Chinese government reveals Microsoft’s secret list of Android-killer patents

http://arstechnica.com/tech-policy/2014/06/chinese-govt-reveals-microsofts-secret-list-of-android-killer-patents

by conductor • 12 years ago

168 84 12 years ago

whoopdedo• 12 years ago

Should patents change to a use-it-or-lose-it system much like trademarks? If you're aware of someone using one of your patents but don't do anything to control it, you're implicitly ceding future rights to that invention.

fokov• 12 years ago

If you did it like that, then it would be trivial to take away all Patents. Just create a shell company that infringes, sell the product to another shell company, and then the patent would be gone since the holder company didn't take action. It is almost impossible for a patent holder to know the implementation details of every other company in order to detect infringement. Typically it is patent is granted, someone comes along and infringes, the holder gets notified somehow, and then legal action is taken. There is always going to be a period of time where the holder doesn't know their patent is being infringed. Also, the holder does have the right to take no legal action against an infringement.

chaostheory• 12 years ago

That's fine. It still makes it more expensive for trolls to do business.

MichaelGG• 12 years ago

OK so entity makes a super expensive "thing" containing the patent. Now what? They have a valid product, sold 1 or 2 copies.

You'd require a judge to start deciding if a company is _really_ using the patent. Which seems hard to tell. Especially if the patent is for something requiring a ton of work/capital. Like a patent for a new spaceship or rocket or jet engine.

NPEs would quickly find a way to stay NPEs but get passed the legal definition of "non practising".

wfjackson3• 12 years ago

There's already a doctrine of laches that limits past damages.

From http://www.wilmerhale.com/pages/publicationsandnewsdetail.as...

Laches defenses have been successful in cases in which a patent owner knows about an infringement and then delays many years before bringing suit. Laches typically bars the recovery of any past damages but allows for the recovery of damages arising after the filing of the lawsuit

danudey• 12 years ago

Make patents non-transferrable.

moron4hire3• 12 years ago

That would harm small inventors more than large corporations. A small inventor working out of her garage would be nearly incapable of capitalizing fully on a patent any sort of for a big-market consumer product. Being able to transfer patents allows the inventor to sell their rights and obtain some degree of compensation despite lacking production capacity.

The idea is just the idea. It's execution that matters.

cwyers• 12 years ago

Microsoft is shipping products (Windows Phone, most notably) that use most if not all of the patents listed, and they're currently raking in money from multiple Android OEMs (Samsung is confirmed to pay Microsoft for patents related to Android). That doesn't really apply here, they've been very active in both using and seeking royalties for these patents.

rayiner• 12 years ago

It's interesting the Rockstar thing doesn't get more coverage on HN. My old company had Nortel as a customer. It was a tremendously inventive, if management-challenged, organization. I remember thinking when it was liquidated what was going to happen to what I was sure was a very deep patent portfolio.

tptacek• 12 years ago

The offices at Nortel (and Qualcomm, too) are lined with patent plaques; they covered every wall I saw. It was definitely a patent culture.

kitsune_• 12 years ago

Seriously, does any programmer have the time to decipher byzantine shit like this before writing a new feature? "Custom local search", US 8255379 B2

