My answer to that last question was that it depends on which part of the "BFC" philosophy you're talking about. In the field of unmanned space exploration, chopping missions into smaller, less ambitious individual pieces is usually better -- however other kinds of cost-cutting measures may not be.

But now to the real meat: what changes, specifically, should we make in the Mars Surveyor program itself in response to this year's double failure? Let me give my opinion of what changes should be made in the program's individual missions.

First, it should be pointed out that the losses of the two 1998 missions don't in themselves force any delay in the sequence of later Mars Surveyor missions. The program was deliberately designed from the start with a large degree of "parallelism"; not many of its missions were supposed to produce data directly necessary to planning and flying later ones.

The Mars Climate Orbiter -- like the Mars Global Surveyor and the 2001 Orbiter -- was intended to refly the experiments lost on the 1993 Mars Observer. Its main experiment -- the Pressure Modulator Infrared Radiometer (PMIRR) -- was intended to study Mars' weather, and so wouldn't have provided data on landing sites for later missions anyway.

As for the Mars Polar Lander, it was selected in 1995, before the 1996 revelation of possible fossil life in Mars meteorite ALH84001 -- which led NASA to refocus the Mars Surveyor program to emphasize a search for fossil evidence of Martian life.

Before that, the program emphasized the search for life as essentially equal with two other major goals: studying the long-term climatic history of Mars, and the search for Martian resources that might be useful for manned expeditions (mainly, water).

The Polar Lander was designed specifically to study the current weather patterns of Mars, its regular cycles of climate change in the geologically recent past (the last few hundred thousand years), and to try to gauge how much greater were the amounts of air and water Mars had during its ancient days.

All of these goals could be best met by a landing in the polar regions. But the upcoming Mars landers are designed to focus on trying to locate evidence of Martian fossils, which can best be done in other areas. It will be a while before any attempt is made to return to the polar regions.

But what effect will the failures have on those specific later missions, as they are currently planned? Let me deal with the future program one piece at a time.

First, the 2001 Mars orbiter - which is likely to end up being called the "Mars Geochemical Mapper". This spacecraft is almost identical to the Mars Climate Orbiter, but with different experiments. It seems best to try to launch it on time.

It turned into something of a "fifth wheel" on the Mars exploration program, and serious consideration was given to cancelling it and proceeding directly to the sample return landers planned for 2003 and 2005. Ironically, however, the failure of MPL may have sharply increased the importance of the 2001 lander -- not so much for science, but as a testbed for new technologies that are needed to increase the chances for success of the expensive sample return landers.

There has been much talk that future Mars landers should return to the relatively simple and cheap bouncing hard-landing technique successfully tested by Pathfinder, which is also much more tolerant of rugged terrain of the type that may have wrecked the Polar Lander with its low-speed soft-landing technique and landing legs. The hard-landing technique does have a definite and major future for these reasons -- the British Beagle 2 Mars lander will use it in 2003, and it will be used many times all over the Solar System.

But the Mars Pathfinder program -- one of whose major goals was to test this technique -- uncovered an unexpected problem: the system was much heavier than expected. The craft's crash airbags ripped much more easily than expected during ground tests on cold, jagged rocks of the type found on Mars -- and to correct this, the bags finally had to be four layers thick.

They were originally supposed to weigh only 15 kg; they ended up weighing fully 85 kg -- one-quarter of the lander's total weight!

It's for this reason that NASA decided to return to the soft-landing technique on its following Mars landers -- it wants to cram as many experiments as possible onto its landers, while keeping them lightweight enough that they can be launched on small, inexpensive boosters.

In the case of MPL and the 2001 Lander, NASA decided that while a complex soft-landing system using throttleable rocket engines controlled by a multi-beam radar system was more expensive than the Pathfinder system, it would allow the craft to be launched on a smaller "Medlite" Delta and would thus be an actual net money-saver.

It can still switch its next Mars landing back to the more durable hard-landing system, if it's willing to swallow the added cost of a full-size Delta launcher like Pathfinder's. But the next spacecraft -- the sample-return landers -- are already much larger (they're scheduled to be launched on Delta 3s), and turning them into hard landers would require a sharp increase in the size and cost of their launchers.

It's possible, of course, that a more durable airbag material will soon be discovered that will allow the bags' weight to be sharply cut -- but a sample return lander is so large and complex that in any case it's hard to make it shock-resistant enough to endure a high-speed, bouncing and tumbling landing of the Pathfinder sort.

We'll have to retain the full-fledged soft-landing system for the sample-return landers -- but these missions are expensive enough that we badly need to increase that system's reliability.

And that's where the 2001 Lander comes in: it can provide a flight test of three new systems needed for that goal. It was already scheduled to test one important new technology: a new onboard navigation system designed to improve the targeting accuracy of Mars landers so that they can land -- with 99% certainty -- within only 10 km of their aiming point, rather than 35 km (as the Vikings and Pathfinder did).

