Overstock’s Blockchain Patent Problem

11 Aug Overstock’s Blockchain Patent Problem

Overstock garnered widespread publicity about a recent patent to record trading of tokenized securities on a public blockchain, awarded to their tZero subsidiary. See Coindesk, TokenPost, The Block Crypto, Tokenist, CoinTelegraph, Street Insider. The articles are essentially this warmed over press release from tZero.  Look carefully at the latter. Where is this patent? Why doesn’t the article link to it?

Here is the patent at the official US government website. So just as I gave you a link, why didn’t they? Maybe they did not want you confused with the details. Or maybe they just did not want you to read it. It is a low quality patent.

In the patent page, scroll down to the Claims. If you get a patent, that means you get claims. These are the cutting edge. When a company says another firm infringes their patent, they say that the firm infringes at least 1 claim in the patent. The problem in the patent is claim 1. Here it is in full glory:

Claim 1. A computerized method comprising: receiving data items during a plurality of first reference level time intervals of a first reference level; generating a genesis hash of genesis data by performing a first genesis hashing function on the genesis data; generating a hash for each of the data items in the order in which the data items are received by performing a first hashing function on each of the data items with a corresponding timestamp; generating an initial first reference level time interval hash of a plurality of first reference level time interval hashes for an initial first reference level time interval of the plurality of first reference level time intervals by performing an initial second hashing function using the genesis hash and the hash for each of the data items in the initial first reference level time interval according to respective timestamps of each of the data items; generating a first reference level time interval hash for each of the first reference level time intervals other than the initial first reference level time interval, wherein each respective first reference level time interval hash is generated by performing a second hashing function using a first reference level time interval hash of a first reference level time interval immediately preceding the respective first reference level time interval and the hashes of the data items in the respective first reference level time interval according to respective timestamps of each of the data items; generating an initial second reference level time interval hash of a plurality of second reference level time interval hashes for an initial second reference level time interval of the plurality of second reference level time intervals by performing an initial third hashing function using the initial first reference level time interval hash and the first reference level time interval hash for each of a predetermined number of the first reference level time intervals; generating a second reference level time interval hash for each of multiple second reference level time intervals other than the initial second reference level time interval, wherein each of the multiple second reference level time intervals is comprised of a predetermined number of the first reference level time intervals, wherein each respective second reference level time interval hash of the multiple second reference level time interval hashes is generated by performing a third hashing function using a second reference level time interval hash of a second reference level time interval immediately preceding the respective second reference level time interval and the first reference level time interval hashes for the first reference level time intervals within the respective second reference level time interval; and generating an initial third reference level time interval hash of a plurality of third reference level time interval hashes for an initial third reference level time interval of the plurality of third reference level time intervals by performing an initial fourth hashing function using the initial second reference level time interval hash and the second reference level time interval hash for each of a predetermined number of the first reference level time intervals; generating a third reference level time interval hash for each of multiple third reference level time intervals other than the initial third reference level time interval, wherein each of the multiple third reference level time intervals is comprised of a predetermined number of the second reference level time intervals, wherein each respective third reference level time interval hash of the multiple third reference level time interval hashes is generated by performing a fourth hashing function using a third reference level time interval hash of a third reference level time interval immediately preceding the respective third reference level time interval and the second reference level time interval hashes for the second reference level time intervals within the respective third reference level time interval.

The patent has 9 steps, separated by semicolons. Many steps are very detailed and long. Not good.  A competitor to Overstock who wants to legally use the claim without paying a royalty can do 8 of the 9 steps and they can do other steps not in the claim. But if they can find one step in the claim that they can avoid then they do not infringe.  This includes being able to mod that claim and still get a similar overall result.

The number of steps and the details in the steps narrow the scope of the claim. With patents there is an inverse property. If you get a claim short in number of steps and details in the steps, then it has broad scope. Harder for others to work around. An experienced patent attorney can glance at the first claim and see the tell.  So can you, even if or especially if you are not an engineer. This is a business analysis of a patent.

Yes, the inventors of this patent might insist that it cannot be worked around. Maybe they are right. I have not done that analysis. But the quick method I described is used by many patent lawyers and it is common advice to inventors:  short is good, long is bad.

You might ask, claim 2 is short:

Claim 2. The computerized method of claim 1, further comprising recording at least one of the second reference level time interval hashes to a distributed ledger.

But claim 2 is a dependent claim. It descends from claim 1. It is all of claim 1 plus the step above. It is 10 steps that must be done to be an infringement. Claim 1 is called an independent claim. If you look at the other claims, 7 and 16 are independent claims and quite long.

Now look back at those news articles. The writers are business writers, so don’t begrudge them not doing this analysis. But the Overstock press release is puzzling. What’s wrong with linking to their new patent? Aren’t they proud of it? Don’t they want others to read it?

My credentials:  13 US patents (so far); 2 are blockchain patents – here and here. The patents are ranked 6th in the world, of the top 10 blockchain patent portfolios.