Steno V Digital (Archive Post)

Consider this a gentle touch on an important topic. There’s been a true memetic shift in the way stenographers are interacting and spreading ideas. Content is honestly popping up faster than we can even really digest it, so let this post serve as a staging point for some of what’s happening this Court Reporting and Captioning Week 2019. This weekend I’ve had the pleasure of reading a flyer from the DRA in California (Photo Archive). Read about Idaho’s need for reporters (Photo Archive). Finally, got to see Cleveland Reporting Partners’ whole take on digital v steno (Photo Archive).

In very brief summary we are seeing many people put into writing what I have opined over Facebook. Yes, technology is amazing. But right now it struggles with certain things. It can transcribe one speaker quite well, but if you throw in some stray sounds or a second speaker, it can have a hard time. This makes the market for captioners and legal reporters a little more promising because we have the skill and training to give them what they need now and train others to do it. Make no mistake, there’s a big market in that, so if a company is having you train a digital, make sure you’re getting at least the next ten years of your annual income upfront.

Technological growth is no longer exponential. Don’t get me wrong, it’s impressive. But until Quantum Computing is cheap and accessible there are probably things we won’t see, like a JARVIS-like AI. We will see imitation AI, that’s for sure, but there is an indeterminate clock on when we will see quantum tech. The running idea and current study that’ll probably lead to true ideas is machine learning. This takes data training sets, like pictures, or recordings, or text — whatever it is programmed to take — and it takes that information and uses it as a basis for its decisions. Sometimes this is entertaining. Sometimes this goes horribly wrong. The bottom line is it is limited by the speed at which it can process its training data and the speed at which it can retrieve that information for later.

I imagine that the training data set for an AI to “do depositions” would look something like recorded depositions paired with their transcripts. There are three big hurdles there, building the training set, processing the training set, and retrieving the right data when it’s time to “do deposition.” In a classic computer we have, in very laypeople terms, little transistors firing on and off to tell the computer what’s going on. Tech is running into a problem where it can’t get these little nanotubes much smaller, and making bigger processors absorbs more electricity. For example, I wrote a Fibonacci-generating program. The basic concept is every number adds itself to the number that would come after it. The computer is happy to make these calculations, but very quickly, the processing power needed to calculate these numbers begins to run dry, and the files we store these numbers in become too large to be opened on a weak laptop. The simplest algorithm in existence busts up a classic computer. This is probably the trouble they have making something that can seamlessly listen to people and transcribe, the computer just doesn’t have the power to process it quickly. Look how long it takes videographers to burn disks or Go To Meeting to process audio. Now imagine adding another layer where the machine is transcribing everything perfectly. In Quantum Computing they’re talking about these very small units being able to calculate everything at once, or large batches of things at once. If they crack that, we’re probably back to exponential technological growth.

In the meantime, fight for your jobs. Fight for market share. It’s not a question of whether we’re outdated. Today the answer to that is no. What matters heavily is perception. Perception can change outcomes. One of the most effective tactics in war has been to get the enemy army to rout, and that’s exactly what digital reporting advocates are trying to get you to do: Give up and go home without a fight. Don’t buy into it, make the technology prove itself. Even the worst stenographer puts in words four or five times faster than the average typist, yet there are still typists.

Keep competing. We are well on our way to winning this thing.