Friday puzzle: switches

Posted on June 12, 2009 by paulmcl

By the way, I moderate comments on the puzzles, just to stop people posting the answer in the comments and so making it hard to avoid seeing accidentally if you want to try the puzzle yourself. So if you comment and it seems to fail, don’t worry. It will appear eventually (but after a week if you include the answer).

Last week’s puzzle is traditionally known as the “impossible problem” although there are a number of variants with different limits and slightly different statements. The answer is that the sum is 17 and the product is 52 (the underlying numbers are 4 and 13).

Start by noting that if the product alone is not enough for P to know the answer, then the product cannot be a product of two primes (since then those would be the only two possible factors and so the sum would be easy) or less obviously the cube of a prime (since then the only factorization would be the prime and its square).

S cannot thus be the sum of two primes (which eliminates all even numbers by Goldbach’s conjecture, or for a problem this small we can just check) or the sum of a prime and its square. For otherwise S could not be certain that P couldn’t solve the problem the first time around. With a bit of thought it is clear that S has to be less than 55 too: look at the first prime greater than 50 and we can see 53*(S-53) would have a unique allowable factorization since we have to keep everything less than 100.

That leaves possible sums of 11, 17, 23, 27, 29, 35, 37, 41, 47, 51, 53.

Now consider all possible splittings of each of these sums into two numbers that sum correctly, and consider the products that result. Since P knows the answer on the third statement, that product must only occur in one of these splitting (otherwise it would still be ambiguous). So we can go through and take out all the products that occur in splittings for more than one of the possible sums.

When you do this, you’ll find that only 17 has a unique splitting left. But since in the fourth statement S knew the numbers then this must be the solution, S=17, the two numbers are 4,13 and P=52.

Today’s puzzle is pure logic, without any arithmetic. It’s another puzzle that seems impossible until you’ve started to think about it for some time. A warden meets with a group of prisoners. They are told they may plan a strategy together today but from then on they will be held in isolation, unable to communicate. One at a time they will be taken into the “switch room”, where there are two switches labeled A and B which are either up or down. The prisoner in the switch room must change the position of either switch A or B (not both, not neither). Then the next prisoner is taken into the switch room, and given the same choice. The order that the prisoners visit is picked by the warden, and prisoners may be taken to the switch room multiple times, even consecutively. The only guarantee is that, given long enough, every prisoner will visit the switch room time after time. At any time, any prisoner may state “all prisoners have visited the switch room.” If this is true, all the prisoners are released, otherwise they are imprisoned forever. The initial position of the switches is unknown. But I’ll give you a hint: first try and solve the problem assuming you know both switches are up to begin with.

Answer next week

Posted in puzzles | Comments Off

Chinese

Posted on June 12, 2009 by paulmcl

I’ve been learning Chinese for some time. Most people know three things about Chinese: it’s a tonal language, it’s hard to read all those pictogram thingies, and it must be really complicated.

Well, the first two of those things are true. But partially due to the writing system, which is not capable of small incremental changes, the language itself is simple in many ways. For example, verbs don’t decline like English (and all other European languages) and, in particular, there are therefore no irregular verbs; there are no tenses. There is no gender like in French or Spanish and therefore no need for adjectives to agree with nouns. Noun endings don’t change like in German, there’s not even an equivalent of adding an “s” to make plurals because there aren’t really any. For simple sentences the word order is the same as English: subject verb object.

There is a standard way of writing Chinese using our regular alphabet, called Pinyin. Created in the late 1950s, it became widespread by the 1970s (and finally became the official Romanization in Taiwan on January 1st this year). That is the reason that what we used to call Peking became Beijing. If you ever fly there, though, you’ll notice that the three-letter airport code is still PEK!

Not all the letters sound like they do in English. So even learning Pinyin is not completely straightforward. And you’ll need a native speaker to teach you the difference between some consonants. Say “jug” and “chug” to yourself in English and then imagine trying to explain the difference to a foreigner. Chinese has several consonants that sound almost like that: “zh”, “j”, “c”, “ch”, “sh”, “z”, “s”, “q” often sound much the same when you first hear them. The Chinese for “please come in” is “qǐng jìn” but when I first heard it on a CD I couldn’t understand how it was the same word twice like the Italian toast “chin-chin” (or strictly “cin-cin” since “ch” is hard in Italian).