CLAIMS(20) A method comprising: receiving a query that is configured by an issuer of the query to perform a local search, wherein the local search is performed over a first dataset and a second data set, wherein the first dataset comprises a first entity and the second dataset comprises a second entity, and wherein the query comprises a first token and a second token; parsing the query such that the first token of the query is mapped to the first entity in the first dataset and the second token of the query is mapped to the second entity in the second dataset; and returning search results based at least in part upon the parsing of the query, wherein the method is executed by a processor of a computing device. 2. The method of claim 1, wherein the first dataset is a custom dataset, wherein the custom dataset comprises a plurality of entities generated by an end user, and wherein the plurality of entities comprise attributes that describe the plurality of entities and shapes pertaining to the plurality of entities. 3. The method of claim 2, wherein the second dataset comprises street-level vector data. 4. The method of claim 3, wherein parsing the query comprises determining an approximate geographic region pertaining to the query. 5. The method of claim 4, further comprising determining the approximate geographic region pertaining to the query based at least in part upon mapping the second token of the query to the second entity in the dataset. 6. The method of claim 1, wherein parsing the query comprises identifying which words and phrases in the query correspond to attributes in the first dataset and the second dataset. 7. The method of claim 1, wherein parsing the query comprises generating a plurality of different partial interpretations of the query. 8. The method of claim 7, wherein each partial interpretation in the plurality of partial interpretations comprises a set of mappings between tokens in the query to attributes in the first dataset and the second dataset and an approximate region of interest pertaining to the query. 9. The method of claim 8, further comprising refining at least one partial interpretation, wherein the at least one partial interpretation is refined by way of spatial processing. 10. The method of claim 9, wherein the at least one partial interpretation is refined by determining that an entity in the first dataset spatially corresponds with a region of interest pertaining to the query. 11. The method of claim 10, wherein the region of interest is altered based at least in part upon a geographic overlap of a previous region of interest and an entity. 12. The method of claim 1, wherein the query is received in a first language, and further comprising translating at least a portion of the query to a second language. 13. The method of claim 1, wherein the query is a free-form text query that comprises misspelled terms, erroneous terms, out of order terms, or valid but wrong terms. 14. A system that facilitates custom local search, the system comprising: a processor; and a memory that comprises a plurality of components that are executed by the processor, the plurality of components comprising: a receiver component that receives a query from a user, wherein the query comprises a plurality of tokens, wherein the query is configured to retrieve custom content with respect to a particular geographic region, and wherein the query comprises a first token and a second token; and an interpreter component that parses the query such that the first token is mapped to a first entity in a first dataset and the second token is mapped to a second entity in a second dataset, wherein the first dataset is a custom dataset that comprises custom data provided by an end user and the second dataset is a contextual dataset that comprises street-level geographic data. 15. The system of claim 14, wherein the first entity comprises a shape that describes a geographic region pertaining to the first entity, an attribute that describes the first entity, and a unique identifier that identifies the entity. 16. The system of claim 14, wherein the query comprises at least one token that is indicative of a custom dataset amongst a plurality of custom datasets to which the query pertains, free-form text, and at least one token that is indicative of geographic scope pertaining to the query. 17. The system of claim 14, wherein the interpreter component comprises a query interpreter component that outputs multiple partial interpretations of the query, wherein each partial interpretation comprises a list of mappings between subsequences of tokens of the query and sets of entities in the custom dataset and the contextual dataset and an approximate region of interest pertaining to the partial query interpretation. 18. The system of claim 17, wherein the interpreter component further comprises a refinement component that refines at least one of the partial interpretations, wherein the refinement component modifies the approximate region of interest pertaining to the partial query interpretation when refining the at least one of the partial interpretations. 19. The system of claim 18, wherein the interpreter component further comprises a ranker component that ranks refined partial interpretations. 20. A computer-readable data storage device comprising instructions that, when executed by a processor, cause the processor to perform acts comprising: receiving a query, wherein the query comprises a plurality of tokens, wherein the query is configured to execute over one or more custom datasets and one or more contextual datasets; generating a partial interpretation of the query, wherein generating the partial interpretation of the query comprises mapping a first subsequence of the tokens to a first attribute in the custom dataset and mapping a second subsequence of the tokens to a second attribute in the contextual dataset; refining the partial interpretation of the query to generate a refined interpretation, wherein refining the partial interpretation comprises mapping a first entity in the custom dataset that has the first attribute to the first subsequence of the tokens and mapping a second entity in the contextual dataset that has the second attribute to the second subsequence of tokens; and causing graphical items representative of the first entity and the second entity to be displayed on a display screen of a computing device.

There should be a non-profit organization that files arbitrary new patents preemptively on behalf of the public and then release them into the public domain.

anon4• 12 years ago

A system that takes a free-form text query for places and tries all possible interpretations to find a match based (somehow) on a history of places for that user (history in the general term, the way I read it, it can be places the user has been, places the user has searched, places whose shape look like the user's face, etc) and a global geospatial database.

So, a fancy parser, a per-user list of places of potential interest and a Solr collection with geospatial data.

