number of clients crash the servers is ‘running it like a business’.
In all fairness, they only had ten years to work out D3 …
It should be called the Morning TED: a bunch of self-serving windbags spouting “solutions” that have no basis in reality, and they expect everyone else to implement them, not do it themselves.
When the recession hit I was working at a large company in one of their tech support call centers. We had probably 20 different customers who had outsourced their service desks to us, each with their own pool of employees dedicated to the specific queue (since every company had different polices and different software that we had to support to varying degrees).
When the recession hit they decided that the best way to save money was to shave the number of employees down to about 20%-30% below the number required to handle the inbound calls (like Stephen describes here), and in some cases below the number of employees they were charging the customer for (in the case of the queue I was working on we were billing for 8 full time employees but they cut us back to 6). Then their brilliant plan was that they would take and cross train about 20% of the employees left on each queue to work on other queues and just shuffle people around as queues got busy. The training consisted of “Here is a temporary phone log in you can use so you can get calls from this queue. If you have questions read the knowledge base.” The biggest problem with this was that all of the queues would get busy at about the same time every day, usually from about 8AM-Noon, so people would get shuffled around, and every queue would usually end up with about the same number of people they had to begin with, but now about 25% of the people taking calls for any given queue didn’t know what they were doing.
The scale and complexity of the ACA exchanges needed is several times as complex as what even Amazon has. The exchange system is very large, and it needs to interact securely and accurately with scores of different databases. Some of those databases are proprietary systems used by insurance companies, and some are government run systems that haven’t been upgraded in decades because congress won’t allocate the money.
As simple as it was in comparison I remember it it took years for on-line ordering to work well. It is all of three weeks in. Big systems take time to get right.
The ACA exchanges are not actually being programed or set up by government workers. Almost all of the work has been contracted out to private industry. Any failure is as much a failure of the private sector as government. It wasn’t government workers who claimed to have stress tested the system before launch. This is a private industry failure to meet contract objectives. Despite being paid very well.
A friend in the programming business mused that the work was likely shoddy because the heads of the corporations hired to do the work are all conservatives who thought the ACA would be killed in return for a budget deal. No need to work hard and do your best on a system that will never be set up.
Even if needed the politics of the bosses contradicts the goals of the system they are hired to create. There is every reason to slow-walk and foot-drag. There is always another excuse. Any congressional investigation will be done by like-minded congressmen who have every reason to excuse the corporation’s unwillingness to fulfill the contract. Obama is the designated whipping boy so any failure, or even outright sabotage, has no down side.
In fairness, there are some legitimate reasons why the rollout of healthcare.gov has been rough. For one thing, it’s having to handle more volume than originally expected due to so many Republican-dominated states refusing to set up their own exchanges.
Also, I heard a report this morning that some 50% of hits are “tourists” — people who already have insurance just surfing to the site out of curiosity.
*Raises hand*
I’m not even American and I wanted to go and see it, to see what socialism looks like.
and I’m betting that in order to supply the resulting rate table the system has to refer to the insurance company computers each and every time, instead of swallowing the rate table each morning (or noon or night or whenever) and doing the table lookup locally. This will add to the service time: time for the (jammed) links to get the request through, time for the mainframe to service the request, which includes the time for disk accesses to the monstrous ins co databases, time to get the answer back, time to reactivate the thread that originated the request, (repeat for each of the insurance companies in the state in question…)