In Chinese the tones distinguish meaning of words that are otherwise spoken identically (although with different characters). You probably have heard that the word for mother and horse are the same “ma” in Chinese but with different tones. Even the tones are not that hard once you realize that we use tones in English sometimes, although we use them to express mood or emotion. Think of how you would say, “It’s a full moon” as a statement versus a question. You vary the tone of “moon.” In fact one habit that it is hard to get out of is putting English emotional tones onto spoken Chinese and inadvertently changing the meaning, especially the last word of a sentence. In Pinyin, the four tones in Chinese are represented with accents with the tone following the direction of the accent (level, rising etc). So Beijing is really written Běijīng.

Once you know the Chinese word for something then you know all about how to use it. You probably know that the everyday Chinese greeting is “nǐ hǎo”. This actually means “you good”. So now you know the word for “you” and the word for “good”. Here’s another simple thing. If you put “ma” on the end of a statement it changes it into a yes/no question. So “nǐ hǎo ma?” means “you good?” or “how are you?” And yes, that “ma” is the same “ma” as means both mother and horse but with a different (in fact no) tone. You might think that the answer to a “yes/no” question would be either “yes” or “no” but Chinese doesn’t really have words for “yes” and “no”. It’s like that party game where you have to avoid saying “yes” or “no” (“your name is Paul?” “it is”). So the possible answers to “nǐ hǎo ma?” are “hǎo” (good) or “bù hǎo” (not good), although, as in English, “how are you?” wasn’t a question expecting a literal answer. And you’ve learned another word “bù” meaning not, which can be used to negate almost anything.

Here’s a verb. The Chinese for “want” is “yào”. And the great thing is that it never changes. It never declines. It never changes tense. If you’ve ever learned a European foreign language (or English as a foreign language) all the common verbs tend to be irregular, so almost every verb you actually use is irregular, making for a steep learning curve. There is none of that in Chinese. “Do you want tea?” is just “nǐ yào chá ma?” And “I don’t want tea” is “wǒ bù yào chá.”

It’s like Legos for a language. It’s a shame the pictures on the blocks are so complicated.

再见. Zài jiàn. Again meet. That is to say, goodbye.

Posted in culture | Comments Off

Channel choices

Posted on June 11, 2009 by paulmcl

Should a separate product be sold through a separate channel? If a new product is pretty much more of the same then the answer is obviously “no.” If the new product is disruptive, sold to a different customer base, or requires different knowledge to sell then the answer is less clear. There seem to be several main inputs into the decision. Cost, reach, conflict, transition and disruption.

First, cost. Each channel costs money. Obviously a separate direct sales force, as Cadence once had for the Alta Group (its system level tools), is expensive. Less obviously, even a distributor or reseller has cost too: upfront cost in training them and ongoing cost in supporting them and in the portion of each sale that they retain. At the very least the separate channel needs to be more productive than it would be to simply sell through the existing channel. By productive, I mean delivering more margin dollars. Sales might be higher with the separate channel, but sales costs might be even higher still making it unattractive. That is one reason that typically when an acquisition is made, the sales force from the acquired company is folded into the sales force for the acquiring company (usually with some of the lower performers being surplus to requirements) rather than being ramped up aggressively as a separate channel.

The second issue is reach. The existing sales force sells to certain customers, and in fact to certain groups within those customers. It will be hard for an existing sales force to sell a new product if it has different customers or even completely different groups within those customers. Their “rolodex” (or CRM system) isn’t any use. They are not already on the right aircraft, they are not already going to the right meetings. In this case, that militates for having a separate channel.

The third issue is conflict. So-called “channel conflict” occurs when a customer might be able to purchase the same product through more than one channel, specifically more than one type of channel, such as direct from the company or via some sort of reseller. This has impact on pricing in a way that might have downsides. For example, go up to Napa Valley and visit a winery. For sure, the winery will be very happy to sell you a few bottles of wine. Since they don’t have any middlemen and have a huge amount of inventory (they don’t just have the few bottles in the store, they have hundreds of barrels of the stuff in the back) then surely they will sell you the wine for less than anyone else. But, in general, they will sell you the wine for the highest price anywhere. If they sold it for less, they would make more money at the winery but they would risk having distributors and restaurants refuse to carry it. In EDA, if there is a product available through distribution and direct, then the direct channel cannot routinely undercut the distribution or the distributor will soon stop actively selling.