P.S. the irony is that at no point in that overly verbose spec, is there anything that in any way tells you how to build that system. It's like patenting a device for enabling one to force a resistance against achieving minimum possible potential energy in a gravity field upon a mass of fluid, i.e. the concept of a bucket (or a glass, or the act of cupping your palms). And while that is surely only the preface of the patent, I seriously doubt the body is any more illuminating. Maybe while completely abolishing software patents remains impossible, we should try to enact reform that forces them to be specific enough that just by reading the patent one would be able to write a program that performs the actions described therein. Even if it's just "take a solr server, configure it like this, take this particular natural language parser, generate the possible forms of the query like so...". At least then they would be useful in a decade or so.

euank2• 12 years ago

This is one of the reasons software patents are dumb imo; they don't actually share sufficient details to implement the idea any more easily.

In other fields, such as mechanical engineering, patents are required to describe an invention in enough detail that a person "skilled in the art" could reproduce it.

In the field of software, the patent merely has to describe the invention (software) well enough that it can be identified, not written. To match other fields, where it's common to require essentially full bluprints, it would make sense to require working source code (if only a reference implementation) for any software patent.

As is, the patenter "gifts" the public with the knowledge of an idea which is rarely in itself useful; the source code could at least be potentially useful.

NAFV_P• 12 years ago

I gooooogled it [0], the patent was filed on 10/11/2009 (I'm British, so read that as dd/mm/yyyy). Some of the other patents listed go back a few years [1].

[0] http://www.google.com/patents/US8255379 [1] https://www.google.com/patents/US5889522

briantakita• 12 years ago

Seriously, does any programmer have the time to decipher byzantine shit like this before writing a new feature?

No. That is the game that is imposed by entrenched players, the wealthy, & lawyers to protect their interests (and maintain their dominance). They found a way to "hack" the system to their benefit over creators.

There should be a non-profit organization that files arbitrary new patents preemptively on behalf of the public and then release them into the public domain.

You need to have a working prototype to file the patent. This non-profit would need the inventors to cooperate.

[edit] Actually you don't need to have a working prototype. Now my question is isn't that the role that open source and free software / hardware fills?

talmand• 12 years ago

Unless something has changed recently, my understanding is that you do not need to provide a working prototype under US patent laws. You just need an explanation on paper of some kind that would allow a third party to be able to build whatever is being patented.

Of course, someone has to look at the explanation to determine whether the prototype could be built and be functional to have the patent granted. But who has time to deal with all that noise when you just grant the patent to let someone else deal with it?

akozak• 12 years ago

Re: your idea at the end: You might find the first section of Charles Stross' Accelerando a fun read.

tankenmate• 12 years ago

The first claim just sounds like a SQL SELECT with a JOIN; wouldn't this be easy enough to get overturned?

moron4hire3• 12 years ago

It would only invalidate that claim, but barring finding other prior art for the other claims, the rest of the patent would stand.

chris_va• 12 years ago

Occasionally I feel like there should be a wall of shame for the patent attorney that wrote the claims (and possibly the named Inventors by extension).

gbog• 12 years ago

The people writing and reading these papers for a living must live in a bubble and may not even realize how far and weird it is.

Could it be possible for a patent crash to happen? Maybe the way to get rid of this absurdity is to push it to the extreme, produce industrial quantities of silly patents and force the whole system to a stop, then a single match would easily burn all of it.

axman64• 12 years ago

The language is precise and defines the legal scope of the invention. The claims are not supposed to tell you how to implement the invention but to definitively define what is and is not within its scope. All areas of technical expertise have their own language to convey precise ideas, and the law is no different. Just because you cant read it doesn't mean it's not understandable.

nostromo• 12 years ago

Regarding your last comment...

I've wondered before if a group of hackers could DDOS the patent system.

It could be a modern form of civil disobedience.

poulson• 12 years ago

A more effective approach might be to maintain a peer-reviewed ranking of various large tech companies based upon their patent history and to semi-regularly publicly base your employment decisions upon said list.

deciplex• 12 years ago

I've wondered before if a group of hackers could DDOS the patent system.

There is such a group and they're already doing it. One of the players is the focus of this article, in fact.

carrotleads• 12 years ago

When I see such legalese style writing, I search for the first fullstop.

here it comes after 716 characters 0r 130 words. After this they seem to be lot more frequent.