MPL apparently was successfully targeted to within 10 km of its landing point using present techniques, but NASA wants that done on a regular basis for two reasons. First, it will greatly increase the number of possible landing sites, by allowing the landers to touch down in small smooth areas in the midst of dangerously rugged terrain as the 2001 Lander will do if it does land in its currently planned primary target area, the rugged highlands south of the Isidis Plain.

Second, since the sampling rovers on future missions will have a range of only a few kilometers, we need to land as close as possible to genuinely good sampling outcrops located from orbit.

The landing system for the Mars 2001 Lander will take in final trajectory data from Earth only a few hours before entry and there is a good chance JPL will land the craft much closer to its target than 10 km.

But two other technology additions to this mission suggest themselves in the wake of the MPL failure. First, there's a serious chance that MPL failed simply because it landed on an overly rugged patch of terrain. As far back as the Seventies, NASA had worked out a system whereby a soft lander could analyze photos from a descent camera on it, chop each picture up into 16 regions, identify the smoothest-looking one and steer the craft toward it.

It would have done so by identifying the region in which there was the least variation in the apparent brightness of the pixels -- and thus the smallest number of shadows.Since then, of course, computer technology has improved enormously -- and since MPL, unlike the Vikings and Pathfinder, was equipped with a descent camera for scientific purposes, it's surprising JPL didn't try to incorporate this ability into its landing navigation system.

The 2003 Japanese Selene Moon lander will use such descent photo analysis to steer itself toward a smooth landing area -- and the Deep Space 4 comet lander, before its cancellation this year, would have used a more sophisticated system in which a scanning laser altimeter would have built up 3-D maps of the landing are, and the lander would then have steered itself toward the best area which was both smooth and at a suitable slope.

Indeed, NASA was already hoping to incorporate a much more sophisticated descent-image navigation system on either the 2003 or the 2005 sample return lander -- in which the craft's computer would actually have compared photos taken during the parachute descent with prestored maps, and then, during its rocket descent after cutting itself loose from the chute, would have used them to try to steer itself to a landing within only 200 meters of its target point.

Such a system -- or at least the obstacle-avoidance part -- could be tested first on the less expensive 2001 Lander before being used on its more expensive successors, to see if it reacted properly to the lighting conditions and surface contrast of the actual Martian surface.

Finally, there's the matter of the lack of communications with MPL during descent -- which will very likely keep us from ever knowing with certainty just what happened to it.Much has been made of the fact that Pathfinder had "communications" with Earth during its descent -- but in fact this was an extremely rudimentary system which was not originally planned, and was added to the mission only as an auxiliary experiment requiring no additional equipment on the craft.

It only modulated a simple tracking carrier signal from the craft when it opened its chute, and again after the first landing shock. Even that crude data could have provided valuable information on the cause of MPL's failure -- but MPL couldn't do it because it was a polar landing, and its antenna (identical to Pathfinder's) was thus at the wrong angle to send any signal at all to Earth.

NASA was gambling that MPL would succeed, and that no postmortem would be needed. It lost that gamble -- and the need for detailed engineering telemetry from every probe during landing is now absolutely clear to everyone. Such a system should also be tested on the 2001 Lander.

It could probably make use of the omnidirectional UHF radio link that the Lander (like MPL) is already scheduled to have with NASA's Mars orbiters after landing. That link will be used to transmit 128,000 bits per second; it could easily transmit the needed engineering telemetry during landing while using a much lower bit rate that would cause less electrical power demand on the Lander during the landing.

If the rate could be raised to several thousand bits per second, it might even be possible to transmit live compressed copies of the descent camera photos, allowing determination of the overall reliability of the obstacle-avoidance system even if the Lander crashes.

These new modifications could be made -- without a wild leap in cost -- to the 2001 Lander despite the fact that it has already been largely constructed. Plans were already under way, in the wake of the MCO failure, to modify the Lander so that (like Pathfinder, MPL and the sample return landers) it had a pointable antenna dish giving it a backup direct radio link to Earth after landing.

This had originally been omitted on the grounds that the '96, '98 and 2001 Orbiters were all supposed to be available at the time of landing to relay its data back to Earth indirectly -- but now it would have been reduced to only two relay orbiters, one of which would be old and quite possibly out of action.

If the 2001 Orbiter had also failed, the Lander could have worked absolutely perfectly -- but been totally unable to send its data back to Earth! Incorporating these communications improvements into the Lander will certainly increase its weight and electrical power requirements, almost surely forcing removal of at least one of the experiments currently planned for it -- but it must be done.

I can't guess which experiments would be removed -- although my personal choice would be its "MARIE" radiation detector, which is solely for the purpose of gauging how hazardous the surface of Mars is to humans, and provides no other scientific data.