The fourth reason to have a separate channel is when the market demands, or the company decides, that it must transition its sales from one channel to another. Maybe they decide to move from direct sales to only doing telesales or only taking online orders. Or perhaps they decide that the day of a standalone product has gone, and they will only be able to sell integrated with a partner going forward. The channel must switch from however they sold before, to simply relying on the partner to sell their product and getting their share of those sales (and, presumably, enlarging their partners market in some way or else the partner wouldn’t be interested). I’ve talked before about how in EDA OEMs only work when the product is actually a component, since otherwise the customer will always want a direct relationship with the real seller. But if you do have a component, rather than a product, you must sell through an OEM type of license (as do companies like Verific or Concept Engineering).

Finally, disruption. If you have a product that is disruptive you have to have a separate channel. Disruptive, in the Innovator’s Dilemma sense, means (usually) that it is sold at a low price point to people who are not served by existing products, and where the low price point product is expected to improve fast and gradually swallow most of the market. Think early PCs versus minicomputers or teeth-whitening strips versus dentist’s providing whitening service. The existing channel is threatened by the disruptive technology, may not even be able to cover its channel cost (think of your dentist selling you teeth whitening strips you could just pick up in Longs) and will be unenthusiastic about selling it compared to more profitable lines. If you are going to be brave enough to try and kill your own baby, then you need separate organization for the baby and the killers. Sometimes, moreover, the disruption is the channel itself (Amazon and bn.com or, for a historical example, Sears starting to sell by catalog as well as department stores). This means a new channel by definition.

EDA has rarely had a separate sales force, since it is just too expensive. One that I mentioned above was Cadence’s Alta Group. For a period that had its own sales team and was successfully growing revenue. But it was expensive and Cadence decided to fold it back into the main sales force. Sales declined and Cadence ended up “selling” that part of the business to CoWare (which you can regard as one way of going back to a separate channel).

How to get a picture like the one above. Who needs a real band, rehearsals, or a recording studio? Thanks to the power of the internet, you can have your very first album the no fuss, no muss way!
1. Go to the Wikipedia random article page en.wikipedia.org/wiki/Special:Random. The title of the article is your band name.
2. Go to the very last quote on random quotations www.quotationspage.com/random.php3 . The last four words of the last quote are your album title.
3. Go to Flickr’s "interesting photos from the last seven days" page www.flickr.com/explore/interesting/7days / . The third picture on the page is your album cover.

Posted in eda industry, sales | Comments Off

The corporate CAD cycle

Posted on June 10, 2009 by paulmcl

Many things in business go in cycles. One in EDA is what I call the “corporate CAD cycle”. It goes like this. I’m sure a similar dynamic plays out in other industries too.

A large multidivisional semiconductor company has dozens of products in development. They have management who decide that the best way to hold groups responsible is to give them complete control of their destiny. Each group decides what its design methodology is going to be, purchases tools and is generally pretty successful at getting their products out. Life is good. Then one day someone in finance or corporate CAD notices two things: the amount of money that the company is spending on tools is very high, and secondly the various groups have made different tool/methodology decisions and so it is much more difficult than it should be to move around people (need re-training) and pieces of designs (wrong formats, wrong standards). The first phase of the cycle has ended.

A decree goes out. Corporate CAD will make corporate-wide volume tool purchase and decide what should be the standard flow. The standard flow will be mandated throughout the corporation, no exceptions. In a fairly short time the big problems are fixed: the tool budget is slashed now that experienced purchasing agents are negotiating very large deals with just a few EDA vendors; moving people and blocks around is simpler since everyone has the same base knowledge. Life is good. The second phase of the cycle has ended.

But management notices that many product groups are being a lot less successful at getting their products out than they used to be. This is a huge multi-million dollar problem. Management takes a look at the most important bet-the-company product. The product group says they are forced to use the wrong tools, that they are spending too much time getting blocks into corporate formats that they don’t use themselves, that they have too few licenses. But this chip is critical, so just this once, since the group is really expert, they are allowed to buy whatever tools they need to get the job done. Of course when corporate CAD said there were no exceptions to the standard flow, they weren’t that stupid. Of course there were exceptions for RF. And the advanced group over there that is doing a pilot project in 45nm obviously needs some tools that are different. Oh, and we just acquired a little fabless semiconductor company who uses something non-standard, we’d better let them at least get their chip out before we force them to use the standard flow, it’s hard enough for them switching to our process. Gradually the iron grip of corporate CAD relaxes. Corporate CAD makes big purchases, but a lot of those tools are sitting on shelves. Design groups are largely making their own decisions about tools. Chips are coming out successfully again. Life is good. Then someone in finance…

This cycle seems to re-play itself in many areas where two contradictory goals collide. In the CAD case above it is the need to have a standard flow, but also allow exceptions when the standard flow is not enough. Of course I exaggerated the story a bit, but the basic cycle between giving too much freedom and not enough seems to be real and I’ve seen a version of the cycle play out in many semiconductor companies.