Looks like the idea was to put people to sleep after 5 seconds.

axman64• 12 years ago

Claims are supposed to be written in the form of one sentense. Claims are not instructions on how to construct the invention, they are a legal definition of the scope of things that would infringe the invention. The language used is supposed to be precise, not generally easy to read prose. Reading a patent properly takes some training and experience, just like reading any other technical literature.

moron4hire3• 12 years ago

The truth is, neither do patent examiners. I've heard they only get 10 hours to review a patent application, at which point they have to approve it if they can't find any published [1] prior art.

[1] And that doesn't mean, "on some long-forgotten website in the dank corners of the internet". That means academic papers or other patents.

kourt• 12 years ago

There should be a non-profit organization that files arbitrary new patents preemptively on behalf of the public and then release them into the public domain.

Good idea. Let's call it "Bell Labs".

x1798DE2• 12 years ago

A note of clarification, but there's actually no reason to patent something if you're going to release it into the public domain. Patents are an incentive for disclosure, so if you write a paper on the subject and never patent it, no one else can patent it later.

ww5203• 12 years ago

The patent language is not too bad. Defeating a patent is not too hard. There are specific terms and patterns to the language, just like a programming language. It's often quite logical. It's just like reading another programming language.

A claim is just a list of "AND" conditions, i.e. all the conditions must be met for the claim to be asserted. To defeat a patent claim, it's a matter of adding enough variance in your code to defeat ONE of the conditions.

I will demonstrate the decrypting of claim 1. The other claims are just dependent claims of 1 to make the whole thing looks meatier. If claim 1 is defeated, all other claims are irrelevant.

Claim 1 basically saying a user issues a query with two parameters on two tables to do search in a program in the local machine, with the following conditions:

1a. the query runs against two data sets, Table1 and Table2.

1b. Table1 has one type of data, e.g. Food. Table2 has another type of data, e.g. Drink.

1c. the query has two parameters: param1, param2. E.g. FoodEaten and DrinkConsumed

1d. parse the query to use Param1 against Table1, and to use Param2 against Table2. E.g. select * from Food where food_type=:FoodEaten union select * from Drink where drink_type=:DrinkConsumed

1e. the query result should based on the parsing of the query and the mapping of parameters to the tables. The result should relate to the parameters in the query.

1f. the whole procedure is run by a process in a computer.

See, it's not too bad once you break it down. The claim conditions are kind of BS individually. The basic idea of the patenting process is to string together a number of day-to-day common items and make the whole thing as patentable.

To defeat claim 1, just make one of the conditions false. The easiest is to attack 1c, which uses the phrase "comprises a first token and a second token," which means the query contains exactly two tokens as parameters. Just add a dummy parameter to all 2-parameter query to fail that condition.

Attacking 1a with more than 2 tables query also works.

Also breaking a 2-table query into two separate queries works, too.

Since claim 1 is gone, all other dependent claims are irrelevant.

Disclaimer: the above are purely for educational purpose only and are not considered as legal advice. Consult a lawyer for real legal advice.

igravious2• 12 years ago

Just add a dummy parameter to all 2-parameter query to fail that condition.

This (along all your other circumventions) seems to me to only side-step the "spirit" of the invention -- to my mind any implementation using the techniques you suggest works more or less precisely as the patent describes.

If this is defeating I want no part of this charade.

quickaccount4• 12 years ago

"Comprises" is a term that has a specific usage in patents. Comprises means that at least all of the listed elements must exist but other elements that are not mentioned in the claim may also be present.

The alternative phrase is "consists of" which means that only the listed elements may be present. I think it's mostly used when claiming chemical or molecular formulas.

CatMtKing• 12 years ago

You seem to assume that a programmer should know about the patent and all other similar patents/patent conditions before he/she implements. I don't think that's so easy to do.

runn1ng• 12 years ago

I am sorry, but it's not "like a programming language". It's more like something that falls out from a compiler.

makomk• 12 years ago

A claim is just a list of "AND" conditions, i.e. all the conditions must be met for the claim to be asserted. To defeat a patent claim, it's a matter of adding enough variance in your code to defeat ONE of the conditions.