Finally, two other major changes may be made in the 2001 Lander. Even before the MPL landing, the MCO failure review board expressed concern over the fact that it would be the first soft landing spacecraft ever to use clusters of small "pulse-modulated" fixed-thrust engines in place of larger, smoothly throttleable engines -- on the grounds that the resulting "jerky" variations in engine thrust might produce vibrations that would make the Lander unstable and cause its fuel to slosh, causing wild variations in the engines' thrust level.

Experienced aerospace engineer James Oberg also regards this as his prime suspect in the MPL failure. Whether this is proven or not, NASA may very well decide to replace the 2001 Lander's thruster clusters with throttleable engines, which are more expensive but also easier for the lander's computer to control.

Second, it's possible (though very doubtful) that NASA might try to add one or both of the unique experiments lost on MPL to the 2001 Lander. If so, I doubt that it would be the "TEGA" soil-analysis oven to look for ground ice and analyze some soil minerals -- the 2001 Lander will almost certainly land at such a low latitude that there will be no near-surface ground ice.

But some of the probe's weather sensors -- especially its "TDL" sensor to measure air humidity and look for trace gases that could provide clues to ancient Mars' past -- may perhaps be added to the 2001 mission.If the Russian "lidar" instrument to measure atmospheric dust is reflown, the Mars Microphone -- which Russia allowed to be piggybacked on their instrument -- may be reflown. Otherwise, it's doubtful on this flight.

The need to make all these changes would almost surely delay the 2001 Lander mission to the next launch window in 2003, and force a similar two-year delay in the start of the later sample return missions. But it looks to me like a very wise precaution to do so, rather than simply cancelling the 2001 Lander and trying to keep the sample return landers on their original schedule.

Moreover, if it's launched in 2003, the Lander would also have at least two more possible Mars orbiters to relay its data back to Earth: the 2003 European Mars Express orbiter, and the first American "Mars Micromission" -- a small craft scheduled for launch in 2003 as the first element in an entire network of Mars-orbiting comsats to drastically improve the radio link between future Mars landers and Earth.

I've been intrigued to see a Dec. 10 item in Keith Cowing's "NASA Watch" Website, in which someone claiming to be an anonymous JPL engineer reports the supposed content of a meeting between JPL's Earth and Space Science Director Charles Elachi and his staff in which he laid out the first post-MPL sketch of JPL's future Mars program. If this information is correct, it matches my own guesses pretty closely:

"...JPL has been directed to re-look at the Mars Program architecture. This time there are no constraints other than the budget [limits]... e.g., no more must [we] launch an orbiter and lander every opportunity, no more must [we] fly the Athena payload, no more must [we] bring back a sample by 2008 [which would require a sample-return lander launch by at least 2005], etc...

"To re-look at the architecture, a Mars Architecture Study Team has been formed under Chris Jones and Dan McCleese... The team will be finished by the end of December. Charles [Elachi] will form an Executive Architecture Group to work the first two weeks of January. Headquarters will form an External Architecture Group to review the results at the end of January.

"Charles' crystal ball:

• "(a) Aggressive pursuit of one or two Micromissions for the 2003 opportunity -- both in equatorial orbit for a telecommunications network.
• "(b) Probably no change to the 2001 Orbiter.
• "(c) A 2001 lander unlikely, unless the cause of the MPL loss is identified quickly.
• "(d) [We] might move the 2001 lander to 2003 as an engineering demonstration.
• "(e) On all our future missions we will have [radio] contact every second during critical moments, 'even if we have to break the laws of physics.' "

However, his team is already working on a much smaller atmospheric sounder that could recover much of its data and might be flown on a Mars Micromission as early as 2005 -- as well as an alternative in which some of the Mars comsats would beam signals to each other through the atmosphere's layers and recover much of the data that way. (Also, two of the instruments on the Mars Express -- SPICAM and PFS -- may recover much of PMIRR's data.)

Second, the Deep Space-2 micro-penetrators also failed -- and their failure may possibly have been due to poor design and testing. But the whole concept of very small penetrators with miniature instruments is an extremely promising one for future Solar System exploration -- in fact, it strikes me as being one of the most important techniques for such future exploration.

Last February, an international workshop in France on future Mars exploration included a section discussing possible Mars Micromission designs -- and Deep Space-2 style penetrators were proposed that would carry everything from seismometers to tiny amino acid detectors. One proposal for the last round of Discovery Program mission selections -- "IMMPACT" -- would have used a single Mars Micromission to scatter a network of a dozen DS-2 type penetrators all over Mars, with their soil water sensors replaced by seismometers.

They will be just as useful for other worlds --at least one proposal has already been made to use them to study Europa. And Mars is the ideal world to carry out initial testing of this concept. I'm convinced that, at some point in the fairly near future, some other mission to test micro-penetrators on Mars will be flown, although their design may be significantly different.