One other area with this sort of cycle is whether to manage functionally or along product lines. Should a large semiconductor or EDA company (or anyone else for that matter) run engineering as an organization, and then have project managers for each product? Or should a product be responsible for all its own groups? Engineering, marketing, finance (probably not), sales (maybe). There are problems with both structures. If product groups are self-contained (even if they don’t have dedicated salesforces) then there is no control of corporate brand image, little standardization of development processes and so on. At once point when I was in Cadence years ago, we had 6 or 7 full-page ads in EE times from different groups, all product-oriented but with a different look and feel and brand image. However, if engineering, marketing and so on are all functionally managed then they can be very unresponsive to urgent needs in the product groups.

The way this seems to play out is to be functionally organized for a few years, then when that seems ossified, become structured into business units or product groups for a few years. One reason that I think that this can actually be effective, not just management churning, is that the relationships from the previous era endure, at least for a couple of years. If you were previously in marketing for a product group, and now you are in the company-wide marketing organization, you still know all the engineers that work on the products you care most about; after all, only a year ago you were all in the same organization. When eventually you get blown part back into product groups, you still know everyone in marketing, you know in your bones the corporate look-and-feel, branding and so on, and it takes a couple of years to gradually forget (as an organization, with people coming and going, more than individuals actually forgetting).

So there really is something to the old management canard that, when you don’t know what to do, centralize everything that is distributed and distribute everything that is centralized.

Posted in semiconductor | Comments Off

Term sheets

Posted on June 9, 2009 by paulmcl

What is a term sheet? If you raise money from a venture capitalist (or an experienced angel) then the most important conditions for the investment will be summarized in a term-sheet. It sounds like this should be a simple document of a single sheet of paper, but in fact these days it is dozens of pages of legalese that is a good way towards the final legal documents. In fact, it is so complex that typically the really really important stuff will indeed be summarized in a few bullet points ln a different piece of paper (or an email).

The most important bullet point, of course, is the pre-money valuation. This is the amount that the investor values your company before putting in the money. The post-money valuation consists of this amount plus the money that they invest.

One item to make sure you understand is the employee option pool. There will be a pool of options created for future employees. The normal way is to carve this out from the previous round before the new round money goes in. This means you pay for the option pool, not the new investors. At one level this sounds unfair, but in fact it just feeds into the valuation. The pre-money valuation is thus usually the valuation of the previous round’s stock, plus the value of option pool. So the previous round’s stock is not worth as much as it sounds. Of course this can be done other ways, but that will affect the valuation differently. Make sure you understand it since not all “$8 million” pre-money valuations are created equal.

The next most important item is probably whatever is agreed about liquidation preferences. The VC will have preferred stock, and preferred means that it gets a better deal than common stock in the event of an acquisition (or wind-up) of the company. Just how much better a deal is important. At the very least, usually a VC wants to get their money back before the common stock gets any. But it can be a multiple of their money they get back, or their money plus a dividend (interest). And then they get their share of the common, which there are a number of ways to do.

Another really important area, especially when there are multiple rounds, is who has to approve change-of-control decisions (essentially acquisitions). Can the VCs over-ride the common stock? Can series B veto series A? In one acquisition I was involved with, each round of preferred stock voting had to approve the acquisition, not just the preferred stockholders voting together as a group, which is not unusual. Sounds innocuous? But one round of investment was a corporate investor (a customer) who thus had veto rights over the acquisition (that they leveraged into a very good deal for themselves since they were also a customer of the eventual acquirer).

An excellent book on term sheets is Term Sheets and Valuations which goes through the most important items on the term sheet and gives you guidance for each item as to what is middle-of-the-road, what is aggressively in favor of the venture capitalist, and what is aggressively in favor of the company (or common stockholders). I’ve purchased this book several times since my copy always goes missing when someone forgets to give it back. If you are going to raise money and have not done it dozens of times before then you should read this book.