That's not true. Thanks to the doctrine of equivalents you can still infringe on a patent even if your invention differs from what the claim describes. (Now, if you want your invention to count as prior art, it does have to exactly match the claim - whether two inventions are considered the same for prior art purposes is different from whether they're considered the same for infringement purposes.)

chillingeffect• 12 years ago

Serious question here, but what is meant by "secret"? The patent database is public and after clicking through three listed patents, I see it quite clearly indicated that the Original Assignee is Microsoft.

Also, in the linked article by Brad Smith, uSoft wrote: We pledge that by April 1 of this year, we will publish on the web information that enables anyone to determine which patents we own.

How was it impossible to determine prior to April 1st? Did they update the patent database or something?

Thank you.

nitrogen4• 12 years ago

The secret was which ~300 of Microsoft and Rockstar's thousands of patents were part of the Android licensing scheme, not that they were actually owned by Microsoft.

kevingadd• 12 years ago

I don't really understand why Microsoft is expected or obligated to produce that list for the public. Calling it 'secret' seems a bit ridiculous in this scenario. The patents aren't 'secret'; the information was already out in the open. The 'secret' is a 'trade secret', that is, which patents they personally believe to be most important. It's quite possible that they have other patents in their arsenal that ALSO apply to Android, and any vendor that intentionally works around these ~300 patents will still get a patent suit.

If you really wanted to be safe from Microsoft and not pay them licensing fees, you could have paid some professionals to review their patent holdings for you, and identify any patents that could potentially be asserted against your smartphones. This list is essentially the result of Microsoft doing that.

patrickaljord• 12 years ago

Microsoft has thousands of patents, only a few hundreds of them are used to extort Android phone manufacturers. If the list of these hundreds of patents was made public, people could start working around them.

IBM• 12 years ago

The companies that license Microsoft's patents know what patents they're licensing.

rmrfrmrf• 12 years ago

There has to be some kind of middle ground here. On one hand, I'm very surprised that Google doesn't pay a license fee to Microsoft for Exchange ActiveSync. I assumed all phone makers with Exchange support had licensing deals with Microsoft. On the other hand, patents for things like a loading bar in the corner of a web browser are absolutely ridiculous.

natch• 12 years ago

It's gone. HN needs a caching system.

throwawaykf053• 12 years ago

Patents today are written in legalese that can only be interpreted by a select tribe of professionals, and Microsoft benefits by keeping lists like this secret.

This is like a blub programmer complaining about lisp being unreadable. I am no attorney or patent agent, let alone part of a "select tribe of professionals", just a typical software geek. Yet I can read patents pretty easily. Took me some practice to get the hang of it, but I needed absolutely no training for it, and neither do you!

Of course, if it's a legal matter (you're being sued), you need a lawyer. But if you just want to evaluate the technical merits of a patent, you can interpret the claims yourselves. All you have to realize is that patentese is just a particular form of English -- English that has to be overly specific (e.g. avoid pronouns) while at the same time trying to be as broad as possible, but English nonetheless. And you usually only need to read the independent claims (those that don't reference other claims) get a sense of the scope, as those are the broadest ones.

For instance, look at the claims conveniently posted at the top of the thread: https://news.ycombinator.com/item?id=7901544

I really cannot see what's so complicated about those. They seem a bit vague (e.g. what's an "entity"?) but applying the common meaning of the terms involved (e.g. maybe "entity" as in "person, place or thing") reveals a pretty straightforward description. It's nowhere near groundbreaking, but popular narrative of lightbulbs and flying machines notwithstanding, patents rarely are.

yuhong• 12 years ago

I have these two links, the second link has my wishlist for Satya and Android patents is on it:

https://news.ycombinator.com/item?id=7281319

http://hal2020.com/2014/03/03/satya-shuffles-his-leadership/...

krfsm• 12 years ago

Given that it's been a mystery which patents Android might be infringing, what is the likelihood that those patents fill their original purpose of promoting progress? (Beyond the effects obtained by simply implementing them and selling the product.)

Chinese government reveals Microsoft’s secret list of Android-killer patents

Related Stories