Another really good resource is that Wilson-Sonsini, the biggest name in legal for most silicon valley stuff, has put a term sheet generator online. You fill in various pages with the data, and it spits out the legal document. If you are really raising money, you should at minimum have your lawyer go over the document, it’s still just a document generator on the web. But if you are thinking about raising money, or want to know more about term-sheets, then you can play with the tool. Even going through the experience of filling out the forms will force you to understand a whole lot of issues that you might not have come across.

There’s also lots of stuff on various venture capitalist blogs. This will often give you an insight into why a VC might care about a particular issue, when it would be important, how big a concession it would be for them to give you more favorable terms and so on.

As always, remember that the VCs have usually done this lots of times before. You, not so much. Get good advice.

Posted in investment | Comments Off

CEO: a dangerous job

Posted on June 8, 2009 by paulmcl

Why do so few startup CEOs last the distance? The Bill Gates, Michale Dell and Scott McNealys who take their companies all the way from the early days as a tiny startup all the way up to enormous multi-division companies are very exceptional. I think that it is obvious that running a little engineering organization developing a technical product requires very different skills from running a large company. Engineering skills dominate in the first; people and strategic management skills dominate in the second. A CEO has to grow a lot along with the organization to be successful at each stage of the company’s growth.

What is less obvious is that the skills getting a company going are very different from running it once the engineering phase is drawing to a close, or in some case just getting started. I’ve read various statistics, but something like 75-80% of startup CEOs are replaced before their company gets acquired (or merges, or goes public etc).

Getting a company started, and raising the first money to fund it, requires a level of focus and obsession that is abnormal. The “doing whatever it takes” attitude is necessary in those very early days, but it tends to leave a trail of turds to be sorted out later. Further, some people like this have difficulty making the transition to being a team-player once the key hires have been made. Nothing will alienate high-performers more than trying micromanage them, or treating them without integrity, or generally not regarding them as close to equals. A startup is more like a jazz-band than a military organization. It is interesting that the highest performing small-scale parts of the military, Navy SEALs or the British SAS, abandon a lot of the military trappings (SAS officers famously are often called by their first names).

I’ve been in several startups where I’ve come in later, well after founding, and had to sort out problems that are left over from getting the company founded. Complete inequities in salary or, especially, stock seem to be the natural debris of getting people out of their current organization and into startups. But not getting them into the company is probably a worse problem.

There are no hard divisions between different stages in the life-cycle of a startup, but roughly speaking there are four. Getting the company founded along with the other initial founders; getting the engineering development solidly under way with a competent team; getting initial sales and starting to ramp up a channel; growth to a more mature organization with an industry standard breakdown of headcount. There is a fifth (and probably more) stages as it become more and more difficult to manage larger organizations. The largest organization I’ve run had about 600 people, and that is like sailing a supertanker. You think you spin the wheel but nothing happens.

At each of those four stages, the CEO may or may not make the transition. VCs are famously ruthless if they think that the CEO is not the best person to look after their investment. The old CEO, no matter how important he or she was in earlier days, is off to “spend some more time with their family” and a new person is at the helm overnight.

The most dangerous phase for many CEOs is the transition from engineering to starting to ship the product. Founding CEOs are often very technical, effectively the primary architect of the product. Like many engineers, they overestimate the importance of technology and they underestimate the importance of marketing and account management. By their nature as founders, they may be much better at driving over objections than at listening. So they fail at the business side since they are out of their natural comfort-zone and they compound the problem because they won’t listen to people who know what they are doing wrong. This happens so often that VCs see it coming from afar and don’t even wait to see if the CEO can handle it before hitting the eject button. They knew when they founded the company that they would change the CEO. Sometimes they even make it a condition of funding, to make the process less traumatic when it happens.

Posted in management | Comments Off

Friday puzzle: impossible

Posted on June 5, 2009 by paulmcl

Last week’s puzzle was to calculate the volume remaining after a hole 6” long is drilled through a sphere. When I was a math nerd in high-school I actually knew how to do the triple integral in spherical polar coordinates to solve this problem properly, and, in particular, verify that the answer does not depend on the radius of the sphere. No way I have a clue how to do that today. However, if you start from the assumption that I wouldn’t be idiotic enough to ask the question if the radius of the sphere was relevant, then the problem is easy to solve. The limiting case is a sphere of diameter 6” so that the hole will remove nothing. The answer must be the volume of a sphere of radius 3” (remember it’s 4/3 π r3) namely 36π cubic inches.

Today’s puzzle also seems not to have enough information. Two numbers not necessarily different are picked from integers greater than 1 that sum to less than 100. Person S gets told only the sum of the two numbers. Person P gets told only the product. The following conversation takes place. P says to S, “I don’t know the two numbers.” S replies, “I knew that you didn’t know.” P then says, “Now I know the two numbers.” S then says, “Now I know the two numbers.” What were they?

Answer next week

Posted in puzzles | Comments Off

Lightweight hiking

Posted on June 5, 2009 by paulmcl

A few years ago I walked the John Muir Trail in the highest parts of the Sierra Nevada mountains from Yosemite Valley to Whitney Portal with my then-girlfriend. Actually, the trail technically ends at the top of Mount Whitney but unless you own a helicopter it’s hard to arrange a pickup from there. It is a total of about 220 miles. We actually didn’t do it in one go since I didn’t have that much vacation. We did the northern half one summer and the southern half the following summer, going in and coming out at Bishop Pass.

Whenever I had been hiking before I had been out for a couple of days at most, so never really reached anywhere that was too isolated for the weekenders. This time it was different and we were in some of the most beautiful valleys you can imagine in complete isolation. A spiritual experience as well.

Even hiking in the wilderness does not allow an escape from the reach of modern technology. There has been a revolution in clothing over the last twenty years: fleece, GoreTex, lycra, Paramo. All light, quick-drying, warm and breathable. But for long-distance hiking the real high-tech revolution has been over weight. In the past, hikers would carry rucksacks weighing 50 pounds. Ray Jardine, an aerospace engineer turned world-class climber turned long-distance hiker changed everything. He developed a low weight philosophy over many long-distance hikes such as doing the Pacific Crest Trail from Mexico to Canada in only 5 months. And then he wrote Trail Life that is now the bible for Ray-Way hikers who carry as little as possible (Ray republishes the book every several years, but confusingly changes the title each time so the version I used as actually this out-of-print one). Instead of heavy frame rucksacks, they can be made from parachute nylon and weigh less than a pound. Use the sleeping pad as part of the rucksack structure. Don’t carry warm clothes and a warm sleeping bag since you can wear the clothes in the sleeping bag and so carry one that only weighs a pound or so. Don’t carry a tent with its heavy poles, carry a waterproofed parachute-nylon tarpaulin and tie it to trees. Carry a key-chain flashlight, the smallest Swiss army knife you can get, a single pan (titanium of course), no change of clothes except socks.

We didn’t go as far as some hikers and worry about cutting the labels out of our clothes, the handles off our toothbrushes or discarding the distant parts of our maps. Our packs probably weighed about 12 pounds plus food, water and fuel. But they were lighter than I have ever hiked with before. So instead of hiking 10 miles per days and feeling tired, we were hiking around 20 miles most days. On the longest day we went 23 miles, partially because the rain made stopping for a meal or a rest unpleasant so we just kept going. That day we discovered another Ray Jardine innovation since we had taken an umbrella with us. Unfortunately we only took one but it was clear that whoever had the umbrella was happy, without even having to put up their hood. Whoever didn’t have the umbrella was miserable, especially when the rain turned to hail.

One thing that is difficult to avoid carrying is a bear-barrel. Yosemite bears have got wise to the old way of fooling them which was to tie your rucksack on one end of a rope and hoist it into the trees. Eventually even bears worked out how to bite through the other end of the rope; or send up a cub to inch out along the bendy branch; or just break the branch with their weight. Yosemite’s policy of moving problem bears from near the touristy valley itself, out to isolated areas of the park has meant that the knowledge of how to defeat bear-bagging spread quickly. So now a bear barrel is mandatory. This is a cylindrical container that has a recessed lid that is held in place by screws. There are no edges for the bears to get a tooth or a claw into, so they cannot get the food out even though they know that it is in the barrel. The old ones were plastic and fairly heavy. High-tech to the rescue. Newer ones are built from aerospace composites and machined aluminum, but they still weigh over a pound, which is sacrilege and probably makes them the heaviest single thing we carried. And that is before filling them up with 10 pounds of food to last for about six days. The food is all dehydrated and the volume of it is just as much of a challenge as the weight since it all must fit into the bear barrels. In the end we never even saw a bear, which is unusual. When I have hiked nearer to Yosemite the campsite has always been visited in the night. Of course doing long distances each day leads to virtuous circle on the food front: less food is required and so the pack can be lighter still, so you can go still further per day.

Another guru of modern long-distance hiking is an Englishman who lives in Scotland called Chris Townsend. He is author of The Advanced Backpacker, another standard book on backpacking, somewhat less extreme than Ray Jardine (and clearly a lot less weird as a person). We met an Englishman on the trail, interested that we were the only other people that he had met on the trail hiking in Teva sandals. Not using hiking boots but sandals or running shoes is another way of hiking further in a day since most hiking boots are like adding weights to your feet to slow yourself down. When we found out he lived in Scotland we asked him if he knew Chris Townsend. Surprise, surprise, he was Chris Townsend.

We didn’t take the ultimate piece of high-tech gear with us, the GPS unit. That seemed both unnecessary since route finding is easy in summer (not so in Winter when the trail is hidden by snow) and back then they were both expensive and heavy.

I’m actually hiking right now, in Romania (although Transylvania sounds so much more, well, vampirish). I’ll let you know if we find Dracula (and, for once, I don’t mean the DRC).

Posted in culture | Comments Off

Patents, CDMA, trolls and standards

Posted on June 4, 2009 by paulmcl

CDMA is also another interesting oddity from a patent point of view. Most patents are tiny pieces of incremental innovation that form the many little pieces you need to build complex technological products. You can’t build a semiconductor without violating thousands if not millions of patents. For example, Motorola (Freescale now, I suppose) owned a patent on the idea of filtering photoresist which surprisingly passed the non-obvious test. This used to be a minor annoyance since the patents were owned by other semiconductor companies, and the problem could be resolved with a manageable payment or royalty and a cross-license. After all, you don’t need to be in the business for long before they can’t build anything without your patents. Now that a huge number of patents are owned by so-called patent trolls, people who have purchased patents for the explicit purpose of trying to generate disproportionate licensing revenue, the cross-licensing approach won’t always work and, as a result, the patent system is effectively broken for technologies like semiconductor (and EDA for that matter) that stand on the shoulders of those who went before in ways too numerous to even take time to examine.

Patents were a problem for GSM phone manufacturers since companies like Philips and Motorola managed to design their own patents into the standard. GSM had the concept of essential and non-essential patents. An essential patent was one that you couldn’t avoid: if you were compliant with GSM you were violating the patent, something that "shouldn’t happen." However, the essential patent owners preferred to keep their heads down for political reasons (don’t want those European governments telling us off in public) and keep quiet about what patents they owned until businesses were rich enough to be worth suing. For example, Philips owned the patent on the specific vocoder (voice encoder) used in GSM. Not the general idea of a vocoder, or that type of vocoder, just the specific parameters used in GSM, for which they would like about $1/phone. It was as if Ford owned the patent on the order of the pedals in a car. Not the idea of an accelerator, clutch and brake but the specific configuration of the clutch to the left of the brake to the left of the accelerator. And then got some car standardization authority to mandate that order for all vehicles. Come to think of it, that’s pretty much what GM did when they got the US government to mandate catalytic converters for all cars, which required all car manufacturers to license catalytic converter patents from a certain car manufacturer beginning with G. And there was some of this with Qualcomm too, since "everybody" knew that the main US carriers would choose GSM, the almost world-wide standard, until someone from the President’s office apparently told them at the last minute that it really ought to be a US standard.

People are smarter these days about making sure that patents don’t get designed  into standards. Look at the fuss over Rambus. However, it is still a grey area. After all, nobody knows what even their own company’s patent portfolio really covers. If you’ve read a patent, you know how hard it is to tell what it really says. You can only read the word “plurality” a limited number of times before your eyes glaze over. And at the company level, nobody knows the whole portfolio. If you are the representative from, say, Nokia on some standardization committee, then you can’t really guarantee that any particular standard doesn’t violate any Nokia patents, and you are certainly not going to sign up for guaranteeing never to sue, say, Samsung over a patent violation. Especially as you are not the corporate counsel, you are some middle level engineer assigned to a standardization committee that may or may not turn out to be strategically important.

But CDMA was a complete patent-protected technology more like a blockbuster drug formula. You couldn’t do anything in CDMA without licensing a portfolio of patents from Qualcomm on whatever terms they felt like giving you. They invented the entire technology and patented it before anyone else really knew it was feasible. They sued Broadcom, they sued Ericsson, they sued everyone and pretty much established that there was no way around this no matter what. In 2G this wasn’t a big issue since GSM doesn’t depend in any way on CDMA. But W-CDMA and all the later technologies use various aspects of CDMA and so Qualcomm is in the happy position of having a tax on every cell phone.

Posted in investment | Comments Off

Guest blog: Lynn Conway

Posted on June 3, 2009 by paulmcl

Today’s guest blog is by Lynn Conway, who is the Conway of "Mead and Conway," the book on VLSI Design that first moved VLSI Design from its priestly guilds within a few semiconductor companies and into first the universities and research institutes and then the mainstream. Lynn Conway was at Xerox PARC and Carver Mead at Caltech. Depending on your age, you either lived through this VLSI revolution or else this is the recent history of how the current approaches to design you use today first came into being. She has recently collected together a lot of the documents from that period into the…

The VLSI Archive

When reflecting back in time with friends and family, we often use albums of snapshots to trigger shared memories—memories that bind us together and reveal how we got to where we are.

But what about the career-adventures of folks in Silicon Valley and beyond? More often than not we see final products of their work, but mementos of what happened along the way are often lost. Only too late do folks realize what “should have been saved.”

Somehow it was different for the VLSI revolution of the late 70’s and early 80’s. Perhaps it was the exciting visual nature of the artifacts. Whatever the reason, many original treasures from that era have been saved—stored away in people’s attics and basements for decades now—from old notes, to chip photos, to huge color check-plots.

During the past two years, members of the VLSI research team along with colleagues in academia and industry have been gathering up many original artifacts—archiving them at the Computer History Museum and scanning and posting many online. It’s a work in progress, and whatever we can find we put in there.

The resulting VLSI Archive is bringing the old adventures back to life and now includes things such as:

  • The 1976 letter from Ivan Sutherland to Bert Sutherland that framed the research opportunity then at hand and proposed the Xerox PARC/Caltech collaboration to develop radically-improved VLSI design methods and EDA tools.
  • Prepublication versions of the Mead-Conway text used to document, teach and evolve the new design methods and tools. Written on Alto personal computers at PARC and demand-printed on the newly-invented laser-printers there, they were used in trial-runs at several universities prior to publication of the text.
  • The handwritten course notes for the first modern VLSI design course, taught at MIT in 1978. These became the Instructor’s Guide and along with the Guide to LSI Implementation and the Mead-Conway text were used to spread the course to over 100 universities around the world.
  • Many documents and notes from MCP79 the internet-based, rapid-chip-prototyping effort that implemented design projects at a dozen universities in 1979, and proved out the Mead-Conway methods. Included are documentation and chip photos of the 124 design projects – projects that led to many innovations in chip architectures, some (for example Jim Clark’s Geometry Engine) leading to major startups such as SGI, Sun Microsystems and MIPS.

As a side note: demand for QTA chip prototyping later led to institutionalization of the MPC79 service as the MOSIS system—used for decades now to support chip prototyping by students and researchers in U.S. universities and industry—and MOSIS-like services later sprang up in many other countries too.

The VLSI movement also started it’s very own VLSI Conferences, which helped build a cohesive, forward-looking research community, and short-intensive courses were offered (such as the one by VLSI Technology) to faculty preparing to teach VLSI design and to early adopters in companies eager to exploit the new methods.

EDA was a vital dimension of the VLSI revolution, and the archive includes key documents from the development of modern design tools, from the early beginnings in Doug Fairbairn’s and Jim Rowson’s work on ICARUS and Lynn’s invention of scalable design rules, to the specification of CIF 2.0 by Bob Sproull and Dick Lyon, to Randy Bryant’s work on MOSSIM, to John Ousterhout’s work on CAESAR and MAGIC, on to the startups of VLSI Technology, Cadence, Valid Logic, Daisy, Mentor Graphics, Viewlogic and more.

The revolution also had its own magazine: Created by Doug Fairbairn and Jim Rowson, LAMBDA (later VLSI Design) showcased important new advances and played a key role in the propagation of the new design methods and design tools.

As a result of all these wild happenings, the VLSI revolution really took off. Electronics Magazine spread the alert in a 1981 feature article, and as they say “the rest is history.”

You’ll sense the excitement of the times by browsing among the mementos, using the archive sprea
dsheet to access individual items. For further reading see my report The MPC Adventures and the National Research Council book Funding a Revolution in which much of Chapter 4 is devoted to the VLSI work

On reflection, it’s clear there are many other adventures out there—stories as yet untold—carried only in the shared memories of the folks who’ve built our industry. It would be great if their treasured mementos could be gathered up and shared online too!

Posted in guest blog | Comments Off